UbiquitousOS
/
OpenCloudOS-Kernel
mirror of https://gitee.com/OpenCloudOS/OpenCloudOS-Kernel.git
11
0
Fork 0
Code Issues Projects Releases Wiki Activity
OpenCloudOS-Kernel/net/xfrm/xfrm_policy.c

3300 lines
78 KiB
C
Raw Normal View History

[NET] XFRM: Fix whitespace errors. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 22:25:29 +08:00
/*
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
* xfrm_policy.c
*
* Changes:
* Mitsuru KANDA @USAGI
* Kazunori MIYAZAWA @USAGI
* Kunihiro Ishiguro <kunihiro@ipinfusion.com>
* IPv6 support
* Kazunori MIYAZAWA @USAGI
* YOSHIFUJI Hideaki
* Split up af-specific portion
* Derek Atkins <derek@ihtfp.com> Add the post_input processor
[LSM-IPSec]: Security association restriction. This patch series implements per packet access control via the extension of the Linux Security Modules (LSM) interface by hooks in the XFRM and pfkey subsystems that leverage IPSec security associations to label packets. Extensions to the SELinux LSM are included that leverage the patch for this purpose. This patch implements the changes necessary to the XFRM subsystem, pfkey interface, ipv4/ipv6, and xfrm_user interface to restrict a socket to use only authorized security associations (or no security association) to send/receive network packets. Patch purpose: The patch is designed to enable access control per packets based on the strongly authenticated IPSec security association. Such access controls augment the existing ones based on network interface and IP address. The former are very coarse-grained, and the latter can be spoofed. By using IPSec, the system can control access to remote hosts based on cryptographic keys generated using the IPSec mechanism. This enables access control on a per-machine basis or per-application if the remote machine is running the same mechanism and trusted to enforce the access control policy. Patch design approach: The overall approach is that policy (xfrm_policy) entries set by user-level programs (e.g., setkey for ipsec-tools) are extended with a security context that is used at policy selection time in the XFRM subsystem to restrict the sockets that can send/receive packets via security associations (xfrm_states) that are built from those policies. A presentation available at www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf from the SELinux symposium describes the overall approach. Patch implementation details: On output, the policy retrieved (via xfrm_policy_lookup or xfrm_sk_policy_lookup) must be authorized for the security context of the socket and the same security context is required for resultant security association (retrieved or negotiated via racoon in ipsec-tools). This is enforced in xfrm_state_find. On input, the policy retrieved must also be authorized for the socket (at __xfrm_policy_check), and the security context of the policy must also match the security association being used. The patch has virtually no impact on packets that do not use IPSec. The existing Netfilter (outgoing) and LSM rcv_skb hooks are used as before. Also, if IPSec is used without security contexts, the impact is minimal. The LSM must allow such policies to be selected for the combination of socket and remote machine, but subsequent IPSec processing proceeds as in the original case. Testing: The pfkey interface is tested using the ipsec-tools. ipsec-tools have been modified (a separate ipsec-tools patch is available for version 0.5) that supports assignment of xfrm_policy entries and security associations with security contexts via setkey and the negotiation using the security contexts via racoon. The xfrm_user interface is tested via ad hoc programs that set security contexts. These programs are also available from me, and contain programs for setting, getting, and deleting policy for testing this interface. Testing of sa functions was done by tracing kernel behavior. Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-14 15:12:27 +08:00
*
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
*/
[IPSEC]: Move flow construction into xfrm_dst_lookup This patch moves the flow construction from the callers of xfrm_dst_lookup into that function. It also changes xfrm_dst_lookup so that it takes an xfrm state as its argument instead of explicit addresses. This removes any address-specific logic from the callers of xfrm_dst_lookup which is needed to correctly support inter-family transforms. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-11-14 13:37:28 +08:00
#include <linux/err.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
#include <linux/slab.h>
#include <linux/kmod.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/workqueue.h>
#include <linux/notifier.h>
#include <linux/netdevice.h>
[NETFILTER]: Handle NAT in IPsec policy checks Handle NAT of decapsulated IPsec packets by reconstructing the struct flowi of the original packet from the conntrack information for IPsec policy checks. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-01-07 15:06:30 +08:00
#include <linux/netfilter.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
#include <linux/module.h>
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
#include <linux/cache.h>
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
#include <linux/cpu.h>
[XFRM]: Assorted IPsec fixups This patch fixes a number of small but potentially troublesome things in the XFRM/IPsec code: * Use the 'audit_enabled' variable already in include/linux/audit.h Removed the need for extern declarations local to each XFRM audit fuction * Convert 'sid' to 'secid' everywhere we can The 'sid' name is specific to SELinux, 'secid' is the common naming convention used by the kernel when refering to tokenized LSM labels, unfortunately we have to leave 'ctx_sid' in 'struct xfrm_sec_ctx' otherwise we risk breaking userspace * Convert address display to use standard NIP* macros Similar to what was recently done with the SPD audit code, this also also includes the removal of some unnecessary memcpy() calls * Move common code to xfrm_audit_common_stateinfo() Code consolidation from the "less is more" book on software development * Proper spacing around commas in function arguments Minor style tweak since I was already touching the code Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:49:33 +08:00
#include <linux/audit.h>
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
#include <net/dst.h>
SELinux: include flow.h where used rather than get it indirectly We use flow_cache_genid in the selinux xfrm files. This is declared in net/flow.h However we do not include that file directly anywhere. We have always just gotten it through a long chain of indirect .h file includes. on x86_64: CC security/selinux/ss/services.o In file included from /next/linux-next-20120216/security/selinux/ss/services.c:69:0: /next/linux-next-20120216/security/selinux/include/xfrm.h: In function 'selinux_xfrm_notify_policyload': /next/linux-next-20120216/security/selinux/include/xfrm.h:51:14: error: 'flow_cache_genid' undeclared (first use in this function) /next/linux-next-20120216/security/selinux/include/xfrm.h:51:14: note: each undeclared identifier is reported only once for each function it appears in make[3]: *** [security/selinux/ss/services.o] Error 1 Reported-by: Randy Dunlap <rdunlap@xenotime.net> Signed-off-by: Eric Paris <eparis@redhat.com>
2012-02-17 04:08:39 +08:00
#include <net/flow.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
#include <net/xfrm.h>
#include <net/ip.h>
[XFRM]: Define packet dropping statistics. This statistics is shown factor dropped by transformation at /proc/net/xfrm_stat for developer. It is a counter designed from current transformation source code and defined as linux private MIB. See Documentation/networking/xfrm_proc.txt for the detail. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:42:57 +08:00
#ifdef CONFIG_XFRM_STATISTICS
#include <net/snmp.h>
#endif
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM]: Extract common hashing code into xfrm_hash.[ch] Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:50:50 +08:00
#include "xfrm_hash.h"
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
#define XFRM_QUEUE_TMO_MIN ((unsigned)(HZ/10))
#define XFRM_QUEUE_TMO_MAX ((unsigned)(60*HZ))
#define XFRM_MAX_QUEUE_LEN 100
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
struct xfrm_flo {
struct dst_entry *dst_orig;
u8 flags;
};
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
static DEFINE_PER_CPU(struct xfrm_dst *, xfrm_last_dst);
static struct work_struct *xfrm_pcpu_work __read_mostly;
Replace rwlock on xfrm_policy_afinfo with rcu xfrm_policy_afinfo is read mosly data structure. Write on xfrm_policy_afinfo is done only at the time of configuration. So rwlocks can be safely replaced with RCU. RCUs usage optimizes the performance. Signed-off-by: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-13 05:22:29 +08:00
static DEFINE_SPINLOCK(xfrm_policy_afinfo_lock);
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
static struct xfrm_policy_afinfo const __rcu *xfrm_policy_afinfo[AF_INET6 + 1]
Replace rwlock on xfrm_policy_afinfo with rcu xfrm_policy_afinfo is read mosly data structure. Write on xfrm_policy_afinfo is done only at the time of configuration. So rwlocks can be safely replaced with RCU. RCUs usage optimizes the performance. Signed-off-by: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-13 05:22:29 +08:00
__read_mostly;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: mark kmem_caches as __ro_after_init Kmem caches aren't relocated once set up. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-02-25 02:21:38 +08:00
static struct kmem_cache *xfrm_dst_cache __ro_after_init;
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
static __read_mostly seqcount_t xfrm_policy_hash_generation;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
static void xfrm_init_pmtu(struct xfrm_dst **bundle, int nr);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
static int stale_bundle(struct dst_entry *dst);
xfrm: Remove family arg from xfrm_bundle_ok The family arg is not used any more, so remove it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-06-30 07:18:20 +08:00
static int xfrm_bundle_ok(struct xfrm_dst *xdst);
xfrm: Convert timers to use timer_setup() In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() helper to pass the timer pointer explicitly. Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: "David S. Miller" <davem@davemloft.net> Cc: netdev@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-17 08:28:56 +08:00
static void xfrm_policy_queue_process(struct timer_list *t);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Use __xfrm_policy_link in xfrm_policy_insert For a long time we couldn't actually use __xfrm_policy_link in xfrm_policy_insert because the latter wanted to do hashing at a specific position. Now that __xfrm_policy_link no longer does hashing it can now be safely used in xfrm_policy_insert to kill some duplicate code, finally reuniting general policies with socket policies. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:50 +08:00
static void __xfrm_policy_link(struct xfrm_policy *pol, int dir);
xfrm: Cleanup for unlink SPD entry Used __xfrm_policy_unlink() to instead of the dup codes when unlink SPD entry. Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-12-03 16:33:09 +08:00
static struct xfrm_policy *__xfrm_policy_unlink(struct xfrm_policy *pol,
int dir);
xfrm: policy: use atomic_inc_not_zero in rcu section If we don't hold the policy lock anymore the refcnt might already be 0, i.e. policy struct is about to be free'd. Switch to atomic_inc_not_zero to avoid this. On removal policies are already unlinked from the tables (lists) before the last _put occurs so we are not supposed to find the same 'dead' entry on the next loop, so its safe to just repeat the lookup. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:55 +08:00
static inline bool xfrm_pol_hold_rcu(struct xfrm_policy *policy)
{
net, xfrm: convert xfrm_policy.refcnt from atomic_t to refcount_t refcount_t type and corresponding API should be used instead of atomic_t when the variable is used as a reference counter. This allows to avoid accidental refcounter overflows that might lead to use-after-free situations. Signed-off-by: Elena Reshetova <elena.reshetova@intel.com> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David Windsor <dwindsor@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-04 20:53:22 +08:00
return refcount_inc_not_zero(&policy->refcnt);
xfrm: policy: use atomic_inc_not_zero in rcu section If we don't hold the policy lock anymore the refcnt might already be 0, i.e. policy struct is about to be free'd. Switch to atomic_inc_not_zero to avoid this. On removal policies are already unlinked from the tables (lists) before the last _put occurs so we are not supposed to find the same 'dead' entry on the next loop, so its safe to just repeat the lookup. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:55 +08:00
}
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
static inline bool
xfrm: Const'ify selector argument to xfrm_selector_match() Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:12:25 +08:00
__xfrm4_selector_match(const struct xfrm_selector *sel, const struct flowi *fl)
[XFRM]: uninline xfrm_selector_match() Six callsites, huge. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-09 14:46:26 +08:00
{
net: Use flowi4 and flowi6 in xfrm layer. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 15:42:11 +08:00
const struct flowi4 *fl4 = &fl->u.ip4;
xfrm: optimize ipv4 selector matching Current addr_match() is errh, under-optimized. Compiler doesn't know that memcmp() branch doesn't trigger for IPv4. Also, pass addresses by value -- they fit into register. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-22 14:46:02 +08:00
return addr4_match(fl4->daddr, sel->daddr.a4, sel->prefixlen_d) &&
addr4_match(fl4->saddr, sel->saddr.a4, sel->prefixlen_s) &&
net: Use flowi4 and flowi6 in xfrm layer. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 15:42:11 +08:00
!((xfrm_flowi_dport(fl, &fl4->uli) ^ sel->dport) & sel->dport_mask) &&
!((xfrm_flowi_sport(fl, &fl4->uli) ^ sel->sport) & sel->sport_mask) &&
(fl4->flowi4_proto == sel->proto || !sel->proto) &&
(fl4->flowi4_oif == sel->ifindex || !sel->ifindex);
[XFRM]: uninline xfrm_selector_match() Six callsites, huge. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-09 14:46:26 +08:00
}
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
static inline bool
xfrm: Const'ify selector argument to xfrm_selector_match() Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:12:25 +08:00
__xfrm6_selector_match(const struct xfrm_selector *sel, const struct flowi *fl)
[XFRM]: uninline xfrm_selector_match() Six callsites, huge. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-09 14:46:26 +08:00
{
net: Use flowi4 and flowi6 in xfrm layer. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 15:42:11 +08:00
const struct flowi6 *fl6 = &fl->u.ip6;
return addr_match(&fl6->daddr, &sel->daddr, sel->prefixlen_d) &&
addr_match(&fl6->saddr, &sel->saddr, sel->prefixlen_s) &&
!((xfrm_flowi_dport(fl, &fl6->uli) ^ sel->dport) & sel->dport_mask) &&
!((xfrm_flowi_sport(fl, &fl6->uli) ^ sel->sport) & sel->sport_mask) &&
(fl6->flowi6_proto == sel->proto || !sel->proto) &&
(fl6->flowi6_oif == sel->ifindex || !sel->ifindex);
[XFRM]: uninline xfrm_selector_match() Six callsites, huge. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-09 14:46:26 +08:00
}
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
bool xfrm_selector_match(const struct xfrm_selector *sel, const struct flowi *fl,
unsigned short family)
[XFRM]: uninline xfrm_selector_match() Six callsites, huge. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-09 14:46:26 +08:00
{
switch (family) {
case AF_INET:
return __xfrm4_selector_match(sel, fl);
case AF_INET6:
return __xfrm6_selector_match(sel, fl);
}
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
return false;
[XFRM]: uninline xfrm_selector_match() Six callsites, huge. Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-09 14:46:26 +08:00
}
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
static const struct xfrm_policy_afinfo *xfrm_policy_get_afinfo(unsigned short family)
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
{
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
const struct xfrm_policy_afinfo *afinfo;
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
if (unlikely(family >= ARRAY_SIZE(xfrm_policy_afinfo)))
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
return NULL;
rcu_read_lock();
afinfo = rcu_dereference(xfrm_policy_afinfo[family]);
if (unlikely(!afinfo))
rcu_read_unlock();
return afinfo;
}
xfrm: Add an IPsec hardware offloading API This patch adds all the bits that are needed to do IPsec hardware offload for IPsec states and ESP packets. We add xfrmdev_ops to the net_device. xfrmdev_ops has function pointers that are needed to manage the xfrm states in the hardware and to do a per packet offloading decision. Joint work with: Ilan Tayari <ilant@mellanox.com> Guy Shapiro <guysh@mellanox.com> Yossi Kuperman <yossiku@mellanox.com> Signed-off-by: Guy Shapiro <guysh@mellanox.com> Signed-off-by: Ilan Tayari <ilant@mellanox.com> Signed-off-by: Yossi Kuperman <yossiku@mellanox.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-04-14 16:06:10 +08:00
struct dst_entry *__xfrm_dst_lookup(struct net *net, int tos, int oif,
const xfrm_address_t *saddr,
const xfrm_address_t *daddr,
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
int family, u32 mark)
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo;
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
struct dst_entry *dst;
afinfo = xfrm_policy_get_afinfo(family);
if (unlikely(afinfo == NULL))
return ERR_PTR(-EAFNOSUPPORT);
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
dst = afinfo->dst_lookup(net, tos, oif, saddr, daddr, mark);
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
return dst;
}
xfrm: Add an IPsec hardware offloading API This patch adds all the bits that are needed to do IPsec hardware offload for IPsec states and ESP packets. We add xfrmdev_ops to the net_device. xfrmdev_ops has function pointers that are needed to manage the xfrm states in the hardware and to do a per packet offloading decision. Joint work with: Ilan Tayari <ilant@mellanox.com> Guy Shapiro <guysh@mellanox.com> Yossi Kuperman <yossiku@mellanox.com> Signed-off-by: Guy Shapiro <guysh@mellanox.com> Signed-off-by: Ilan Tayari <ilant@mellanox.com> Signed-off-by: Yossi Kuperman <yossiku@mellanox.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-04-14 16:06:10 +08:00
EXPORT_SYMBOL(__xfrm_dst_lookup);
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
xfrm: Add oif to dst lookups Rules can be installed that direct route lookups to specific tables based on oif. Plumb the oif through the xfrm lookups so it gets set in the flow struct and passed to the resolver routines. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-08-11 06:58:11 +08:00
static inline struct dst_entry *xfrm_dst_lookup(struct xfrm_state *x,
int tos, int oif,
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
xfrm_address_t *prev_saddr,
xfrm_address_t *prev_daddr,
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
int family, u32 mark)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
netns xfrm: ->dst_lookup in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:51:25 +08:00
struct net *net = xs_net(x);
[IPSEC]: Move flow construction into xfrm_dst_lookup This patch moves the flow construction from the callers of xfrm_dst_lookup into that function. It also changes xfrm_dst_lookup so that it takes an xfrm state as its argument instead of explicit addresses. This removes any address-specific logic from the callers of xfrm_dst_lookup which is needed to correctly support inter-family transforms. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-11-14 13:37:28 +08:00
xfrm_address_t *saddr = &x->props.saddr;
xfrm_address_t *daddr = &x->id.daddr;
struct dst_entry *dst;
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
if (x->type->flags & XFRM_TYPE_LOCAL_COADDR) {
[IPSEC]: Move flow construction into xfrm_dst_lookup This patch moves the flow construction from the callers of xfrm_dst_lookup into that function. It also changes xfrm_dst_lookup so that it takes an xfrm state as its argument instead of explicit addresses. This removes any address-specific logic from the callers of xfrm_dst_lookup which is needed to correctly support inter-family transforms. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-11-14 13:37:28 +08:00
saddr = x->coaddr;
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
daddr = prev_daddr;
}
if (x->type->flags & XFRM_TYPE_REMOTE_COADDR) {
saddr = prev_saddr;
[IPSEC]: Move flow construction into xfrm_dst_lookup This patch moves the flow construction from the callers of xfrm_dst_lookup into that function. It also changes xfrm_dst_lookup so that it takes an xfrm state as its argument instead of explicit addresses. This removes any address-specific logic from the callers of xfrm_dst_lookup which is needed to correctly support inter-family transforms. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-11-14 13:37:28 +08:00
daddr = x->coaddr;
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
dst = __xfrm_dst_lookup(net, tos, oif, saddr, daddr, family, mark);
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
if (!IS_ERR(dst)) {
if (prev_saddr != saddr)
memcpy(prev_saddr, saddr, sizeof(*prev_saddr));
if (prev_daddr != daddr)
memcpy(prev_daddr, daddr, sizeof(*prev_daddr));
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IPSEC]: Move flow construction into xfrm_dst_lookup This patch moves the flow construction from the callers of xfrm_dst_lookup into that function. It also changes xfrm_dst_lookup so that it takes an xfrm state as its argument instead of explicit addresses. This removes any address-specific logic from the callers of xfrm_dst_lookup which is needed to correctly support inter-family transforms. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-11-14 13:37:28 +08:00
return dst;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
static inline unsigned long make_jiffies(long secs)
{
if (secs >= (MAX_SCHEDULE_TIMEOUT-1)/HZ)
return MAX_SCHEDULE_TIMEOUT-1;
else
[NET] XFRM: Fix whitespace errors. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 22:25:29 +08:00
return secs*HZ;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: Convert timers to use timer_setup() In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() helper to pass the timer pointer explicitly. Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: "David S. Miller" <davem@davemloft.net> Cc: netdev@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-17 08:28:56 +08:00
static void xfrm_policy_timer(struct timer_list *t)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: Convert timers to use timer_setup() In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() helper to pass the timer pointer explicitly. Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: "David S. Miller" <davem@davemloft.net> Cc: netdev@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-17 08:28:56 +08:00
struct xfrm_policy *xp = from_timer(xp, t, timer);
[NET]: Convert xtime.tv_sec to get_seconds() Where appropriate, convert references to xtime.tv_sec to the get_seconds() helper function. Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-05 08:12:44 +08:00
unsigned long now = get_seconds();
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
long next = LONG_MAX;
int warn = 0;
int dir;
read_lock(&xp->lock);
xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-31 08:17:05 +08:00
if (unlikely(xp->walk.dead))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
goto out;
[IPSEC]: Document that policy direction is derived from the index. Here is a patch that adds a helper called xfrm_policy_id2dir to document the fact that the policy direction can be and is derived from the index. This is based on a patch by YOSHIFUJI Hideaki and 210313105@suda.edu.cn. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-10-06 03:15:12 +08:00
dir = xfrm_policy_id2dir(xp->index);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (xp->lft.hard_add_expires_seconds) {
long tmo = xp->lft.hard_add_expires_seconds +
xp->curlft.add_time - now;
if (tmo <= 0)
goto expired;
if (tmo < next)
next = tmo;
}
if (xp->lft.hard_use_expires_seconds) {
long tmo = xp->lft.hard_use_expires_seconds +
(xp->curlft.use_time ? : xp->curlft.add_time) - now;
if (tmo <= 0)
goto expired;
if (tmo < next)
next = tmo;
}
if (xp->lft.soft_add_expires_seconds) {
long tmo = xp->lft.soft_add_expires_seconds +
xp->curlft.add_time - now;
if (tmo <= 0) {
warn = 1;
tmo = XFRM_KM_TIMEOUT;
}
if (tmo < next)
next = tmo;
}
if (xp->lft.soft_use_expires_seconds) {
long tmo = xp->lft.soft_use_expires_seconds +
(xp->curlft.use_time ? : xp->curlft.add_time) - now;
if (tmo <= 0) {
warn = 1;
tmo = XFRM_KM_TIMEOUT;
}
if (tmo < next)
next = tmo;
}
if (warn)
[IPSEC]: Sync series - policy expires This is similar to the SA expire insertion patch - only it inserts expires for SP. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-21 11:17:25 +08:00
km_policy_expired(xp, dir, 0, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (next != LONG_MAX &&
!mod_timer(&xp->timer, jiffies + make_jiffies(next)))
xfrm_pol_hold(xp);
out:
read_unlock(&xp->lock);
xfrm_pol_put(xp);
return;
expired:
read_unlock(&xp->lock);
[IPSEC] Kill spurious hard expire messages This patch ensures that the hard state/policy expire notifications are only sent when the state/policy is successfully removed from their respective tables. As it is, it's possible for a state/policy to both expire through reaching a hard limit, as well as being deleted by the user. Note that this behaviour isn't actually forbidden by RFC 2367. However, it is a quality of implementation issue. As an added bonus, the restructuring in this patch will help eventually in moving the expire notifications from softirq context into process context, thus improving their reliability. One important side-effect from this change is that SAs reaching their hard byte/packet limits are now deleted immediately, just like SAs that have reached their hard time limits. Previously they were announced immediately but only deleted after 30 seconds. This is bad because it prevents the system from issuing an ACQUIRE command until the existing state was deleted by the user or expires after the time is up. In the scenario where the expire notification was lost this introduces a 30 second delay into the system for no good reason. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2005-06-19 13:43:22 +08:00
if (!xfrm_policy_delete(xp, dir))
[IPSEC]: Sync series - policy expires This is similar to the SA expire insertion patch - only it inserts expires for SP. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-21 11:17:25 +08:00
km_policy_expired(xp, dir, 1, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_pol_put(xp);
}
/* Allocate xfrm_policy. Not used here, it is supposed to be used by pfkeyv2
* SPD calls.
*/
netns xfrm: add struct xfrm_policy::xp_net Again, to avoid complications with passing netns when not necessary. Again, ->xp_net is set-once field, once set it never changes. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:21:45 +08:00
struct xfrm_policy *xfrm_policy_alloc(struct net *net, gfp_t gfp)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct xfrm_policy *policy;
[NET]: Conversions from kmalloc+memset to k(z|c)alloc. Signed-off-by: Panagiotis Issaris <takis@issaris.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-07-22 05:51:30 +08:00
policy = kzalloc(sizeof(struct xfrm_policy), gfp);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (policy) {
netns xfrm: add struct xfrm_policy::xp_net Again, to avoid complications with passing netns when not necessary. Again, ->xp_net is set-once field, once set it never changes. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:21:45 +08:00
write_pnet(&policy->xp_net, net);
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
INIT_LIST_HEAD(&policy->walk.all);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
INIT_HLIST_NODE(&policy->bydst);
INIT_HLIST_NODE(&policy->byidx);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
rwlock_init(&policy->lock);
net, xfrm: convert xfrm_policy.refcnt from atomic_t to refcount_t refcount_t type and corresponding API should be used instead of atomic_t when the variable is used as a reference counter. This allows to avoid accidental refcounter overflows that might lead to use-after-free situations. Signed-off-by: Elena Reshetova <elena.reshetova@intel.com> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David Windsor <dwindsor@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-04 20:53:22 +08:00
refcount_set(&policy->refcnt, 1);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
skb_queue_head_init(&policy->polq.hold_queue);
xfrm: Convert timers to use timer_setup() In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() helper to pass the timer pointer explicitly. Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: "David S. Miller" <davem@davemloft.net> Cc: netdev@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-17 08:28:56 +08:00
timer_setup(&policy->timer, xfrm_policy_timer, 0);
timer_setup(&policy->polq.hold_timer,
xfrm_policy_queue_process, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
return policy;
}
EXPORT_SYMBOL(xfrm_policy_alloc);
xfrm: add rcu grace period in xfrm_policy_destroy() We will soon switch sk->sk_policy[] to RCU protection, as SYNACK packets are sent while listener socket is not locked. This patch simply adds RCU grace period before struct xfrm_policy freeing, and the corresponding rcu_head in struct xfrm_policy. Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:01 +08:00
static void xfrm_policy_destroy_rcu(struct rcu_head *head)
{
struct xfrm_policy *policy = container_of(head, struct xfrm_policy, rcu);
security_xfrm_policy_free(policy->security);
kfree(policy);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* Destroy xfrm_policy: descendant resources must be released to this moment. */
[XFRM] xfrm_policy_destroy: Rename and relative fixes. Since __xfrm_policy_destroy is used to destory the resources allocated by xfrm_policy_alloc. So using the name __xfrm_policy_destroy is not correspond with xfrm_policy_alloc. Rename it to xfrm_policy_destroy. And along with some instances that call xfrm_policy_alloc but not using xfrm_policy_destroy to destroy the resource, fix them. Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-08 14:34:29 +08:00
void xfrm_policy_destroy(struct xfrm_policy *policy)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
BUG_ON(!policy->walk.dead);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Delete hold_timer when destroy policy Both policy timer and hold_timer need to be deleted when destroy policy Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-08-01 18:08:36 +08:00
if (del_timer(&policy->timer) || del_timer(&policy->polq.hold_timer))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
BUG();
xfrm: add rcu grace period in xfrm_policy_destroy() We will soon switch sk->sk_policy[] to RCU protection, as SYNACK packets are sent while listener socket is not locked. This patch simply adds RCU grace period before struct xfrm_policy freeing, and the corresponding rcu_head in struct xfrm_policy. Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:01 +08:00
call_rcu(&policy->rcu, xfrm_policy_destroy_rcu);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM] xfrm_policy_destroy: Rename and relative fixes. Since __xfrm_policy_destroy is used to destory the resources allocated by xfrm_policy_alloc. So using the name __xfrm_policy_destroy is not correspond with xfrm_policy_alloc. Rename it to xfrm_policy_destroy. And along with some instances that call xfrm_policy_alloc but not using xfrm_policy_destroy to destroy the resource, fix them. Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-08 14:34:29 +08:00
EXPORT_SYMBOL(xfrm_policy_destroy);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: trivial typos o s/descentant/descendant o s/workarbound/workaround Signed-off-by: Alexander Alemayhu <alexander@alemayhu.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-01-04 00:13:20 +08:00
/* Rule must be locked. Release descendant resources, announce
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
* entry dead. The rule must be unlinked from lists to the moment.
*/
static void xfrm_policy_kill(struct xfrm_policy *policy)
{
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
policy->walk.dead = 1;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: remove policy garbage collection Policies are now properly reference counted and destroyed from all code paths. The delayed gc is just an overhead now and can be removed. Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:06 +08:00
atomic_inc(&policy->genid);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
if (del_timer(&policy->polq.hold_timer))
xfrm_pol_put(policy);
xfrm: remove the xfrm_queue_purge definition The task of xfrm_queue_purge is same as skb_queue_purge, so remove it Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-22 15:51:16 +08:00
skb_queue_purge(&policy->polq.hold_queue);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
xfrm: remove policy garbage collection Policies are now properly reference counted and destroyed from all code paths. The delayed gc is just an overhead now and can be removed. Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:06 +08:00
if (del_timer(&policy->timer))
xfrm_pol_put(policy);
xfrm_pol_put(policy);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
static unsigned int xfrm_policy_hashmax __read_mostly = 1 * 1024 * 1024;
netns xfrm: propagate netns into policy byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:32:41 +08:00
static inline unsigned int idx_hash(struct net *net, u32 index)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: propagate netns into policy byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:32:41 +08:00
return __idx_hash(index, net->xfrm.policy_idx_hmask);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
/* calculate policy hash thresholds */
static void __get_hash_thresh(struct net *net,
unsigned short family, int dir,
u8 *dbits, u8 *sbits)
{
switch (family) {
case AF_INET:
*dbits = net->xfrm.policy_bydst[dir].dbits4;
*sbits = net->xfrm.policy_bydst[dir].sbits4;
break;
case AF_INET6:
*dbits = net->xfrm.policy_bydst[dir].dbits6;
*sbits = net->xfrm.policy_bydst[dir].sbits6;
break;
default:
*dbits = 0;
*sbits = 0;
}
}
xfrm: Const'ify address args to hash helpers. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:33:19 +08:00
static struct hlist_head *policy_hash_bysel(struct net *net,
const struct xfrm_selector *sel,
unsigned short family, int dir)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
unsigned int hmask = net->xfrm.policy_bydst[dir].hmask;
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
unsigned int hash;
u8 dbits;
u8 sbits;
__get_hash_thresh(net, family, dir, &dbits, &sbits);
hash = __sel_hash(sel, family, hmask, dbits, sbits);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: prepare policy_bydst hash for rcu lookups Since commit 56f047305dd4b6b617 ("xfrm: add rcu grace period in xfrm_policy_destroy()") xfrm policy objects are already free'd via rcu. In order to make more places lockless (i.e. use rcu_read_lock instead of grabbing read-side of policy rwlock) we only need to: - use rcu_assign_pointer to store address of new hash table backend memory - add rcu barrier so that freeing of old memory is delayed (expansion and free happens from system workqueue, so synchronize_rcu is fine) - use rcu_dereference to fetch current address of the hash table. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:53 +08:00
if (hash == hmask + 1)
return &net->xfrm.policy_inexact[dir];
return rcu_dereference_check(net->xfrm.policy_bydst[dir].table,
lockdep_is_held(&net->xfrm.xfrm_policy_lock)) + hash;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
xfrm: Const'ify address args to hash helpers. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:33:19 +08:00
static struct hlist_head *policy_hash_direct(struct net *net,
const xfrm_address_t *daddr,
const xfrm_address_t *saddr,
unsigned short family, int dir)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
unsigned int hmask = net->xfrm.policy_bydst[dir].hmask;
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
unsigned int hash;
u8 dbits;
u8 sbits;
__get_hash_thresh(net, family, dir, &dbits, &sbits);
hash = __addr_hash(daddr, saddr, family, hmask, dbits, sbits);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: prepare policy_bydst hash for rcu lookups Since commit 56f047305dd4b6b617 ("xfrm: add rcu grace period in xfrm_policy_destroy()") xfrm policy objects are already free'd via rcu. In order to make more places lockless (i.e. use rcu_read_lock instead of grabbing read-side of policy rwlock) we only need to: - use rcu_assign_pointer to store address of new hash table backend memory - add rcu barrier so that freeing of old memory is delayed (expansion and free happens from system workqueue, so synchronize_rcu is fine) - use rcu_dereference to fetch current address of the hash table. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:53 +08:00
return rcu_dereference_check(net->xfrm.policy_bydst[dir].table,
lockdep_is_held(&net->xfrm.xfrm_policy_lock)) + hash;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
static void xfrm_dst_hash_transfer(struct net *net,
struct hlist_head *list,
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct hlist_head *ndsttable,
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
unsigned int nhashmask,
int dir)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
struct hlist_node *tmp, *entry0 = NULL;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol;
[XFRM]: Fix ordering issue in xfrm_dst_hash_transfer(). Keep ordering of policy entries with same selector in xfrm_dst_hash_transfer(). Issue should not appear in usual cases because multiple policy entries with same selector are basically not allowed so far. Bug was pointed out by Sebastien Decugis <sdecugis@hongo.wide.ad.jp>. We could convert bydst from hlist to list and use list_add_tail() instead. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Acked-by: Sebastien Decugis <sdecugis@hongo.wide.ad.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-18 15:29:30 +08:00
unsigned int h0 = 0;
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
u8 dbits;
u8 sbits;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
[XFRM]: Fix ordering issue in xfrm_dst_hash_transfer(). Keep ordering of policy entries with same selector in xfrm_dst_hash_transfer(). Issue should not appear in usual cases because multiple policy entries with same selector are basically not allowed so far. Bug was pointed out by Sebastien Decugis <sdecugis@hongo.wide.ad.jp>. We could convert bydst from hlist to list and use list_add_tail() instead. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Acked-by: Sebastien Decugis <sdecugis@hongo.wide.ad.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-18 15:29:30 +08:00
redo:
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry_safe(pol, tmp, list, bydst) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
unsigned int h;
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
__get_hash_thresh(net, pol->family, dir, &dbits, &sbits);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
h = __addr_hash(&pol->selector.daddr, &pol->selector.saddr,
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
pol->family, nhashmask, dbits, sbits);
[XFRM]: Fix ordering issue in xfrm_dst_hash_transfer(). Keep ordering of policy entries with same selector in xfrm_dst_hash_transfer(). Issue should not appear in usual cases because multiple policy entries with same selector are basically not allowed so far. Bug was pointed out by Sebastien Decugis <sdecugis@hongo.wide.ad.jp>. We could convert bydst from hlist to list and use list_add_tail() instead. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Acked-by: Sebastien Decugis <sdecugis@hongo.wide.ad.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-18 15:29:30 +08:00
if (!entry0) {
xfrm: policy: use rcu versions for iteration and list add/del This is required once we allow lockless readers. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:52 +08:00
hlist_del_rcu(&pol->bydst);
hlist_add_head_rcu(&pol->bydst, ndsttable + h);
[XFRM]: Fix ordering issue in xfrm_dst_hash_transfer(). Keep ordering of policy entries with same selector in xfrm_dst_hash_transfer(). Issue should not appear in usual cases because multiple policy entries with same selector are basically not allowed so far. Bug was pointed out by Sebastien Decugis <sdecugis@hongo.wide.ad.jp>. We could convert bydst from hlist to list and use list_add_tail() instead. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Acked-by: Sebastien Decugis <sdecugis@hongo.wide.ad.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-18 15:29:30 +08:00
h0 = h;
} else {
if (h != h0)
continue;
xfrm: policy: use rcu versions for iteration and list add/del This is required once we allow lockless readers. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:52 +08:00
hlist_del_rcu(&pol->bydst);
hlist_add_behind_rcu(&pol->bydst, entry0);
[XFRM]: Fix ordering issue in xfrm_dst_hash_transfer(). Keep ordering of policy entries with same selector in xfrm_dst_hash_transfer(). Issue should not appear in usual cases because multiple policy entries with same selector are basically not allowed so far. Bug was pointed out by Sebastien Decugis <sdecugis@hongo.wide.ad.jp>. We could convert bydst from hlist to list and use list_add_tail() instead. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Acked-by: Sebastien Decugis <sdecugis@hongo.wide.ad.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-18 15:29:30 +08:00
}
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
entry0 = &pol->bydst;
[XFRM]: Fix ordering issue in xfrm_dst_hash_transfer(). Keep ordering of policy entries with same selector in xfrm_dst_hash_transfer(). Issue should not appear in usual cases because multiple policy entries with same selector are basically not allowed so far. Bug was pointed out by Sebastien Decugis <sdecugis@hongo.wide.ad.jp>. We could convert bydst from hlist to list and use list_add_tail() instead. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Acked-by: Sebastien Decugis <sdecugis@hongo.wide.ad.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-18 15:29:30 +08:00
}
if (!hlist_empty(list)) {
entry0 = NULL;
goto redo;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
}
static void xfrm_idx_hash_transfer(struct hlist_head *list,
struct hlist_head *nidxtable,
unsigned int nhashmask)
{
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
struct hlist_node *tmp;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry_safe(pol, tmp, list, byidx) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
unsigned int h;
h = __idx_hash(pol->index, nhashmask);
hlist_add_head(&pol->byidx, nidxtable+h);
}
}
static unsigned long xfrm_new_hash_mask(unsigned int old_hmask)
{
return ((old_hmask + 1) << 1) - 1;
}
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
static void xfrm_bydst_resize(struct net *net, int dir)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
unsigned int hmask = net->xfrm.policy_bydst[dir].hmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
unsigned int nhashmask = xfrm_new_hash_mask(hmask);
unsigned int nsize = (nhashmask + 1) * sizeof(struct hlist_head);
[XFRM]: Extract common hashing code into xfrm_hash.[ch] Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:50:50 +08:00
struct hlist_head *ndst = xfrm_hash_alloc(nsize);
xfrm: policy: prepare policy_bydst hash for rcu lookups Since commit 56f047305dd4b6b617 ("xfrm: add rcu grace period in xfrm_policy_destroy()") xfrm policy objects are already free'd via rcu. In order to make more places lockless (i.e. use rcu_read_lock instead of grabbing read-side of policy rwlock) we only need to: - use rcu_assign_pointer to store address of new hash table backend memory - add rcu barrier so that freeing of old memory is delayed (expansion and free happens from system workqueue, so synchronize_rcu is fine) - use rcu_dereference to fetch current address of the hash table. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:53 +08:00
struct hlist_head *odst;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
int i;
if (!ndst)
return;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
write_seqcount_begin(&xfrm_policy_hash_generation);
odst = rcu_dereference_protected(net->xfrm.policy_bydst[dir].table,
lockdep_is_held(&net->xfrm.xfrm_policy_lock));
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: prepare policy_bydst hash for rcu lookups Since commit 56f047305dd4b6b617 ("xfrm: add rcu grace period in xfrm_policy_destroy()") xfrm policy objects are already free'd via rcu. In order to make more places lockless (i.e. use rcu_read_lock instead of grabbing read-side of policy rwlock) we only need to: - use rcu_assign_pointer to store address of new hash table backend memory - add rcu barrier so that freeing of old memory is delayed (expansion and free happens from system workqueue, so synchronize_rcu is fine) - use rcu_dereference to fetch current address of the hash table. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:53 +08:00
odst = rcu_dereference_protected(net->xfrm.policy_bydst[dir].table,
lockdep_is_held(&net->xfrm.xfrm_policy_lock));
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
for (i = hmask; i >= 0; i--)
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
xfrm_dst_hash_transfer(net, odst + i, ndst, nhashmask, dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: prepare policy_bydst hash for rcu lookups Since commit 56f047305dd4b6b617 ("xfrm: add rcu grace period in xfrm_policy_destroy()") xfrm policy objects are already free'd via rcu. In order to make more places lockless (i.e. use rcu_read_lock instead of grabbing read-side of policy rwlock) we only need to: - use rcu_assign_pointer to store address of new hash table backend memory - add rcu barrier so that freeing of old memory is delayed (expansion and free happens from system workqueue, so synchronize_rcu is fine) - use rcu_dereference to fetch current address of the hash table. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:53 +08:00
rcu_assign_pointer(net->xfrm.policy_bydst[dir].table, ndst);
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
net->xfrm.policy_bydst[dir].hmask = nhashmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
write_seqcount_end(&xfrm_policy_hash_generation);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: prepare policy_bydst hash for rcu lookups Since commit 56f047305dd4b6b617 ("xfrm: add rcu grace period in xfrm_policy_destroy()") xfrm policy objects are already free'd via rcu. In order to make more places lockless (i.e. use rcu_read_lock instead of grabbing read-side of policy rwlock) we only need to: - use rcu_assign_pointer to store address of new hash table backend memory - add rcu barrier so that freeing of old memory is delayed (expansion and free happens from system workqueue, so synchronize_rcu is fine) - use rcu_dereference to fetch current address of the hash table. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:53 +08:00
synchronize_rcu();
[XFRM]: Extract common hashing code into xfrm_hash.[ch] Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:50:50 +08:00
xfrm_hash_free(odst, (hmask + 1) * sizeof(struct hlist_head));
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
static void xfrm_byidx_resize(struct net *net, int total)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
unsigned int hmask = net->xfrm.policy_idx_hmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
unsigned int nhashmask = xfrm_new_hash_mask(hmask);
unsigned int nsize = (nhashmask + 1) * sizeof(struct hlist_head);
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
struct hlist_head *oidx = net->xfrm.policy_byidx;
[XFRM]: Extract common hashing code into xfrm_hash.[ch] Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:50:50 +08:00
struct hlist_head *nidx = xfrm_hash_alloc(nsize);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
int i;
if (!nidx)
return;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
for (i = hmask; i >= 0; i--)
xfrm_idx_hash_transfer(oidx + i, nidx, nhashmask);
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
net->xfrm.policy_byidx = nidx;
net->xfrm.policy_idx_hmask = nhashmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
[XFRM]: Extract common hashing code into xfrm_hash.[ch] Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:50:50 +08:00
xfrm_hash_free(oidx, (hmask + 1) * sizeof(struct hlist_head));
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
static inline int xfrm_bydst_should_resize(struct net *net, int dir, int *total)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
unsigned int cnt = net->xfrm.policy_count[dir];
unsigned int hmask = net->xfrm.policy_bydst[dir].hmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (total)
*total += cnt;
if ((hmask + 1) < xfrm_policy_hashmax &&
cnt > hmask)
return 1;
return 0;
}
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
static inline int xfrm_byidx_should_resize(struct net *net, int total)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
unsigned int hmask = net->xfrm.policy_idx_hmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if ((hmask + 1) < xfrm_policy_hashmax &&
total > hmask)
return 1;
return 0;
}
netns xfrm: fix "ip xfrm state|policy count" misreport "ip xfrm state|policy count" report SA/SP count from init_net, not from netns of caller process. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-23 21:37:10 +08:00
void xfrm_spd_getinfo(struct net *net, struct xfrmk_spdinfo *si)
[XFRM]: Export SPD info With this patch you can use iproute2 in user space to efficiently see how many policies exist in different directions. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-04-29 12:20:32 +08:00
{
netns xfrm: fix "ip xfrm state|policy count" misreport "ip xfrm state|policy count" report SA/SP count from init_net, not from netns of caller process. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-23 21:37:10 +08:00
si->incnt = net->xfrm.policy_count[XFRM_POLICY_IN];
si->outcnt = net->xfrm.policy_count[XFRM_POLICY_OUT];
si->fwdcnt = net->xfrm.policy_count[XFRM_POLICY_FWD];
si->inscnt = net->xfrm.policy_count[XFRM_POLICY_IN+XFRM_POLICY_MAX];
si->outscnt = net->xfrm.policy_count[XFRM_POLICY_OUT+XFRM_POLICY_MAX];
si->fwdscnt = net->xfrm.policy_count[XFRM_POLICY_FWD+XFRM_POLICY_MAX];
si->spdhcnt = net->xfrm.policy_idx_hmask;
[XFRM]: Export SPD info With this patch you can use iproute2 in user space to efficiently see how many policies exist in different directions. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-04-29 12:20:32 +08:00
si->spdhmcnt = xfrm_policy_hashmax;
}
EXPORT_SYMBOL(xfrm_spd_getinfo);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
[XFRM]: Export SPD info With this patch you can use iproute2 in user space to efficiently see how many policies exist in different directions. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-04-29 12:20:32 +08:00
static DEFINE_MUTEX(hash_resize_mutex);
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
static void xfrm_hash_resize(struct work_struct *work)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
{
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
struct net *net = container_of(work, struct net, xfrm.policy_hash_work);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
int dir, total;
mutex_lock(&hash_resize_mutex);
total = 0;
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
for (dir = 0; dir < XFRM_POLICY_MAX; dir++) {
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
if (xfrm_bydst_should_resize(net, dir, &total))
xfrm_bydst_resize(net, dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
if (xfrm_byidx_should_resize(net, total))
xfrm_byidx_resize(net, total);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
mutex_unlock(&hash_resize_mutex);
}
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
static void xfrm_hash_rebuild(struct work_struct *work)
{
struct net *net = container_of(work, struct net,
xfrm.policy_hthresh.work);
unsigned int hmask;
struct xfrm_policy *pol;
struct xfrm_policy *policy;
struct hlist_head *chain;
struct hlist_head *odst;
struct hlist_node *newpos;
int i;
int dir;
unsigned seq;
u8 lbits4, rbits4, lbits6, rbits6;
mutex_lock(&hash_resize_mutex);
/* read selector prefixlen thresholds */
do {
seq = read_seqbegin(&net->xfrm.policy_hthresh.lock);
lbits4 = net->xfrm.policy_hthresh.lbits4;
rbits4 = net->xfrm.policy_hthresh.rbits4;
lbits6 = net->xfrm.policy_hthresh.lbits6;
rbits6 = net->xfrm.policy_hthresh.rbits6;
} while (read_seqretry(&net->xfrm.policy_hthresh.lock, seq));
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
/* reset the bydst and inexact table in all directions */
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
for (dir = 0; dir < XFRM_POLICY_MAX; dir++) {
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
INIT_HLIST_HEAD(&net->xfrm.policy_inexact[dir]);
hmask = net->xfrm.policy_bydst[dir].hmask;
odst = net->xfrm.policy_bydst[dir].table;
for (i = hmask; i >= 0; i--)
INIT_HLIST_HEAD(odst + i);
if ((dir & XFRM_POLICY_MASK) == XFRM_POLICY_OUT) {
/* dir out => dst = remote, src = local */
net->xfrm.policy_bydst[dir].dbits4 = rbits4;
net->xfrm.policy_bydst[dir].sbits4 = lbits4;
net->xfrm.policy_bydst[dir].dbits6 = rbits6;
net->xfrm.policy_bydst[dir].sbits6 = lbits6;
} else {
/* dir in/fwd => dst = local, src = remote */
net->xfrm.policy_bydst[dir].dbits4 = lbits4;
net->xfrm.policy_bydst[dir].sbits4 = rbits4;
net->xfrm.policy_bydst[dir].dbits6 = lbits6;
net->xfrm.policy_bydst[dir].sbits6 = rbits6;
}
}
/* re-insert all policies by order of creation */
list_for_each_entry_reverse(policy, &net->xfrm.policy_all, walk.all) {
xfrm: skip policies marked as dead while rehashing syzkaller triggered following KASAN splat: BUG: KASAN: slab-out-of-bounds in xfrm_hash_rebuild+0xdbe/0xf00 net/xfrm/xfrm_policy.c:618 read of size 2 at addr ffff8801c8e92fe4 by task kworker/1:1/23 [..] Workqueue: events xfrm_hash_rebuild [..] __asan_report_load2_noabort+0x14/0x20 mm/kasan/report.c:428 xfrm_hash_rebuild+0xdbe/0xf00 net/xfrm/xfrm_policy.c:618 process_one_work+0xbbf/0x1b10 kernel/workqueue.c:2112 worker_thread+0x223/0x1990 kernel/workqueue.c:2246 [..] The reproducer triggers: 1016 if (error) { 1017 list_move_tail(&walk->walk.all, &x->all); 1018 goto out; 1019 } in xfrm_policy_walk() via pfkey (it sets tiny rcv space, dump callback returns -ENOBUFS). In this case, *walk is located the pfkey socket struct, so this socket becomes visible in the global policy list. It looks like this is intentional -- phony walker has walk.dead set to 1 and all other places skip such "policies". Ccing original authors of the two commits that seem to expose this issue (first patch missed ->dead check, second patch adds pfkey sockets to policies dumper list). Fixes: 880a6fab8f6ba5b ("xfrm: configure policy hash table thresholds by netlink") Fixes: 12a169e7d8f4b1c ("ipsec: Put dumpers on the dump list") Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Timo Teras <timo.teras@iki.fi> Cc: Christophe Gouault <christophe.gouault@6wind.com> Reported-by: syzbot <bot+c028095236fcb6f4348811565b75084c754dc729@syzkaller.appspotmail.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-12-28 06:25:45 +08:00
if (policy->walk.dead ||
xfrm_policy_id2dir(policy->index) >= XFRM_POLICY_MAX) {
xfrm: Ignore socket policies when rebuilding hash tables Whenever thresholds are changed the hash tables are rebuilt. This is done by enumerating all policies and hashing and inserting them into the right table according to the thresholds and direction. Because socket policies are also contained in net->xfrm.policy_all but no hash tables are defined for their direction (dir + XFRM_POLICY_MAX) this causes a NULL or invalid pointer dereference after returning from policy_hash_bysel() if the rebuild is done while any socket policies are installed. Since the rebuild after changing thresholds is scheduled this crash could even occur if the userland sets thresholds seemingly before installing any socket policies. Fixes: 53c2e285f970 ("xfrm: Do not hash socket policies") Signed-off-by: Tobias Brunner <tobias@strongswan.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-07-29 15:57:32 +08:00
/* skip socket policies */
continue;
}
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
newpos = NULL;
chain = policy_hash_bysel(net, &policy->selector,
policy->family,
xfrm_policy_id2dir(policy->index));
hlist_for_each_entry(pol, chain, bydst) {
if (policy->priority >= pol->priority)
newpos = &pol->bydst;
else
break;
}
if (newpos)
hlist_add_behind(&policy->bydst, newpos);
else
hlist_add_head(&policy->bydst, chain);
}
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
mutex_unlock(&hash_resize_mutex);
}
void xfrm_policy_hash_rebuild(struct net *net)
{
schedule_work(&net->xfrm.policy_hthresh.work);
}
EXPORT_SYMBOL(xfrm_policy_hash_rebuild);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* Generate new index... KAME seems to generate them ordered by cost
* of an absolute inpredictability of ordering of rules. This will not pass. */
xfrm: Try to honor policy index if it's supplied by user xfrm code always searches for unused policy index for newly created policy regardless whether or not user space policy index hint supplied. This patch enables such feature so that using "ip xfrm ... index=xxx" can be used by user to set specific policy index. Currently this beahvior is broken, so this patch make it happen as expected. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:48 +08:00
static u32 xfrm_gen_index(struct net *net, int dir, u32 index)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
static u32 idx_generator;
for (;;) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct hlist_head *list;
struct xfrm_policy *p;
u32 idx;
int found;
xfrm: Try to honor policy index if it's supplied by user xfrm code always searches for unused policy index for newly created policy regardless whether or not user space policy index hint supplied. This patch enables such feature so that using "ip xfrm ... index=xxx" can be used by user to set specific policy index. Currently this beahvior is broken, so this patch make it happen as expected. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:48 +08:00
if (!index) {
idx = (idx_generator | dir);
idx_generator += 8;
} else {
idx = index;
index = 0;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (idx == 0)
idx = 8;
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
list = net->xfrm.policy_byidx + idx_hash(net, idx);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
found = 0;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(p, list, byidx) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (p->index == idx) {
found = 1;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (!found)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return idx;
}
}
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
static inline int selector_cmp(struct xfrm_selector *s1, struct xfrm_selector *s2)
{
u32 *p1 = (u32 *) s1;
u32 *p2 = (u32 *) s2;
int len = sizeof(struct xfrm_selector) / sizeof(u32);
int i;
for (i = 0; i < len; i++) {
if (p1[i] != p2[i])
return 1;
}
return 0;
}
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
static void xfrm_policy_requeue(struct xfrm_policy *old,
struct xfrm_policy *new)
{
struct xfrm_policy_queue *pq = &old->polq;
struct sk_buff_head list;
xfrm: move the checking for old xfrm_policy hold_queue to beginning if hold_queue of old xfrm_policy is NULL, return directly, then not need to run other codes, especially take the spin lock Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-30 17:25:19 +08:00
if (skb_queue_empty(&pq->hold_queue))
return;
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
__skb_queue_head_init(&list);
spin_lock_bh(&pq->hold_queue.lock);
skb_queue_splice_init(&pq->hold_queue, &list);
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
if (del_timer(&pq->hold_timer))
xfrm_pol_put(old);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
spin_unlock_bh(&pq->hold_queue.lock);
pq = &new->polq;
spin_lock_bh(&pq->hold_queue.lock);
skb_queue_splice(&list, &pq->hold_queue);
pq->timeout = XFRM_QUEUE_TMO_MIN;
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
if (!mod_timer(&pq->hold_timer, jiffies))
xfrm_pol_hold(new);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
spin_unlock_bh(&pq->hold_queue.lock);
}
xfrm: Allow inserting policies with matching mark and different priorities We currently can not insert policies with mark and mask such that some flows would be matched from both policies. We make this possible when the priority of these policies are different. If both policies match a flow, the one with the higher priority is used. Reported-by: Emmanuel Thierry <emmanuel.thierry@telecom-bretagne.eu> Reported-by: Romain Kuntz <r.kuntz@ipflavors.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-11 14:02:36 +08:00
static bool xfrm_policy_mark_match(struct xfrm_policy *policy,
struct xfrm_policy *pol)
{
u32 mark = policy->mark.v & policy->mark.m;
if (policy->mark.v == pol->mark.v && policy->mark.m == pol->mark.m)
return true;
if ((mark & pol->mark.m) == pol->mark.v &&
policy->priority == pol->priority)
return true;
return false;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int xfrm_policy_insert(int dir, struct xfrm_policy *policy, int excl)
{
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
struct net *net = xp_net(policy);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol;
struct xfrm_policy *delpol;
struct hlist_head *chain;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
struct hlist_node *newpos;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
chain = policy_hash_bysel(net, &policy->selector, policy->family, dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
delpol = NULL;
newpos = NULL;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol, chain, bydst) {
[IPSEC]: Policy list disorder The recent hashing introduced an off-by-one bug in policy list insertion. Instead of adding after the last entry with a lesser or equal priority, we're adding after the successor of that entry. This patch fixes this and also adds a warning if we detect a duplicate entry in the policy list. This should never happen due to this if clause. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-01-17 08:52:02 +08:00
if (pol->type == policy->type &&
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
!selector_cmp(&pol->selector, &policy->selector) &&
xfrm: Allow inserting policies with matching mark and different priorities We currently can not insert policies with mark and mask such that some flows would be matched from both policies. We make this possible when the priority of these policies are different. If both policies match a flow, the one with the higher priority is used. Reported-by: Emmanuel Thierry <emmanuel.thierry@telecom-bretagne.eu> Reported-by: Romain Kuntz <r.kuntz@ipflavors.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-11 14:02:36 +08:00
xfrm_policy_mark_match(policy, pol) &&
[IPSEC]: Policy list disorder The recent hashing introduced an off-by-one bug in policy list insertion. Instead of adding after the last entry with a lesser or equal priority, we're adding after the successor of that entry. This patch fixes this and also adds a warning if we detect a duplicate entry in the policy list. This should never happen due to this if clause. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-01-17 08:52:02 +08:00
xfrm_sec_ctx_match(pol->security, policy->security) &&
!WARN_ON(delpol)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (excl) {
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -EEXIST;
}
delpol = pol;
if (policy->priority > pol->priority)
continue;
} else if (policy->priority >= pol->priority) {
[IPSEC]: Policy list disorder The recent hashing introduced an off-by-one bug in policy list insertion. Instead of adding after the last entry with a lesser or equal priority, we're adding after the successor of that entry. This patch fixes this and also adds a warning if we detect a duplicate entry in the policy list. This should never happen due to this if clause. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-01-17 08:52:02 +08:00
newpos = &pol->bydst;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
continue;
}
if (delpol)
break;
}
if (newpos)
list: fix order of arguments for hlist_add_after(_rcu) All other add functions for lists have the new item as first argument and the position where it is added as second argument. This was changed for no good reason in this function and makes using it unnecessary confusing. The name was changed to hlist_add_behind() to cause unconverted code to generate a compile error instead of using the wrong parameter order. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Ken Helias <kenhelias@firemail.de> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> [intel driver bits] Cc: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-07 07:09:16 +08:00
hlist_add_behind(&policy->bydst, newpos);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
else
hlist_add_head(&policy->bydst, chain);
xfrm: Use __xfrm_policy_link in xfrm_policy_insert For a long time we couldn't actually use __xfrm_policy_link in xfrm_policy_insert because the latter wanted to do hashing at a specific position. Now that __xfrm_policy_link no longer does hashing it can now be safely used in xfrm_policy_insert to kill some duplicate code, finally reuniting general policies with socket policies. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:50 +08:00
__xfrm_policy_link(policy, dir);
net: split rt_genid for ipv4 and ipv6 Current net name space has only one genid for both IPv4 and IPv6, it has below drawbacks: - Add/delete an IPv4 address will invalidate all IPv6 routing table entries. - Insert/remove XFRM policy will also invalidate both IPv4/IPv6 routing table entries even when the policy is only applied for one address family. Thus, this patch attempt to split one genid for two to cater for IPv4 and IPv6 separately in a fine granularity. Signed-off-by: Fan Du <fan.du@windriver.com> Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-07-30 08:33:53 +08:00
/* After previous checking, family can either be AF_INET or AF_INET6 */
if (policy->family == AF_INET)
rt_genid_bump_ipv4(net);
else
rt_genid_bump_ipv6(net);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
if (delpol) {
xfrm_policy_requeue(delpol, policy);
xfrm: Cleanup for unlink SPD entry Used __xfrm_policy_unlink() to instead of the dup codes when unlink SPD entry. Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-12-03 16:33:09 +08:00
__xfrm_policy_unlink(delpol, dir);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
}
xfrm: Try to honor policy index if it's supplied by user xfrm code always searches for unused policy index for newly created policy regardless whether or not user space policy index hint supplied. This patch enables such feature so that using "ip xfrm ... index=xxx" can be used by user to set specific policy index. Currently this beahvior is broken, so this patch make it happen as expected. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:48 +08:00
policy->index = delpol ? delpol->index : xfrm_gen_index(net, dir, policy->index);
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
hlist_add_head(&policy->byidx, net->xfrm.policy_byidx+idx_hash(net, policy->index));
[NET]: Convert xtime.tv_sec to get_seconds() Where appropriate, convert references to xtime.tv_sec to the get_seconds() helper function. Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-05 08:12:44 +08:00
policy->curlft.add_time = get_seconds();
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
policy->curlft.use_time = 0;
if (!mod_timer(&policy->timer, jiffies + HZ))
xfrm_pol_hold(policy);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IPSEC]: Fix policy updates missed by sockets The problem is that when new policies are inserted, sockets do not see the update (but all new route lookups do). This bug is related to the SA insertion stale route issue solved recently, and this policy visibility problem can be fixed in a similar way. The fix is to flush out the bundles of all policies deeper than the policy being inserted. Consider beginning state of "outgoing" direction policy list: policy A --> policy B --> policy C --> policy D First, realize that inserting a policy into a list only potentially changes IPSEC routes for that direction. Therefore we need not bother considering the policies for other directions. We need only consider the existing policies in the list we are doing the inserting. Consider new policy "B'", inserted after B. policy A --> policy B --> policy B' --> policy C --> policy D Two rules: 1) If policy A or policy B matched before the insertion, they appear before B' and thus would still match after inserting B' 2) Policy C and D, now "shadowed" and after policy B', potentially contain stale routes because policy B' might be selected instead of them. Therefore we only need flush routes assosciated with policies appearing after a newly inserted policy, if any. Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-22 23:39:48 +08:00
if (delpol)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_policy_kill(delpol);
netns xfrm: policy insertion in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:06 +08:00
else if (xfrm_bydst_should_resize(net, dir, NULL))
schedule_work(&net->xfrm.policy_hash_work);
[IPSEC]: Fix policy updates missed by sockets The problem is that when new policies are inserted, sockets do not see the update (but all new route lookups do). This bug is related to the SA insertion stale route issue solved recently, and this policy visibility problem can be fixed in a similar way. The fix is to flush out the bundles of all policies deeper than the policy being inserted. Consider beginning state of "outgoing" direction policy list: policy A --> policy B --> policy C --> policy D First, realize that inserting a policy into a list only potentially changes IPSEC routes for that direction. Therefore we need not bother considering the policies for other directions. We need only consider the existing policies in the list we are doing the inserting. Consider new policy "B'", inserted after B. policy A --> policy B --> policy B' --> policy C --> policy D Two rules: 1) If policy A or policy B matched before the insertion, they appear before B' and thus would still match after inserting B' 2) Policy C and D, now "shadowed" and after policy B', potentially contain stale routes because policy B' might be selected instead of them. Therefore we only need flush routes assosciated with policies appearing after a newly inserted policy, if any. Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-22 23:39:48 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 0;
}
EXPORT_SYMBOL(xfrm_policy_insert);
xfrm: SP lookups signature with mark pass mark to all SP lookups to prepare them for when we add code to have them search. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-22 19:32:57 +08:00
struct xfrm_policy *xfrm_policy_bysel_ctx(struct net *net, u32 mark, u8 type,
int dir, struct xfrm_selector *sel,
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
struct xfrm_sec_ctx *ctx, int delete,
int *err)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol, *ret;
struct hlist_head *chain;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
*err = 0;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
netns xfrm: finding policy in netns Add netns parameter to xfrm_policy_bysel_ctx(), xfrm_policy_byidx(). Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:34:20 +08:00
chain = policy_hash_bysel(net, sel, sel->family, dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
ret = NULL;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol, chain, bydst) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (pol->type == type &&
xfrm: SP lookups with mark Allow mark to be used when doing SP lookup Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-22 19:32:58 +08:00
(mark & pol->mark.m) == pol->mark.v &&
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
!selector_cmp(sel, &pol->selector) &&
xfrm_sec_ctx_match(ctx, pol->security)) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_pol_hold(pol);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (delete) {
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
*err = security_xfrm_policy_delete(
pol->security);
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
if (*err) {
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
return pol;
}
xfrm: Cleanup for unlink SPD entry Used __xfrm_policy_unlink() to instead of the dup codes when unlink SPD entry. Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-12-03 16:33:09 +08:00
__xfrm_policy_unlink(pol, dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
ret = pol;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
}
}
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
flow: virtualize flow cache entry methods This allows to validate the cached object before returning it. It also allows to destruct object properly, if the last reference was held in flow cache. This is also a prepartion for caching bundles in the flow cache. In return for virtualizing the methods, we save on: - not having to regenerate the whole flow cache on policy removal: each flow matching a killed policy gets refreshed as the getter function notices it smartly. - we do not have to call flow_cache_flush from policy gc, since the flow cache now properly deletes the object if it had any references Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:04 +08:00
if (ret && delete)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm_policy_kill(ret);
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[LSM-IPSec]: Security association restriction. This patch series implements per packet access control via the extension of the Linux Security Modules (LSM) interface by hooks in the XFRM and pfkey subsystems that leverage IPSec security associations to label packets. Extensions to the SELinux LSM are included that leverage the patch for this purpose. This patch implements the changes necessary to the XFRM subsystem, pfkey interface, ipv4/ipv6, and xfrm_user interface to restrict a socket to use only authorized security associations (or no security association) to send/receive network packets. Patch purpose: The patch is designed to enable access control per packets based on the strongly authenticated IPSec security association. Such access controls augment the existing ones based on network interface and IP address. The former are very coarse-grained, and the latter can be spoofed. By using IPSec, the system can control access to remote hosts based on cryptographic keys generated using the IPSec mechanism. This enables access control on a per-machine basis or per-application if the remote machine is running the same mechanism and trusted to enforce the access control policy. Patch design approach: The overall approach is that policy (xfrm_policy) entries set by user-level programs (e.g., setkey for ipsec-tools) are extended with a security context that is used at policy selection time in the XFRM subsystem to restrict the sockets that can send/receive packets via security associations (xfrm_states) that are built from those policies. A presentation available at www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf from the SELinux symposium describes the overall approach. Patch implementation details: On output, the policy retrieved (via xfrm_policy_lookup or xfrm_sk_policy_lookup) must be authorized for the security context of the socket and the same security context is required for resultant security association (retrieved or negotiated via racoon in ipsec-tools). This is enforced in xfrm_state_find. On input, the policy retrieved must also be authorized for the socket (at __xfrm_policy_check), and the security context of the policy must also match the security association being used. The patch has virtually no impact on packets that do not use IPSec. The existing Netfilter (outgoing) and LSM rcv_skb hooks are used as before. Also, if IPSec is used without security contexts, the impact is minimal. The LSM must allow such policies to be selected for the combination of socket and remote machine, but subsequent IPSec processing proceeds as in the original case. Testing: The pfkey interface is tested using the ipsec-tools. ipsec-tools have been modified (a separate ipsec-tools patch is available for version 0.5) that supports assignment of xfrm_policy entries and security associations with security contexts via setkey and the negotiation using the security contexts via racoon. The xfrm_user interface is tested via ad hoc programs that set security contexts. These programs are also available from me, and contain programs for setting, getting, and deleting policy for testing this interface. Testing of sa functions was done by tracing kernel behavior. Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-14 15:12:27 +08:00
EXPORT_SYMBOL(xfrm_policy_bysel_ctx);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: SP lookups signature with mark pass mark to all SP lookups to prepare them for when we add code to have them search. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-22 19:32:57 +08:00
struct xfrm_policy *xfrm_policy_byid(struct net *net, u32 mark, u8 type,
int dir, u32 id, int delete, int *err)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol, *ret;
struct hlist_head *chain;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IPSEC]: Check validity of direction in xfrm_policy_byid The function xfrm_policy_byid takes a dir argument but finds the policy using the index instead. We only use the dir argument to update the policy count for that direction. Since the user can supply any value for dir, this can corrupt our policy count. I know this is the problem because a few days ago I was deleting policies by hand using indicies and accidentally typed in the wrong direction. It still deleted the policy and at the time I thought that was cool. In retrospect it isn't such a good idea :) I decided against letting it delete the policy anyway just in case we ever remove the connection between indicies and direction. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-05-14 17:15:47 +08:00
*err = -ENOENT;
if (xfrm_policy_id2dir(id) != dir)
return NULL;
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
*err = 0;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
netns xfrm: finding policy in netns Add netns parameter to xfrm_policy_bysel_ctx(), xfrm_policy_byidx(). Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:34:20 +08:00
chain = net->xfrm.policy_byidx + idx_hash(net, id);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
ret = NULL;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol, chain, byidx) {
xfrm: SP lookups with mark Allow mark to be used when doing SP lookup Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-22 19:32:58 +08:00
if (pol->type == type && pol->index == id &&
(mark & pol->mark.m) == pol->mark.v) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_pol_hold(pol);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (delete) {
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
*err = security_xfrm_policy_delete(
pol->security);
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
if (*err) {
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
[IPSEC]: xfrm_policy delete security check misplaced The security hooks to check permissions to remove an xfrm_policy were actually done after the policy was removed. Since the unlinking and deletion are done in xfrm_policy_by* functions this moves the hooks inside those 2 functions. There we have all the information needed to do the security check and it can be done before the deletion. Since auditing requires the result of that security check err has to be passed back and forth from the xfrm_policy_by* functions. This patch also fixes a bug where a deletion that failed the security check could cause improper accounting on the xfrm_policy (xfrm_get_policy didn't have a put on the exit path for the hold taken by xfrm_policy_by*) It also fixes the return code when no policy is found in xfrm_add_pol_expire. In old code (at least back in the 2.6.18 days) err wasn't used before the return when no policy is found and so the initialization would cause err to be ENOENT. But since err has since been used above when we don't get a policy back from the xfrm_policy_by* function we would always return 0 instead of the intended ENOENT. Also fixed some white space damage in the same area. Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Venkat Yekkirala <vyekkirala@trustedcs.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-08 07:37:58 +08:00
return pol;
}
xfrm: Cleanup for unlink SPD entry Used __xfrm_policy_unlink() to instead of the dup codes when unlink SPD entry. Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-12-03 16:33:09 +08:00
__xfrm_policy_unlink(pol, dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
ret = pol;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
}
}
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
flow: virtualize flow cache entry methods This allows to validate the cached object before returning it. It also allows to destruct object properly, if the last reference was held in flow cache. This is also a prepartion for caching bundles in the flow cache. In return for virtualizing the methods, we save on: - not having to regenerate the whole flow cache on policy removal: each flow matching a killed policy gets refreshed as the getter function notices it smartly. - we do not have to call flow_cache_flush from policy gc, since the flow cache now properly deletes the object if it had any references Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:04 +08:00
if (ret && delete)
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm_policy_kill(ret);
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(xfrm_policy_byid);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
#ifdef CONFIG_SECURITY_NETWORK_XFRM
static inline int
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_policy_flush_secctx_check(struct net *net, u8 type, bool task_valid)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
int dir, err = 0;
for (dir = 0; dir < XFRM_POLICY_MAX; dir++) {
struct xfrm_policy *pol;
int i;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol,
netns xfrm: policy flushing in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:32 +08:00
&net->xfrm.policy_inexact[dir], bydst) {
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
if (pol->type != type)
continue;
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
err = security_xfrm_policy_delete(pol->security);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
if (err) {
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_audit_policy_delete(pol, 0, task_valid);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
return err;
}
[NET] XFRM: Fix whitespace errors. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2007-07-19 09:45:15 +08:00
}
netns xfrm: policy flushing in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:32 +08:00
for (i = net->xfrm.policy_bydst[dir].hmask; i >= 0; i--) {
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol,
netns xfrm: policy flushing in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:32 +08:00
net->xfrm.policy_bydst[dir].table + i,
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
bydst) {
if (pol->type != type)
continue;
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
err = security_xfrm_policy_delete(
pol->security);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
if (err) {
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
xfrm_audit_policy_delete(pol, 0,
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
task_valid);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
return err;
}
}
}
}
return err;
}
#else
static inline int
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_policy_flush_secctx_check(struct net *net, u8 type, bool task_valid)
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
{
return 0;
}
#endif
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
int xfrm_policy_flush(struct net *net, u8 type, bool task_valid)
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
{
xfrm: Flushing empty SPD generates false events To see the effect make sure you have an empty SPD. On window1 "ip xfrm mon" and on window2 issue "ip xfrm policy flush" You get prompt back in window2 and you see the flush event on window1. With this fix, you still get prompt on window1 but no event on window2. Thanks to Alexey Dobriyan for finding a bug in earlier version when using pfkey to do the flushing. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-19 10:00:42 +08:00
int dir, err = 0, cnt = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
err = xfrm_policy_flush_secctx_check(net, type, task_valid);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
if (err)
goto out;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
for (dir = 0; dir < XFRM_POLICY_MAX; dir++) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol;
xfrm: Cleanup for unlink SPD entry Used __xfrm_policy_unlink() to instead of the dup codes when unlink SPD entry. Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-12-03 16:33:09 +08:00
int i;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
again1:
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol,
netns xfrm: policy flushing in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:32 +08:00
&net->xfrm.policy_inexact[dir], bydst) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (pol->type != type)
continue;
xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-31 08:17:05 +08:00
__xfrm_policy_unlink(pol, dir);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-31 08:17:05 +08:00
cnt++;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_audit_policy_delete(pol, 1, task_valid);
audit: Add auditing to ipsec An audit message occurs when an ipsec SA or ipsec policy is created/deleted. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-11-28 03:11:54 +08:00
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm_policy_kill(pol);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Fix xfrm_policy_lock imbalance An earlier patch accidentally replaced a write_lock_bh with a spin_unlock_bh. Fix this by using spin_lock_bh instead. Fixes: 9d0380df6217 ("xfrm: policy: convert policy_lock to spinlock") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-24 19:08:40 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
goto again1;
}
netns xfrm: policy flushing in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:32 +08:00
for (i = net->xfrm.policy_bydst[dir].hmask; i >= 0; i--) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
again2:
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol,
netns xfrm: policy flushing in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:33:32 +08:00
net->xfrm.policy_bydst[dir].table + i,
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
bydst) {
if (pol->type != type)
continue;
xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-31 08:17:05 +08:00
__xfrm_policy_unlink(pol, dir);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-31 08:17:05 +08:00
cnt++;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_audit_policy_delete(pol, 1, task_valid);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm_policy_kill(pol);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
goto again2;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: Flushing empty SPD generates false events To see the effect make sure you have an empty SPD. On window1 "ip xfrm mon" and on window2 issue "ip xfrm policy flush" You get prompt back in window2 and you see the flush event on window1. With this fix, you still get prompt on window1 but no event on window2. Thanks to Alexey Dobriyan for finding a bug in earlier version when using pfkey to do the flushing. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-19 10:00:42 +08:00
if (!cnt)
err = -ESRCH;
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
out:
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: Add security check before flushing SAD/SPD Currently we check for permission before deleting entries from SAD and SPD, (see security_xfrm_policy_delete() security_xfrm_state_delete()) However we are not checking for authorization when flushing the SPD and the SAD completely. It was perhaps missed in the original security hooks patch. This patch adds a security check when flushing entries from the SAD and SPD. It runs the entire database and checks each entry for a denial. If the process attempting the flush is unable to remove all of the entries a denial is logged the the flush function returns an error without removing anything. This is particularly useful when a process may need to create or delete its own xfrm entries used for things like labeled networking but that same process should not be able to delete other entries or flush the entire database. Signed-off-by: Joy Latten<latten@austin.ibm.com> Signed-off-by: Eric Paris <eparis@parisplace.org> Signed-off-by: James Morris <jmorris@namei.org>
2007-06-05 07:05:57 +08:00
return err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(xfrm_policy_flush);
netns xfrm: policy walking in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:34:49 +08:00
int xfrm_policy_walk(struct net *net, struct xfrm_policy_walk *walk,
[XFRM]: Speed up xfrm_policy and xfrm_state walking Change xfrm_policy and xfrm_state walking algorithm from O(n^2) to O(n). This is achieved adding the entries to one more list which is used solely for walking the entries. This also fixes some races where the dump can have duplicate or missing entries when the SPD/SADB is modified during an ongoing dump. Dumping SADB with 20000 entries using "time ip xfrm state" the sys time dropped from 1.012s to 0.080s. Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-29 13:31:08 +08:00
int (*func)(struct xfrm_policy *, int, int, void*),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void *data)
{
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
struct xfrm_policy *pol;
struct xfrm_policy_walk_entry *x;
[XFRM]: Speed up xfrm_policy and xfrm_state walking Change xfrm_policy and xfrm_state walking algorithm from O(n^2) to O(n). This is achieved adding the entries to one more list which is used solely for walking the entries. This also fixes some races where the dump can have duplicate or missing entries when the SPD/SADB is modified during an ongoing dump. Dumping SADB with 20000 entries using "time ip xfrm state" the sys time dropped from 1.012s to 0.080s. Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-29 13:31:08 +08:00
int error = 0;
if (walk->type >= XFRM_POLICY_TYPE_MAX &&
walk->type != XFRM_POLICY_TYPE_ANY)
return -EINVAL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
if (list_empty(&walk->walk.all) && walk->seq != 0)
[XFRM]: Speed up xfrm_policy and xfrm_state walking Change xfrm_policy and xfrm_state walking algorithm from O(n^2) to O(n). This is achieved adding the entries to one more list which is used solely for walking the entries. This also fixes some races where the dump can have duplicate or missing entries when the SPD/SADB is modified during an ongoing dump. Dumping SADB with 20000 entries using "time ip xfrm state" the sys time dropped from 1.012s to 0.080s. Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-29 13:31:08 +08:00
return 0;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
if (list_empty(&walk->walk.all))
netns xfrm: policy walking in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:34:49 +08:00
x = list_first_entry(&net->xfrm.policy_all, struct xfrm_policy_walk_entry, all);
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
else
xfrm: optimise the use of walk list header in xfrm_policy/state_walk The walk from input is the list header, and marked as dead, and will be skipped in loop. list_first_entry() can be used to return the true usable value from walk if walk is not empty Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-22 17:09:54 +08:00
x = list_first_entry(&walk->walk.all,
struct xfrm_policy_walk_entry, all);
netns xfrm: policy walking in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:34:49 +08:00
list_for_each_entry_from(x, &net->xfrm.policy_all, all) {
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
if (x->dead)
[XFRM]: Speed up xfrm_policy and xfrm_state walking Change xfrm_policy and xfrm_state walking algorithm from O(n^2) to O(n). This is achieved adding the entries to one more list which is used solely for walking the entries. This also fixes some races where the dump can have duplicate or missing entries when the SPD/SADB is modified during an ongoing dump. Dumping SADB with 20000 entries using "time ip xfrm state" the sys time dropped from 1.012s to 0.080s. Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-02-29 13:31:08 +08:00
continue;
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
pol = container_of(x, struct xfrm_policy, walk);
if (walk->type != XFRM_POLICY_TYPE_ANY &&
walk->type != pol->type)
continue;
error = func(pol, xfrm_policy_id2dir(pol->index),
walk->seq, data);
if (error) {
list_move_tail(&walk->walk.all, &x->all);
goto out;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
walk->seq++;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
if (walk->seq == 0) {
[XFRM] Optimize policy dumping This change optimizes the dumping of Security policies. 1) Before this change .. speedopolis:~# time ./ip xf pol real 0m22.274s user 0m0.000s sys 0m22.269s 2) Turn off sub-policies speedopolis:~# ./ip xf pol real 0m13.496s user 0m0.000s sys 0m13.493s i suppose the above is to be expected 3) With this change .. speedopolis:~# time ./ip x policy real 0m7.901s user 0m0.008s sys 0m7.896s
2006-12-05 12:02:37 +08:00
error = -ENOENT;
goto out;
}
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
list_del_init(&walk->walk.all);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
out:
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return error;
}
EXPORT_SYMBOL(xfrm_policy_walk);
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
void xfrm_policy_walk_init(struct xfrm_policy_walk *walk, u8 type)
{
INIT_LIST_HEAD(&walk->walk.all);
walk->walk.dead = 1;
walk->type = type;
walk->seq = 0;
}
EXPORT_SYMBOL(xfrm_policy_walk_init);
xfrm: Namespacify xfrm state/policy locks By semantics, xfrm layer is fully name space aware, so will the locks, e.g. xfrm_state/pocliy_lock. Ensure exclusive access into state/policy link list for different name space with one global lock is not right in terms of semantics aspect at first place, as they are indeed mutually independent with each other, but also more seriously causes scalability problem. One practical scenario is on a Open Network Stack, more than hundreds of lxc tenants acts as routers within one host, a global xfrm_state/policy_lock becomes the bottleneck. But onces those locks are decoupled in a per-namespace fashion, locks contend is just with in specific name space scope, without causing additional SPD/SAD access delay for other name space. Also this patch improve scalability while as without changing original xfrm behavior. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:50 +08:00
void xfrm_policy_walk_done(struct xfrm_policy_walk *walk, struct net *net)
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
{
if (list_empty(&walk->walk.all))
return;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock); /*FIXME where is net? */
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
list_del(&walk->walk.all);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
}
EXPORT_SYMBOL(xfrm_policy_walk_done);
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
/*
* Find policy to apply to this flow.
*
* Returns 0 if policy found, else an -errno.
*/
xfrm: Const'ify policy arg and local selector in xfrm_policy_match. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:23:30 +08:00
static int xfrm_policy_match(const struct xfrm_policy *pol,
const struct flowi *fl,
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
u8 type, u16 family, int dir)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: Const'ify policy arg and local selector in xfrm_policy_match. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:23:30 +08:00
const struct xfrm_selector *sel = &pol->selector;
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
int ret = -ESRCH;
bool match;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (pol->family != family ||
net: Put flowi_* prefix on AF independent members of struct flowi I intend to turn struct flowi into a union of AF specific flowi structs. There will be a common structure that each variant includes first, much like struct sock_common. This is the first step to move in that direction. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 13:29:39 +08:00
(fl->flowi_mark & pol->mark.m) != pol->mark.v ||
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
pol->type != type)
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
match = xfrm_selector_match(sel, fl, family);
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
if (match)
net: Put flowi_* prefix on AF independent members of struct flowi I intend to turn struct flowi into a union of AF specific flowi structs. There will be a common structure that each variant includes first, much like struct sock_common. This is the first step to move in that direction. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 13:29:39 +08:00
ret = security_xfrm_policy_lookup(pol->security, fl->flowi_secid,
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
dir);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
return ret;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
netns xfrm: lookup in netns Pass netns to xfrm_lookup()/__xfrm_lookup(). For that pass netns to flow_cache_lookup() and resolver callback. Take it from socket or netdevice. Stub DECnet to init_net. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:35:18 +08:00
static struct xfrm_policy *xfrm_policy_lookup_bytype(struct net *net, u8 type,
xfrm: Mark flowi arg to xfrm_policy_{lookup_by_type,match}() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 10:31:08 +08:00
const struct flowi *fl,
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
u16 family, u8 dir)
{
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
int err;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy *pol, *ret;
xfrm: Const'ify local xfrm_address_t pointers in xfrm_policy_lookup_bytype. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:22:48 +08:00
const xfrm_address_t *daddr, *saddr;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct hlist_head *chain;
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
unsigned int sequence;
u32 priority;
[LSM-IPSec]: Security association restriction. This patch series implements per packet access control via the extension of the Linux Security Modules (LSM) interface by hooks in the XFRM and pfkey subsystems that leverage IPSec security associations to label packets. Extensions to the SELinux LSM are included that leverage the patch for this purpose. This patch implements the changes necessary to the XFRM subsystem, pfkey interface, ipv4/ipv6, and xfrm_user interface to restrict a socket to use only authorized security associations (or no security association) to send/receive network packets. Patch purpose: The patch is designed to enable access control per packets based on the strongly authenticated IPSec security association. Such access controls augment the existing ones based on network interface and IP address. The former are very coarse-grained, and the latter can be spoofed. By using IPSec, the system can control access to remote hosts based on cryptographic keys generated using the IPSec mechanism. This enables access control on a per-machine basis or per-application if the remote machine is running the same mechanism and trusted to enforce the access control policy. Patch design approach: The overall approach is that policy (xfrm_policy) entries set by user-level programs (e.g., setkey for ipsec-tools) are extended with a security context that is used at policy selection time in the XFRM subsystem to restrict the sockets that can send/receive packets via security associations (xfrm_states) that are built from those policies. A presentation available at www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf from the SELinux symposium describes the overall approach. Patch implementation details: On output, the policy retrieved (via xfrm_policy_lookup or xfrm_sk_policy_lookup) must be authorized for the security context of the socket and the same security context is required for resultant security association (retrieved or negotiated via racoon in ipsec-tools). This is enforced in xfrm_state_find. On input, the policy retrieved must also be authorized for the socket (at __xfrm_policy_check), and the security context of the policy must also match the security association being used. The patch has virtually no impact on packets that do not use IPSec. The existing Netfilter (outgoing) and LSM rcv_skb hooks are used as before. Also, if IPSec is used without security contexts, the impact is minimal. The LSM must allow such policies to be selected for the combination of socket and remote machine, but subsequent IPSec processing proceeds as in the original case. Testing: The pfkey interface is tested using the ipsec-tools. ipsec-tools have been modified (a separate ipsec-tools patch is available for version 0.5) that supports assignment of xfrm_policy entries and security associations with security contexts via setkey and the negotiation using the security contexts via racoon. The xfrm_user interface is tested via ad hoc programs that set security contexts. These programs are also available from me, and contain programs for setting, getting, and deleting policy for testing this interface. Testing of sa functions was done by tracing kernel behavior. Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-14 15:12:27 +08:00
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
daddr = xfrm_flowi_daddr(fl, family);
saddr = xfrm_flowi_saddr(fl, family);
if (unlikely(!daddr || !saddr))
return NULL;
xfrm: policy: make xfrm_policy_lookup_bytype lockless side effect: no longer disables BH (should be fine). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:56 +08:00
rcu_read_lock();
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
retry:
do {
sequence = read_seqcount_begin(&xfrm_policy_hash_generation);
chain = policy_hash_direct(net, daddr, saddr, family, dir);
} while (read_seqcount_retry(&xfrm_policy_hash_generation, sequence));
priority = ~0U;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
ret = NULL;
xfrm: policy: use rcu versions for iteration and list add/del This is required once we allow lockless readers. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:52 +08:00
hlist_for_each_entry_rcu(pol, chain, bydst) {
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
err = xfrm_policy_match(pol, fl, type, family, dir);
if (err) {
if (err == -ESRCH)
continue;
else {
ret = ERR_PTR(err);
goto fail;
}
} else {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
ret = pol;
[XFRM]: Respect priority in policy lookups. Even if we find an exact match in the hash table, we must inspect the inexact list to look for a match with a better priority. Noticed by Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-26 06:46:46 +08:00
priority = ret->priority;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
break;
}
}
netns xfrm: lookup in netns Pass netns to xfrm_lookup()/__xfrm_lookup(). For that pass netns to flow_cache_lookup() and resolver callback. Take it from socket or netdevice. Stub DECnet to init_net. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:35:18 +08:00
chain = &net->xfrm.policy_inexact[dir];
xfrm: policy: use rcu versions for iteration and list add/del This is required once we allow lockless readers. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:52 +08:00
hlist_for_each_entry_rcu(pol, chain, bydst) {
xfrm: optimise to search the inexact policy list The policies are organized into list by priority ascent of policy, so it is unnecessary to continue to loop the policy if the priority of current looped police is larger than or equal priority which is from the policy_bydst list. This allows to match policy with ~0U priority in inexact list too. Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-05-14 11:16:59 +08:00
if ((pol->priority >= priority) && ret)
break;
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
err = xfrm_policy_match(pol, fl, type, family, dir);
if (err) {
if (err == -ESRCH)
continue;
else {
ret = ERR_PTR(err);
goto fail;
}
xfrm: optimise to search the inexact policy list The policies are organized into list by priority ascent of policy, so it is unnecessary to continue to loop the policy if the priority of current looped police is larger than or equal priority which is from the policy_bydst list. This allows to match policy with ~0U priority in inexact list too. Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-05-14 11:16:59 +08:00
} else {
[XFRM]: Respect priority in policy lookups. Even if we find an exact match in the hash table, we must inspect the inexact list to look for a match with a better priority. Noticed by Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-26 06:46:46 +08:00
ret = pol;
break;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
}
xfrm: remove the unnecessary checking before call xfrm_pol_hold xfrm_pol_hold will check its input with NULL Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-30 17:13:41 +08:00
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
if (read_seqcount_retry(&xfrm_policy_hash_generation, sequence))
goto retry;
xfrm: policy: use atomic_inc_not_zero in rcu section If we don't hold the policy lock anymore the refcnt might already be 0, i.e. policy struct is about to be free'd. Switch to atomic_inc_not_zero to avoid this. On removal policies are already unlinked from the tables (lists) before the last _put occurs so we are not supposed to find the same 'dead' entry on the next loop, so its safe to just repeat the lookup. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:55 +08:00
if (ret && !xfrm_pol_hold_rcu(ret))
goto retry;
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
fail:
xfrm: policy: make xfrm_policy_lookup_bytype lockless side effect: no longer disables BH (should be fine). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:56 +08:00
rcu_read_unlock();
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
return ret;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
static struct xfrm_policy *
xfrm_policy: remove xfrm_policy_lookup This removes the wrapper and renames the __xfrm_policy_lookup variant to get rid of another place that used flow cache objects. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:24 +08:00
xfrm_policy_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir)
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
{
#ifdef CONFIG_XFRM_SUB_POLICY
struct xfrm_policy *pol;
pol = xfrm_policy_lookup_bytype(net, XFRM_POLICY_TYPE_SUB, fl, family, dir);
if (pol != NULL)
return pol;
#endif
return xfrm_policy_lookup_bytype(net, XFRM_POLICY_TYPE_MAIN, fl, family, dir);
}
inet: constify ip_route_output_flow() socket argument Very soon, TCP stack might call inet_csk_route_req(), which calls inet_csk_route_req() with an unlocked listener socket, so we need to make sure ip_route_output_flow() is not trying to change any field from its socket argument. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-25 22:39:10 +08:00
static struct xfrm_policy *xfrm_sk_policy_lookup(const struct sock *sk, int dir,
xfrm: Don't use sk_family for socket policy lookups On IPv4-mapped IPv6 addresses sk_family is AF_INET6, but the flow informations are created based on AF_INET. So the routing set up 'struct flowi4' but we try to access 'struct flowi6' what leads to an out of bounds access. Fix this by using the family we get with the dst_entry, like we do it for the standard policy lookup. Reported-by: Dmitry Vyukov <dvyukov@google.com> Tested-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-14 14:43:56 +08:00
const struct flowi *fl, u16 family)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct xfrm_policy *pol;
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
rcu_read_lock();
xfrm: policy: only use rcu in xfrm_sk_policy_lookup Don't acquire the readlock anymore and rely on rcu alone. In case writer on other CPU changed policy at the wrong moment (after we obtained sk policy pointer but before we could obtain the reference) just repeat the lookup. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:57 +08:00
again:
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
pol = rcu_dereference(sk->sk_policy[dir]);
if (pol != NULL) {
xfrm: Fix stack-out-of-bounds read on socket policy lookup. When we do tunnel or beet mode, we pass saddr and daddr from the template to xfrm_state_find(), this is ok. On transport mode, we pass the addresses from the flowi, assuming that the IP addresses (and address family) don't change during transformation. This assumption is wrong in the IPv4 mapped IPv6 case, packet is IPv4 and template is IPv6. Fix this by catching address family missmatches of the policy and the flow already before we do the lookup. Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-29 13:53:55 +08:00
bool match;
[NET] XFRM: Fix whitespace errors. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 22:25:29 +08:00
int err = 0;
[LSM-IPSec]: Security association restriction. This patch series implements per packet access control via the extension of the Linux Security Modules (LSM) interface by hooks in the XFRM and pfkey subsystems that leverage IPSec security associations to label packets. Extensions to the SELinux LSM are included that leverage the patch for this purpose. This patch implements the changes necessary to the XFRM subsystem, pfkey interface, ipv4/ipv6, and xfrm_user interface to restrict a socket to use only authorized security associations (or no security association) to send/receive network packets. Patch purpose: The patch is designed to enable access control per packets based on the strongly authenticated IPSec security association. Such access controls augment the existing ones based on network interface and IP address. The former are very coarse-grained, and the latter can be spoofed. By using IPSec, the system can control access to remote hosts based on cryptographic keys generated using the IPSec mechanism. This enables access control on a per-machine basis or per-application if the remote machine is running the same mechanism and trusted to enforce the access control policy. Patch design approach: The overall approach is that policy (xfrm_policy) entries set by user-level programs (e.g., setkey for ipsec-tools) are extended with a security context that is used at policy selection time in the XFRM subsystem to restrict the sockets that can send/receive packets via security associations (xfrm_states) that are built from those policies. A presentation available at www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf from the SELinux symposium describes the overall approach. Patch implementation details: On output, the policy retrieved (via xfrm_policy_lookup or xfrm_sk_policy_lookup) must be authorized for the security context of the socket and the same security context is required for resultant security association (retrieved or negotiated via racoon in ipsec-tools). This is enforced in xfrm_state_find. On input, the policy retrieved must also be authorized for the socket (at __xfrm_policy_check), and the security context of the policy must also match the security association being used. The patch has virtually no impact on packets that do not use IPSec. The existing Netfilter (outgoing) and LSM rcv_skb hooks are used as before. Also, if IPSec is used without security contexts, the impact is minimal. The LSM must allow such policies to be selected for the combination of socket and remote machine, but subsequent IPSec processing proceeds as in the original case. Testing: The pfkey interface is tested using the ipsec-tools. ipsec-tools have been modified (a separate ipsec-tools patch is available for version 0.5) that supports assignment of xfrm_policy entries and security associations with security contexts via setkey and the negotiation using the security contexts via racoon. The xfrm_user interface is tested via ad hoc programs that set security contexts. These programs are also available from me, and contain programs for setting, getting, and deleting policy for testing this interface. Testing of sa functions was done by tracing kernel behavior. Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-14 15:12:27 +08:00
xfrm: Fix stack-out-of-bounds read on socket policy lookup. When we do tunnel or beet mode, we pass saddr and daddr from the template to xfrm_state_find(), this is ok. On transport mode, we pass the addresses from the flowi, assuming that the IP addresses (and address family) don't change during transformation. This assumption is wrong in the IPv4 mapped IPv6 case, packet is IPv4 and template is IPv6. Fix this by catching address family missmatches of the policy and the flow already before we do the lookup. Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-29 13:53:55 +08:00
if (pol->family != family) {
pol = NULL;
goto out;
}
match = xfrm_selector_match(&pol->selector, fl, family);
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
if (match) {
xfrm: SP lookups with mark Allow mark to be used when doing SP lookup Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-22 19:32:58 +08:00
if ((sk->sk_mark & pol->mark.m) != pol->mark.v) {
pol = NULL;
goto out;
}
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
err = security_xfrm_policy_lookup(pol->security,
net: Put flowi_* prefix on AF independent members of struct flowi I intend to turn struct flowi into a union of AF specific flowi structs. There will be a common structure that each variant includes first, much like struct sock_common. This is the first step to move in that direction. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 13:29:39 +08:00
fl->flowi_secid,
xfrm_policy: kill flow to policy dir conversion XFRM_POLICY_IN/OUT/FWD are identical to FLOW_DIR_*, so gcc already removed this function as its just returns the argument. Again, no code change. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:23 +08:00
dir);
xfrm: unbreak xfrm_sk_policy_lookup if we succeed grabbing the refcount, then if (err && !xfrm_pol_hold_rcu) will evaluate to false so this hits last else branch which then sets policy to ERR_PTR(0). Fixes: ae33786f73a7ce ("xfrm: policy: only use rcu in xfrm_sk_policy_lookup") Reported-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Tested-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-11-17 20:21:46 +08:00
if (!err) {
if (!xfrm_pol_hold_rcu(pol))
goto again;
} else if (err == -ESRCH) {
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
pol = NULL;
xfrm: unbreak xfrm_sk_policy_lookup if we succeed grabbing the refcount, then if (err && !xfrm_pol_hold_rcu) will evaluate to false so this hits last else branch which then sets policy to ERR_PTR(0). Fixes: ae33786f73a7ce ("xfrm: policy: only use rcu in xfrm_sk_policy_lookup") Reported-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Tested-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-11-17 20:21:46 +08:00
} else {
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
pol = ERR_PTR(err);
xfrm: unbreak xfrm_sk_policy_lookup if we succeed grabbing the refcount, then if (err && !xfrm_pol_hold_rcu) will evaluate to false so this hits last else branch which then sets policy to ERR_PTR(0). Fixes: ae33786f73a7ce ("xfrm: policy: only use rcu in xfrm_sk_policy_lookup") Reported-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Tested-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-11-17 20:21:46 +08:00
}
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
} else
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
pol = NULL;
}
xfrm: SP lookups with mark Allow mark to be used when doing SP lookup Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-22 19:32:58 +08:00
out:
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
rcu_read_unlock();
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return pol;
}
static void __xfrm_policy_link(struct xfrm_policy *pol, int dir)
{
netns xfrm: trivial netns propagations Take netns from xfrm_state or xfrm_policy. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:29:47 +08:00
struct net *net = xp_net(pol);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
netns xfrm: trivial netns propagations Take netns from xfrm_state or xfrm_policy. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:29:47 +08:00
list_add(&pol->walk.all, &net->xfrm.policy_all);
net->xfrm.policy_count[dir]++;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_pol_hold(pol);
}
static struct xfrm_policy *__xfrm_policy_unlink(struct xfrm_policy *pol,
int dir)
{
netns xfrm: trivial netns propagations Take netns from xfrm_state or xfrm_policy. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:29:47 +08:00
struct net *net = xp_net(pol);
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
if (list_empty(&pol->walk.all))
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
return NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
/* Socket policies are not hashed. */
if (!hlist_unhashed(&pol->bydst)) {
xfrm: policy: use rcu versions for iteration and list add/del This is required once we allow lockless readers. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:52 +08:00
hlist_del_rcu(&pol->bydst);
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
hlist_del(&pol->byidx);
}
list_del_init(&pol->walk.all);
netns xfrm: trivial netns propagations Take netns from xfrm_state or xfrm_policy. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:29:47 +08:00
net->xfrm.policy_count[dir]--;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
return pol;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
static void xfrm_sk_policy_link(struct xfrm_policy *pol, int dir)
{
__xfrm_policy_link(pol, XFRM_POLICY_MAX + dir);
}
static void xfrm_sk_policy_unlink(struct xfrm_policy *pol, int dir)
{
__xfrm_policy_unlink(pol, XFRM_POLICY_MAX + dir);
}
[IPSEC] Kill spurious hard expire messages This patch ensures that the hard state/policy expire notifications are only sent when the state/policy is successfully removed from their respective tables. As it is, it's possible for a state/policy to both expire through reaching a hard limit, as well as being deleted by the user. Note that this behaviour isn't actually forbidden by RFC 2367. However, it is a quality of implementation issue. As an added bonus, the restructuring in this patch will help eventually in moving the expire notifications from softirq context into process context, thus improving their reliability. One important side-effect from this change is that SAs reaching their hard byte/packet limits are now deleted immediately, just like SAs that have reached their hard time limits. Previously they were announced immediately but only deleted after 30 seconds. This is bad because it prevents the system from issuing an ACQUIRE command until the existing state was deleted by the user or expires after the time is up. In the scenario where the expire notification was lost this introduces a 30 second delay into the system for no good reason. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2005-06-19 13:43:22 +08:00
int xfrm_policy_delete(struct xfrm_policy *pol, int dir)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: Namespacify xfrm state/policy locks By semantics, xfrm layer is fully name space aware, so will the locks, e.g. xfrm_state/pocliy_lock. Ensure exclusive access into state/policy link list for different name space with one global lock is not right in terms of semantics aspect at first place, as they are indeed mutually independent with each other, but also more seriously causes scalability problem. One practical scenario is on a Open Network Stack, more than hundreds of lxc tenants acts as routers within one host, a global xfrm_state/policy_lock becomes the bottleneck. But onces those locks are decoupled in a per-namespace fashion, locks contend is just with in specific name space scope, without causing additional SPD/SAD access delay for other name space. Also this patch improve scalability while as without changing original xfrm behavior. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:50 +08:00
struct net *net = xp_net(pol);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
pol = __xfrm_policy_unlink(pol, dir);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (pol) {
xfrm_policy_kill(pol);
[IPSEC] Kill spurious hard expire messages This patch ensures that the hard state/policy expire notifications are only sent when the state/policy is successfully removed from their respective tables. As it is, it's possible for a state/policy to both expire through reaching a hard limit, as well as being deleted by the user. Note that this behaviour isn't actually forbidden by RFC 2367. However, it is a quality of implementation issue. As an added bonus, the restructuring in this patch will help eventually in moving the expire notifications from softirq context into process context, thus improving their reliability. One important side-effect from this change is that SAs reaching their hard byte/packet limits are now deleted immediately, just like SAs that have reached their hard time limits. Previously they were announced immediately but only deleted after 30 seconds. This is bad because it prevents the system from issuing an ACQUIRE command until the existing state was deleted by the user or expires after the time is up. In the scenario where the expire notification was lost this introduces a 30 second delay into the system for no good reason. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2005-06-19 13:43:22 +08:00
return 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[IPSEC] Kill spurious hard expire messages This patch ensures that the hard state/policy expire notifications are only sent when the state/policy is successfully removed from their respective tables. As it is, it's possible for a state/policy to both expire through reaching a hard limit, as well as being deleted by the user. Note that this behaviour isn't actually forbidden by RFC 2367. However, it is a quality of implementation issue. As an added bonus, the restructuring in this patch will help eventually in moving the expire notifications from softirq context into process context, thus improving their reliability. One important side-effect from this change is that SAs reaching their hard byte/packet limits are now deleted immediately, just like SAs that have reached their hard time limits. Previously they were announced immediately but only deleted after 30 seconds. This is bad because it prevents the system from issuing an ACQUIRE command until the existing state was deleted by the user or expires after the time is up. In the scenario where the expire notification was lost this introduces a 30 second delay into the system for no good reason. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2005-06-19 13:43:22 +08:00
return -ENOENT;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Add some missing exports. To fix the case of modular xfrm_user. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-21 11:18:52 +08:00
EXPORT_SYMBOL(xfrm_policy_delete);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int xfrm_sk_policy_insert(struct sock *sk, int dir, struct xfrm_policy *pol)
{
net: xfrm: allow clearing socket xfrm policies. Currently it is possible to add or update socket policies, but not clear them. Therefore, once a socket policy has been applied, the socket cannot be used for unencrypted traffic. This patch allows (privileged) users to clear socket policies by passing in a NULL pointer and zero length argument to the {IP,IPV6}_{IPSEC,XFRM}_POLICY setsockopts. This results in both the incoming and outgoing policies being cleared. The simple approach taken in this patch cannot clear socket policies in only one direction. If desired this could be added in the future, for example by continuing to pass in a length of zero (which currently is guaranteed to return EMSGSIZE) and making the policy be a pointer to an integer that contains one of the XFRM_POLICY_{IN,OUT} enum values. An alternative would have been to interpret the length as a signed integer and use XFRM_POLICY_IN (i.e., 0) to clear the input policy and -XFRM_POLICY_OUT (i.e., -1) to clear the output policy. Tested: https://android-review.googlesource.com/539816 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-20 18:26:02 +08:00
struct net *net = sock_net(sk);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct xfrm_policy *old_pol;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
#ifdef CONFIG_XFRM_SUB_POLICY
if (pol && pol->type != XFRM_POLICY_TYPE_MAIN)
return -EINVAL;
#endif
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
old_pol = rcu_dereference_protected(sk->sk_policy[dir],
lockdep_is_held(&net->xfrm.xfrm_policy_lock));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (pol) {
[NET]: Convert xtime.tv_sec to get_seconds() Where appropriate, convert references to xtime.tv_sec to the get_seconds() helper function. Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-05 08:12:44 +08:00
pol->curlft.add_time = get_seconds();
xfrm: Try to honor policy index if it's supplied by user xfrm code always searches for unused policy index for newly created policy regardless whether or not user space policy index hint supplied. This patch enables such feature so that using "ip xfrm ... index=xxx" can be used by user to set specific policy index. Currently this beahvior is broken, so this patch make it happen as expected. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:48 +08:00
pol->index = xfrm_gen_index(net, XFRM_POLICY_MAX+dir, 0);
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
xfrm_sk_policy_link(pol, dir);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
rcu_assign_pointer(sk->sk_policy[dir], pol);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
if (old_pol) {
if (pol)
xfrm_policy_requeue(old_pol, pol);
xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-31 08:17:05 +08:00
/* Unlinking succeeds always. This is the only function
* allowed to delete or replace socket policy.
*/
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
xfrm_sk_policy_unlink(old_pol, dir);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
}
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (old_pol) {
xfrm_policy_kill(old_pol);
}
return 0;
}
xfrm: Const'ify policy arg to clone_policy. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:25:41 +08:00
static struct xfrm_policy *clone_policy(const struct xfrm_policy *old, int dir)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
netns xfrm: add struct xfrm_policy::xp_net Again, to avoid complications with passing netns when not necessary. Again, ->xp_net is set-once field, once set it never changes. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:21:45 +08:00
struct xfrm_policy *newp = xfrm_policy_alloc(xp_net(old), GFP_ATOMIC);
xfrm: Namespacify xfrm state/policy locks By semantics, xfrm layer is fully name space aware, so will the locks, e.g. xfrm_state/pocliy_lock. Ensure exclusive access into state/policy link list for different name space with one global lock is not right in terms of semantics aspect at first place, as they are indeed mutually independent with each other, but also more seriously causes scalability problem. One practical scenario is on a Open Network Stack, more than hundreds of lxc tenants acts as routers within one host, a global xfrm_state/policy_lock becomes the bottleneck. But onces those locks are decoupled in a per-namespace fashion, locks contend is just with in specific name space scope, without causing additional SPD/SAD access delay for other name space. Also this patch improve scalability while as without changing original xfrm behavior. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:50 +08:00
struct net *net = xp_net(old);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (newp) {
newp->selector = old->selector;
LSM: Make the Labeled IPsec hooks more stack friendly The xfrm_get_policy() and xfrm_add_pol_expire() put some rather large structs on the stack to work around the LSM API. This patch attempts to fix that problem by changing the LSM API to require only the relevant "security" pointers instead of the entire SPD entry; we do this for all of the security_xfrm_policy*() functions to keep things consistent. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-13 10:07:52 +08:00
if (security_xfrm_policy_clone(old->security,
&newp->security)) {
[LSM-IPSec]: Security association restriction. This patch series implements per packet access control via the extension of the Linux Security Modules (LSM) interface by hooks in the XFRM and pfkey subsystems that leverage IPSec security associations to label packets. Extensions to the SELinux LSM are included that leverage the patch for this purpose. This patch implements the changes necessary to the XFRM subsystem, pfkey interface, ipv4/ipv6, and xfrm_user interface to restrict a socket to use only authorized security associations (or no security association) to send/receive network packets. Patch purpose: The patch is designed to enable access control per packets based on the strongly authenticated IPSec security association. Such access controls augment the existing ones based on network interface and IP address. The former are very coarse-grained, and the latter can be spoofed. By using IPSec, the system can control access to remote hosts based on cryptographic keys generated using the IPSec mechanism. This enables access control on a per-machine basis or per-application if the remote machine is running the same mechanism and trusted to enforce the access control policy. Patch design approach: The overall approach is that policy (xfrm_policy) entries set by user-level programs (e.g., setkey for ipsec-tools) are extended with a security context that is used at policy selection time in the XFRM subsystem to restrict the sockets that can send/receive packets via security associations (xfrm_states) that are built from those policies. A presentation available at www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf from the SELinux symposium describes the overall approach. Patch implementation details: On output, the policy retrieved (via xfrm_policy_lookup or xfrm_sk_policy_lookup) must be authorized for the security context of the socket and the same security context is required for resultant security association (retrieved or negotiated via racoon in ipsec-tools). This is enforced in xfrm_state_find. On input, the policy retrieved must also be authorized for the socket (at __xfrm_policy_check), and the security context of the policy must also match the security association being used. The patch has virtually no impact on packets that do not use IPSec. The existing Netfilter (outgoing) and LSM rcv_skb hooks are used as before. Also, if IPSec is used without security contexts, the impact is minimal. The LSM must allow such policies to be selected for the combination of socket and remote machine, but subsequent IPSec processing proceeds as in the original case. Testing: The pfkey interface is tested using the ipsec-tools. ipsec-tools have been modified (a separate ipsec-tools patch is available for version 0.5) that supports assignment of xfrm_policy entries and security associations with security contexts via setkey and the negotiation using the security contexts via racoon. The xfrm_user interface is tested via ad hoc programs that set security contexts. These programs are also available from me, and contain programs for setting, getting, and deleting policy for testing this interface. Testing of sa functions was done by tracing kernel behavior. Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-14 15:12:27 +08:00
kfree(newp);
return NULL; /* ENOMEM */
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
newp->lft = old->lft;
newp->curlft = old->curlft;
xfrm: clone mark when cloning policy When we clone the SP, we should also clone the mark. Useful for socket based SPs. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-24 07:09:53 +08:00
newp->mark = old->mark;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
newp->action = old->action;
newp->flags = old->flags;
newp->xfrm_nr = old->xfrm_nr;
newp->index = old->index;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
newp->type = old->type;
xfrm: Copy policy family in clone_policy The syzbot found an ancient bug in the IPsec code. When we cloned a socket policy (for example, for a child TCP socket derived from a listening socket), we did not copy the family field. This results in a live policy with a zero family field. This triggers a BUG_ON check in the af_key code when the cloned policy is retrieved. This patch fixes it by copying the family field over. Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-10 11:14:06 +08:00
newp->family = old->family;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
memcpy(newp->xfrm_vec, old->xfrm_vec,
newp->xfrm_nr*sizeof(struct xfrm_tmpl));
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
xfrm_sk_policy_link(newp, dir);
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_pol_put(newp);
}
return newp;
}
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
int __xfrm_sk_clone_policy(struct sock *sk, const struct sock *osk)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: add rcu protection to sk->sk_policy[] XFRM can deal with SYNACK messages, sent while listener socket is not locked. We add proper rcu protection to __xfrm_sk_clone_policy() and xfrm_sk_policy_lookup() This might serve as the first step to remove xfrm.xfrm_policy_lock use in fast path. Fixes: fa76ce7328b2 ("inet: get rid of central tcp/dccp listener timer") Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 23:22:02 +08:00
const struct xfrm_policy *p;
struct xfrm_policy *np;
int i, ret = 0;
rcu_read_lock();
for (i = 0; i < 2; i++) {
p = rcu_dereference(osk->sk_policy[i]);
if (p) {
np = clone_policy(p, i);
if (unlikely(!np)) {
ret = -ENOMEM;
break;
}
rcu_assign_pointer(sk->sk_policy[i], np);
}
}
rcu_read_unlock();
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Fix wildcard as tunnel source Hashing SAs by source address breaks templates with wildcards as tunnel source since the source address used for hashing/lookup is still 0/0. Move source address lookup to xfrm_tmpl_resolve_one() so we can use the real address in the lookup. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-20 03:57:34 +08:00
static int
xfrm: Add oif to dst lookups Rules can be installed that direct route lookups to specific tables based on oif. Plumb the oif through the xfrm lookups so it gets set in the flow struct and passed to the resolver routines. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-08-11 06:58:11 +08:00
xfrm_get_saddr(struct net *net, int oif, xfrm_address_t *local,
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
xfrm_address_t *remote, unsigned short family, u32 mark)
[XFRM]: Fix wildcard as tunnel source Hashing SAs by source address breaks templates with wildcards as tunnel source since the source address used for hashing/lookup is still 0/0. Move source address lookup to xfrm_tmpl_resolve_one() so we can use the real address in the lookup. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-20 03:57:34 +08:00
{
int err;
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo = xfrm_policy_get_afinfo(family);
[XFRM]: Fix wildcard as tunnel source Hashing SAs by source address breaks templates with wildcards as tunnel source since the source address used for hashing/lookup is still 0/0. Move source address lookup to xfrm_tmpl_resolve_one() so we can use the real address in the lookup. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-20 03:57:34 +08:00
if (unlikely(afinfo == NULL))
return -EINVAL;
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
err = afinfo->get_saddr(net, oif, local, remote, mark);
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
[XFRM]: Fix wildcard as tunnel source Hashing SAs by source address breaks templates with wildcards as tunnel source since the source address used for hashing/lookup is still 0/0. Move source address lookup to xfrm_tmpl_resolve_one() so we can use the real address in the lookup. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-20 03:57:34 +08:00
return err;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* Resolve list of templates for the flow, given policy. */
static int
xfrm: Mark flowi arg to xfrm_tmpl_resolve{,_one}() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 10:35:39 +08:00
xfrm_tmpl_resolve_one(struct xfrm_policy *policy, const struct flowi *fl,
struct xfrm_state **xfrm, unsigned short family)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
netns xfrm: ->get_saddr in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:56:49 +08:00
struct net *net = xp_net(policy);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int nx;
int i, error;
Revert "xfrm: Fix stack-out-of-bounds read in xfrm_state_find." This reverts commit c9f3f813d462c72dbe412cee6a5cbacf13c4ad5e. This commit breaks transport mode when the policy template has widlcard addresses configured, so revert it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-15 13:40:57 +08:00
xfrm_address_t *daddr = xfrm_flowi_daddr(fl, family);
xfrm_address_t *saddr = xfrm_flowi_saddr(fl, family);
[XFRM]: Fix wildcard as tunnel source Hashing SAs by source address breaks templates with wildcards as tunnel source since the source address used for hashing/lookup is still 0/0. Move source address lookup to xfrm_tmpl_resolve_one() so we can use the real address in the lookup. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-20 03:57:34 +08:00
xfrm_address_t tmp;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: checkpatch errors with space This patch cleanup some space errors. Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:46 +08:00
for (nx = 0, i = 0; i < policy->xfrm_nr; i++) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct xfrm_state *x;
Revert "xfrm: Fix stack-out-of-bounds read in xfrm_state_find." This reverts commit c9f3f813d462c72dbe412cee6a5cbacf13c4ad5e. This commit breaks transport mode when the policy template has widlcard addresses configured, so revert it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-15 13:40:57 +08:00
xfrm_address_t *remote = daddr;
xfrm_address_t *local = saddr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct xfrm_tmpl *tmpl = &policy->xfrm_vec[i];
Revert "xfrm: Fix stack-out-of-bounds read in xfrm_state_find." This reverts commit c9f3f813d462c72dbe412cee6a5cbacf13c4ad5e. This commit breaks transport mode when the policy template has widlcard addresses configured, so revert it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-15 13:40:57 +08:00
if (tmpl->mode == XFRM_MODE_TUNNEL ||
tmpl->mode == XFRM_MODE_BEET) {
remote = &tmpl->id.daddr;
local = &tmpl->saddr;
if (xfrm_addr_any(local, tmpl->encap_family)) {
error = xfrm_get_saddr(net, fl->flowi_oif,
&tmp, remote,
tmpl->encap_family, 0);
if (error)
goto fail;
local = &tmp;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
x = xfrm_state_find(remote, local, fl, tmpl, policy, &error, family);
if (x && x->km.state == XFRM_STATE_VALID) {
xfrm[nx++] = x;
Revert "xfrm: Fix stack-out-of-bounds read in xfrm_state_find." This reverts commit c9f3f813d462c72dbe412cee6a5cbacf13c4ad5e. This commit breaks transport mode when the policy template has widlcard addresses configured, so revert it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-15 13:40:57 +08:00
daddr = remote;
saddr = local;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
continue;
}
if (x) {
error = (x->km.state == XFRM_STATE_ERROR ?
-EINVAL : -EAGAIN);
xfrm_state_put(x);
xfrm: fix checkpatch error Fix that "else should follow close brace '}'". Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:49 +08:00
} else if (error == -ESRCH) {
xfrm: do not leak ESRCH to user space I noticed that, under certain conditions, ESRCH can be leaked from the xfrm layer to user space through sys_connect. In particular, this seems to happen reliably when the kernel fails to resolve a template either because the AF_KEY receive buffer being used by racoon is full or because the SA entry we are trying to use is in XFRM_STATE_EXPIRED state. However, since this could be a transient issue it could be argued that EAGAIN would be more appropriate. Besides this error code is not even documented in the man page for sys_connect (as of man-pages 3.07). Signed-off-by: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-23 12:27:19 +08:00
error = -EAGAIN;
xfrm: fix checkpatch error Fix that "else should follow close brace '}'". Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:49 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (!tmpl->optional)
goto fail;
}
return nx;
fail:
xfrm: checkpatch errors with space This patch cleanup some space errors. Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:46 +08:00
for (nx--; nx >= 0; nx--)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_state_put(xfrm[nx]);
return error;
}
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
static int
xfrm: Mark flowi arg to xfrm_tmpl_resolve{,_one}() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 10:35:39 +08:00
xfrm_tmpl_resolve(struct xfrm_policy **pols, int npols, const struct flowi *fl,
struct xfrm_state **xfrm, unsigned short family)
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
{
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
struct xfrm_state *tp[XFRM_MAX_DEPTH];
struct xfrm_state **tpp = (npols > 1) ? tp : xfrm;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
int cnx = 0;
int error;
int ret;
int i;
for (i = 0; i < npols; i++) {
if (cnx + pols[i]->xfrm_nr >= XFRM_MAX_DEPTH) {
error = -ENOBUFS;
goto fail;
}
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
ret = xfrm_tmpl_resolve_one(pols[i], fl, &tpp[cnx], family);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
if (ret < 0) {
error = ret;
goto fail;
} else
cnx += ret;
}
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
/* found states are sorted for outbound processing */
if (npols > 1)
xfrm_state_sort(xfrm, tpp, cnx, family);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
return cnx;
fail:
xfrm: checkpatch errors with space This patch cleanup some space errors. Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:46 +08:00
for (cnx--; cnx >= 0; cnx--)
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
xfrm_state_put(tpp[cnx]);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
return error;
}
xfrm: policy: xfrm_get_tos cannot fail The comment makes it look like get_tos() is used to validate something, but it turns out the comment was about xfrm_find_bundle() which got removed years ago. xfrm_get_tos will return either the tos (ipv4) or 0 (ipv6). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:14 +08:00
static int xfrm_get_tos(const struct flowi *fl, int family)
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo;
xfrm: do not call rcu_read_unlock when afinfo is NULL in xfrm_get_tos When xfrm_policy_get_afinfo returns NULL, it will not hold rcu read lock. In this case, rcu_read_unlock should not be called in xfrm_get_tos, just like other places where it's calling xfrm_policy_get_afinfo. Fixes: f5e2bb4f5b22 ("xfrm: policy: xfrm_get_tos cannot fail") Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-02-17 15:16:22 +08:00
int tos;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: policy: xfrm_get_tos cannot fail The comment makes it look like get_tos() is used to validate something, but it turns out the comment was about xfrm_find_bundle() which got removed years ago. xfrm_get_tos will return either the tos (ipv4) or 0 (ipv6). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:14 +08:00
afinfo = xfrm_policy_get_afinfo(family);
xfrm: do not call rcu_read_unlock when afinfo is NULL in xfrm_get_tos When xfrm_policy_get_afinfo returns NULL, it will not hold rcu read lock. In this case, rcu_read_unlock should not be called in xfrm_get_tos, just like other places where it's calling xfrm_policy_get_afinfo. Fixes: f5e2bb4f5b22 ("xfrm: policy: xfrm_get_tos cannot fail") Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-02-17 15:16:22 +08:00
if (!afinfo)
return 0;
tos = afinfo->get_tos(fl);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
return tos;
}
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
static inline struct xfrm_dst *xfrm_alloc_dst(struct net *net, int family)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo = xfrm_policy_get_afinfo(family);
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
struct dst_ops *dst_ops;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
struct xfrm_dst *xdst;
if (!afinfo)
return ERR_PTR(-EINVAL);
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
switch (family) {
case AF_INET:
dst_ops = &net->xfrm.xfrm4_dst_ops;
break;
net: use IS_ENABLED(CONFIG_IPV6) Instead of testing defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-12-10 17:48:31 +08:00
#if IS_ENABLED(CONFIG_IPV6)
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
case AF_INET6:
dst_ops = &net->xfrm.xfrm6_dst_ops;
break;
#endif
default:
BUG();
}
net: remove DST_NOGC flag Now that all the components have been changed to release dst based on refcnt only and not depend on dst gc anymore, we can remove the temporary flag DST_NOGC. Note that we also need to remove the DST_NOCACHE check in dst_release() and dst_hold_safe() because now all the dst are released based on refcnt and behaves as DST_NOCACHE. Signed-off-by: Wei Wang <weiwan@google.com> Acked-by: Martin KaFai Lau <kafai@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-06-18 01:42:41 +08:00
xdst = dst_alloc(dst_ops, NULL, 1, DST_OBSOLETE_NONE, 0);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
net: check return value for dst_alloc return value of dst_alloc must be checked before use Signed-off-by: Madalin Bucur <madalin.bucur@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-09-26 15:04:36 +08:00
if (likely(xdst)) {
xfrm: Initialize the struct xfrm_dst behind the dst_enty field We start initializing the struct xfrm_dst at the first field behind the struct dst_enty. This is error prone because it might leave a new field uninitialized. So start initializing the struct xfrm_dst right behind the dst_entry. Suggested-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-07-06 07:39:34 +08:00
struct dst_entry *dst = &xdst->u.dst;
memset(dst + 1, 0, sizeof(*xdst) - sizeof(*dst));
net: check return value for dst_alloc return value of dst_alloc must be checked before use Signed-off-by: Madalin Bucur <madalin.bucur@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-09-26 15:04:36 +08:00
} else
xfrm: avoid possible oopse in xfrm_alloc_dst Commit 80c802f3073e84 (xfrm: cache bundles instead of policies for outgoing flows) introduced possible oopse when dst_alloc returns NULL. Signed-off-by: Hiroaki SHIMODA <shimoda.hiroaki@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-11 15:08:33 +08:00
xdst = ERR_PTR(-ENOBUFS);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
net: check return value for dst_alloc return value of dst_alloc must be checked before use Signed-off-by: Madalin Bucur <madalin.bucur@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-09-26 15:04:36 +08:00
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
return xdst;
}
[XFRM] IPv6: Fix dst/routing check at transformation. IPv6 specific thing is wrongly removed from transformation at net-2.6.25. This patch recovers it with current design. o Update "path" of xfrm_dst since IPv6 transformation should care about routing changes. It is required by MIPv6 and off-link destined IPsec. o Rename nfheader_len which is for non-fragment transformation used by MIPv6 to rt6i_nfheader_len as IPv6 name space. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:41:12 +08:00
static inline int xfrm_init_path(struct xfrm_dst *path, struct dst_entry *dst,
int nfheader_len)
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo =
[XFRM] IPv6: Fix dst/routing check at transformation. IPv6 specific thing is wrongly removed from transformation at net-2.6.25. This patch recovers it with current design. o Update "path" of xfrm_dst since IPv6 transformation should care about routing changes. It is required by MIPv6 and off-link destined IPsec. o Rename nfheader_len which is for non-fragment transformation used by MIPv6 to rt6i_nfheader_len as IPv6 name space. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:41:12 +08:00
xfrm_policy_get_afinfo(dst->ops->family);
int err;
if (!afinfo)
return -EINVAL;
err = afinfo->init_path(path, dst, nfheader_len);
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
[XFRM] IPv6: Fix dst/routing check at transformation. IPv6 specific thing is wrongly removed from transformation at net-2.6.25. This patch recovers it with current design. o Update "path" of xfrm_dst since IPv6 transformation should care about routing changes. It is required by MIPv6 and off-link destined IPsec. o Rename nfheader_len which is for non-fragment transformation used by MIPv6 to rt6i_nfheader_len as IPv6 name space. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:41:12 +08:00
return err;
}
ipsec: Fix bogus bundle flowi When I merged the bundle creation code, I introduced a bogus flowi value in the bundle. Instead of getting from the caller, it was instead set to the flow in the route object, which is totally different. The end result is that the bundles we created never match, and we instead end up with an ever growing bundle list. Thanks to Jamal for find this problem. Reported-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Acked-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-02 10:51:56 +08:00
static inline int xfrm_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
xfrm: Mark flowi arg to ->fill_dst() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 09:48:57 +08:00
const struct flowi *fl)
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo =
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
xfrm_policy_get_afinfo(xdst->u.dst.ops->family);
int err;
if (!afinfo)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -EINVAL;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
ipsec: Fix bogus bundle flowi When I merged the bundle creation code, I introduced a bogus flowi value in the bundle. Instead of getting from the caller, it was instead set to the flow in the route object, which is totally different. The end result is that the bundles we created never match, and we instead end up with an ever growing bundle list. Thanks to Jamal for find this problem. Reported-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Acked-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-03-02 10:51:56 +08:00
err = afinfo->fill_dst(xdst, dev, fl);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return err;
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
/* Allocate chain of dst_entry's, attach known xfrm's, calculate
* all the metrics... Shortly, bundle a bundle.
*/
static struct dst_entry *xfrm_bundle_create(struct xfrm_policy *policy,
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
struct xfrm_state **xfrm,
struct xfrm_dst **bundle,
int nx,
xfrm: Mark flowi arg to xfrm_bundle_create() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 10:36:50 +08:00
const struct flowi *fl,
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
struct dst_entry *dst)
{
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
struct net *net = xp_net(policy);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
unsigned long now = jiffies;
struct net_device *dev;
xfrm: Assign the inner mode output function to the dst entry As it is, we assign the outer modes output function to the dst entry when we create the xfrm bundle. This leads to two problems on interfamily scenarios. We might insert ipv4 packets into ip6_fragment when called from xfrm6_output. The system crashes if we try to fragment an ipv4 packet with ip6_fragment. This issue was introduced with git commit ad0081e4 (ipv6: Fragment locally generated tunnel-mode IPSec6 packets as needed). The second issue is, that we might insert ipv4 packets in netfilter6 and vice versa on interfamily scenarios. With this patch we assign the inner mode output function to the dst entry when we create the xfrm bundle. So xfrm4_output/xfrm6_output from the inner mode is used and the right fragmentation and netfilter functions are called. We switch then to outer mode with the output_finish functions. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-05-10 03:36:38 +08:00
struct xfrm_mode *inner_mode;
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
struct xfrm_dst *xdst_prev = NULL;
struct xfrm_dst *xdst0 = NULL;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
int i = 0;
int err;
int header_len = 0;
[XFRM] IPv6: Fix dst/routing check at transformation. IPv6 specific thing is wrongly removed from transformation at net-2.6.25. This patch recovers it with current design. o Update "path" of xfrm_dst since IPv6 transformation should care about routing changes. It is required by MIPv6 and off-link destined IPsec. o Rename nfheader_len which is for non-fragment transformation used by MIPv6 to rt6i_nfheader_len as IPv6 name space. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:41:12 +08:00
int nfheader_len = 0;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
int trailer_len = 0;
int tos;
int family = policy->selector.family;
[XFRM] MIP6: Fix address keys for routing search. Each MIPv6 XFRM state (DSTOPT/RH2) holds either destination or source address to be mangled in the IPv6 header (that is "CoA"). On Inter-MN communication after both nodes binds each other, they use route optimized traffic two MIPv6 states applied, and both source and destination address in the IPv6 header are replaced by the states respectively. The packet format is correct, however, next-hop routing search are not. This patch fixes it by remembering address pairs for later states. Based on patch from Masahide NAKAMURA <nakam@linux-ipv6.org>. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-02-22 13:48:22 +08:00
xfrm_address_t saddr, daddr;
xfrm_flowi_addr_get(fl, &saddr, &daddr, family);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
tos = xfrm_get_tos(fl, family);
dst_hold(dst);
for (; i < nx; i++) {
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
struct xfrm_dst *xdst = xfrm_alloc_dst(net, family);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
struct dst_entry *dst1 = &xdst->u.dst;
err = PTR_ERR(xdst);
if (IS_ERR(xdst)) {
dst_release(dst);
goto put_states;
}
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
bundle[i] = xdst;
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
if (!xdst_prev)
xdst0 = xdst;
ipsec: Fix dst leak in xfrm_bundle_create(). If we cannot find a suitable inner_mode value, we will leak the currently allocated 'xdst'. The fix is to make sure it is linked into the chain before erroring out. Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-10-11 11:59:38 +08:00
else
/* Ref count is taken during xfrm_alloc_dst()
* No need to do dst_clone() on dst1
*/
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
xfrm_dst_set_child(xdst_prev, &xdst->u.dst);
ipsec: Fix dst leak in xfrm_bundle_create(). If we cannot find a suitable inner_mode value, we will leak the currently allocated 'xdst'. The fix is to make sure it is linked into the chain before erroring out. Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-10-11 11:59:38 +08:00
xfrm: Assign the inner mode output function to the dst entry As it is, we assign the outer modes output function to the dst entry when we create the xfrm bundle. This leads to two problems on interfamily scenarios. We might insert ipv4 packets into ip6_fragment when called from xfrm6_output. The system crashes if we try to fragment an ipv4 packet with ip6_fragment. This issue was introduced with git commit ad0081e4 (ipv6: Fragment locally generated tunnel-mode IPSec6 packets as needed). The second issue is, that we might insert ipv4 packets in netfilter6 and vice versa on interfamily scenarios. With this patch we assign the inner mode output function to the dst entry when we create the xfrm bundle. So xfrm4_output/xfrm6_output from the inner mode is used and the right fragmentation and netfilter functions are called. We switch then to outer mode with the output_finish functions. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-05-10 03:36:38 +08:00
if (xfrm[i]->sel.family == AF_UNSPEC) {
inner_mode = xfrm_ip2inner_mode(xfrm[i],
xfrm_af2proto(family));
if (!inner_mode) {
err = -EAFNOSUPPORT;
dst_release(dst);
goto put_states;
}
} else
inner_mode = xfrm[i]->inner_mode;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
xdst->route = dst;
net: Abstract away all dst_entry metrics accesses. Use helper functions to hide all direct accesses, especially writes, to dst_entry metrics values. This will allow us to: 1) More easily change how the metrics are stored. 2) Implement COW for metrics. In particular this will help us put metrics into the inetpeer cache if that is what we end up doing. We can make the _metrics member a pointer instead of an array, initially have it point at the read-only metrics in the FIB, and then on the first set grab an inetpeer entry and point the _metrics member there. Signed-off-by: David S. Miller <davem@davemloft.net> Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
2010-12-09 13:16:57 +08:00
dst_copy_metrics(dst1, dst);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
if (xfrm[i]->props.mode != XFRM_MODE_TRANSPORT) {
family = xfrm[i]->props.family;
xfrm: Add oif to dst lookups Rules can be installed that direct route lookups to specific tables based on oif. Plumb the oif through the xfrm lookups so it gets set in the flow struct and passed to the resolver routines. Signed-off-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-08-11 06:58:11 +08:00
dst = xfrm_dst_lookup(xfrm[i], tos, fl->flowi_oif,
net: xfrm: support setting an output mark. On systems that use mark-based routing it may be necessary for routing lookups to use marks in order for packets to be routed correctly. An example of such a system is Android, which uses socket marks to route packets via different networks. Currently, routing lookups in tunnel mode always use a mark of zero, making routing incorrect on such systems. This patch adds a new output_mark element to the xfrm state and a corresponding XFRMA_OUTPUT_MARK netlink attribute. The output mark differs from the existing xfrm mark in two ways: 1. The xfrm mark is used to match xfrm policies and states, while the xfrm output mark is used to set the mark (and influence the routing) of the packets emitted by those states. 2. The existing mark is constrained to be a subset of the bits of the originating socket or transformed packet, but the output mark is arbitrary and depends only on the state. The use of a separate mark provides additional flexibility. For example: - A packet subject to two transforms (e.g., transport mode inside tunnel mode) can have two different output marks applied to it, one for the transport mode SA and one for the tunnel mode SA. - On a system where socket marks determine routing, the packets emitted by an IPsec tunnel can be routed based on a mark that is determined by the tunnel, not by the marks of the unencrypted packets. - Support for setting the output marks can be introduced without breaking any existing setups that employ both mark-based routing and xfrm tunnel mode. Simply changing the code to use the xfrm mark for routing output packets could xfrm mark could change behaviour in a way that breaks these setups. If the output mark is unspecified or set to zero, the mark is not set or changed. Tested: make allyesconfig; make -j64 Tested: https://android-review.googlesource.com/452776 Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-11 01:11:33 +08:00
&saddr, &daddr, family,
xfrm[i]->props.output_mark);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
err = PTR_ERR(dst);
if (IS_ERR(dst))
goto put_states;
} else
dst_hold(dst);
dst1->xfrm = xfrm[i];
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xdst->xfrm_genid = xfrm[i]->genid;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
net: Document dst->obsolete better. Add a big comment explaining how the field works, and use defines instead of magic constants for the values assigned to it. Suggested by Joe Perches. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-07-20 03:31:33 +08:00
dst1->obsolete = DST_OBSOLETE_FORCE_CHK;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
dst1->flags |= DST_HOST;
dst1->lastuse = now;
dst1->input = dst_discard;
xfrm: Assign the inner mode output function to the dst entry As it is, we assign the outer modes output function to the dst entry when we create the xfrm bundle. This leads to two problems on interfamily scenarios. We might insert ipv4 packets into ip6_fragment when called from xfrm6_output. The system crashes if we try to fragment an ipv4 packet with ip6_fragment. This issue was introduced with git commit ad0081e4 (ipv6: Fragment locally generated tunnel-mode IPSec6 packets as needed). The second issue is, that we might insert ipv4 packets in netfilter6 and vice versa on interfamily scenarios. With this patch we assign the inner mode output function to the dst entry when we create the xfrm bundle. So xfrm4_output/xfrm6_output from the inner mode is used and the right fragmentation and netfilter functions are called. We switch then to outer mode with the output_finish functions. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-05-10 03:36:38 +08:00
dst1->output = inner_mode->afinfo->output;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
xdst_prev = xdst;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
header_len += xfrm[i]->props.header_len;
[XFRM] IPv6: Fix dst/routing check at transformation. IPv6 specific thing is wrongly removed from transformation at net-2.6.25. This patch recovers it with current design. o Update "path" of xfrm_dst since IPv6 transformation should care about routing changes. It is required by MIPv6 and off-link destined IPsec. o Rename nfheader_len which is for non-fragment transformation used by MIPv6 to rt6i_nfheader_len as IPv6 name space. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:41:12 +08:00
if (xfrm[i]->type->flags & XFRM_TYPE_NON_FRAGMENT)
nfheader_len += xfrm[i]->props.header_len;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
trailer_len += xfrm[i]->props.trailer_len;
}
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
xfrm_dst_set_child(xdst_prev, dst);
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
xdst0->path = dst;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
err = -ENODEV;
dev = dst->dev;
if (!dev)
goto free_dst;
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
xfrm_init_path(xdst0, dst, nfheader_len);
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
xfrm_init_pmtu(bundle, nx);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
for (xdst_prev = xdst0; xdst_prev != (struct xfrm_dst *)dst;
xdst_prev = (struct xfrm_dst *) xfrm_dst_child(&xdst_prev->u.dst)) {
err = xfrm_fill_dst(xdst_prev, dev, fl);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
if (err)
goto free_dst;
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
xdst_prev->u.dst.header_len = header_len;
xdst_prev->u.dst.trailer_len = trailer_len;
header_len -= xdst_prev->u.dst.xfrm->props.header_len;
trailer_len -= xdst_prev->u.dst.xfrm->props.trailer_len;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
}
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
return &xdst0->u.dst;
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
put_states:
for (; i < nx; i++)
xfrm_state_put(xfrm[i]);
free_dst:
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
if (xdst0)
dst_release_immediate(&xdst0->u.dst);
xfrm Fix potential error pointer dereference in xfrm_bundle_create. We may derference an invalid pointer in the error path of xfrm_bundle_create(). Fix this by returning this error pointer directly instead of assigning it to xdst0. Fixes: 45b018beddb6 ("ipsec: Create and use new helpers for dst child access.") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-05-31 15:45:18 +08:00
return ERR_PTR(err);
[IPSEC]: Merge common code into xfrm_bundle_create Half of the code in xfrm4_bundle_create and xfrm6_bundle_create are common. This patch extracts that logic and puts it into xfrm_bundle_create. The rest of it are then accessed through afinfo. As a result this fixes the problem with inter-family transforms where we treat every xfrm dst in the bundle as if it belongs to the top family. This patch also fixes a long-standing error-path bug where we may free the xfrm states twice. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-12 01:32:34 +08:00
}
xfrm: Mark flowi arg to xfrm_expand_policies() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 10:33:42 +08:00
static int xfrm_expand_policies(const struct flowi *fl, u16 family,
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
struct xfrm_policy **pols,
int *num_pols, int *num_xfrms)
{
int i;
if (*num_pols == 0 || !pols[0]) {
*num_pols = 0;
*num_xfrms = 0;
return 0;
}
if (IS_ERR(pols[0]))
return PTR_ERR(pols[0]);
*num_xfrms = pols[0]->xfrm_nr;
#ifdef CONFIG_XFRM_SUB_POLICY
if (pols[0] && pols[0]->action == XFRM_POLICY_ALLOW &&
pols[0]->type != XFRM_POLICY_TYPE_MAIN) {
pols[1] = xfrm_policy_lookup_bytype(xp_net(pols[0]),
XFRM_POLICY_TYPE_MAIN,
fl, family,
XFRM_POLICY_OUT);
if (pols[1]) {
if (IS_ERR(pols[1])) {
xfrm_pols_put(pols, *num_pols);
return PTR_ERR(pols[1]);
}
xfrm: checkpatch erros with space prohibited Fix checkpatch error "space prohibited xxx". Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:48 +08:00
(*num_pols)++;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
(*num_xfrms) += pols[1]->xfrm_nr;
}
}
#endif
for (i = 0; i < *num_pols; i++) {
if (pols[i]->action != XFRM_POLICY_ALLOW) {
*num_xfrms = -1;
break;
}
}
return 0;
}
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
static void xfrm_last_dst_update(struct xfrm_dst *xdst, struct xfrm_dst *old)
{
this_cpu_write(xfrm_last_dst, xdst);
if (old)
dst_release(&old->u.dst);
}
static void __xfrm_pcpu_work_fn(void)
{
struct xfrm_dst *old;
old = this_cpu_read(xfrm_last_dst);
if (old && !xfrm_bundle_ok(old))
xfrm_last_dst_update(NULL, old);
}
static void xfrm_pcpu_work_fn(struct work_struct *work)
{
local_bh_disable();
rcu_read_lock();
__xfrm_pcpu_work_fn();
rcu_read_unlock();
local_bh_enable();
}
void xfrm_policy_cache_flush(void)
{
struct xfrm_dst *old;
xfrm_policy: use true and false for boolean values Assign true or false to boolean variables instead of an integer value. This issue was detected with the help of Coccinelle. Signed-off-by: Gustavo A. R. Silva <garsilva@embeddedor.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-03-06 05:49:59 +08:00
bool found = false;
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
int cpu;
xfrm: don't call xfrm_policy_cache_flush while holding spinlock xfrm_policy_cache_flush can sleep, so it cannot be called while holding a spinlock. We could release the lock first, but I don't see why we need to invoke this function here in first place, the packet path won't reuse an xdst entry unless its still valid. While at it, add an annotation to xfrm_policy_cache_flush, it would have probably caught this bug sooner. Fixes: ec30d78c14a813 ("xfrm: add xdst pcpu cache") Reported-by: syzbot+e149f7d1328c26f9c12f@syzkaller.appspotmail.com Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-01-06 08:13:08 +08:00
might_sleep();
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
local_bh_disable();
rcu_read_lock();
for_each_possible_cpu(cpu) {
old = per_cpu(xfrm_last_dst, cpu);
if (old && !xfrm_bundle_ok(old)) {
if (smp_processor_id() == cpu) {
__xfrm_pcpu_work_fn();
continue;
}
found = true;
break;
}
}
rcu_read_unlock();
local_bh_enable();
if (!found)
return;
get_online_cpus();
for_each_possible_cpu(cpu) {
bool bundle_release;
rcu_read_lock();
old = per_cpu(xfrm_last_dst, cpu);
bundle_release = old && !xfrm_bundle_ok(old);
rcu_read_unlock();
if (!bundle_release)
continue;
if (cpu_online(cpu)) {
schedule_work_on(cpu, &xfrm_pcpu_work[cpu]);
continue;
}
rcu_read_lock();
old = per_cpu(xfrm_last_dst, cpu);
if (old && !xfrm_bundle_ok(old)) {
per_cpu(xfrm_last_dst, cpu) = NULL;
dst_release(&old->u.dst);
}
rcu_read_unlock();
}
put_online_cpus();
}
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
static bool xfrm_xdst_can_reuse(struct xfrm_dst *xdst,
struct xfrm_state * const xfrm[],
int num)
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
{
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
const struct dst_entry *dst = &xdst->u.dst;
int i;
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
if (xdst->num_xfrms != num)
return false;
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
for (i = 0; i < num; i++) {
if (!dst || dst->xfrm != xfrm[i])
return false;
net: Create and use new helper xfrm_dst_child(). Only IPSEC routes have a non-NULL dst->child pointer. And IPSEC routes are identified by a non-NULL dst->xfrm pointer. Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-29 04:40:22 +08:00
dst = xfrm_dst_child(dst);
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
}
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
return xfrm_bundle_ok(xdst);
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
static struct xfrm_dst *
xfrm_resolve_and_create_bundle(struct xfrm_policy **pols, int num_pols,
xfrm: Mark flowi arg to xfrm_resolve_and_create_bundle() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 10:38:51 +08:00
const struct flowi *fl, u16 family,
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
struct dst_entry *dst_orig)
{
struct net *net = xp_net(pols[0]);
struct xfrm_state *xfrm[XFRM_MAX_DEPTH];
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
struct xfrm_dst *bundle[XFRM_MAX_DEPTH];
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
struct xfrm_dst *xdst, *old;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
struct dst_entry *dst;
int err;
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
/* Try to instantiate a bundle */
err = xfrm_tmpl_resolve(pols, num_pols, fl, xfrm, family);
if (err <= 0) {
if (err != 0 && err != -EAGAIN)
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTPOLERROR);
return ERR_PTR(err);
}
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
xdst = this_cpu_read(xfrm_last_dst);
if (xdst &&
xdst->u.dst.dev == dst_orig->dev &&
xdst->num_pols == num_pols &&
memcmp(xdst->pols, pols,
xfrm: check that cached bundle is still valid Quoting Ilan Tayari: 1. Set up a host-to-host IPSec tunnel (or transport, doesn't matter) 2. Ping over IPSec, or do something to populate the pcpu cache 3. Join a MC group, then leave MC group 4. Try to ping again using same CPU as before -> traffic doesn't egress the machine at all Ilan debugged the problem down to the fact that one of the path dsts devices point to lo due to earlier dst_dev_put(). In this case, dst is marked as DEAD and we cannot reuse the bundle. The cache only asserted that the requested policy and that of the cached bundle match, but its not enough - also verify the path is still valid. Fixes: ec30d78c14a813 ("xfrm: add xdst pcpu cache") Reported-by: Ayham Masood <ayhamm@mellanox.com> Tested-by: Ilan Tayari <ilant@mellanox.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-08-06 16:19:07 +08:00
sizeof(struct xfrm_policy *) * num_pols) == 0 &&
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
xfrm_xdst_can_reuse(xdst, xfrm, err)) {
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
dst_hold(&xdst->u.dst);
xfrm: put policies when reusing pcpu xdst entry We need to put the policies when re-using the pcpu xdst entry, else this leaks the reference. Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-12-12 01:23:09 +08:00
xfrm_pols_put(pols, num_pols);
xfrm: do unconditional template resolution before pcpu cache check Stephen Smalley says: Since 4.14-rc1, the selinux-testsuite has been encountering sporadic failures during testing of labeled IPSEC. git bisect pointed to commit ec30d ("xfrm: add xdst pcpu cache"). The xdst pcpu cache is only checking that the policies are the same, but does not validate that the policy, state, and flow match with respect to security context labeling. As a result, the wrong SA could be used and the receiver could end up performing permission checking and providing SO_PEERSEC or SCM_SECURITY values for the wrong security context. This fix makes it so that we always do the template resolution, and then checks that the found states match those in the pcpu bundle. This has the disadvantage of doing a bit more work (lookup in state hash table) if we can reuse the xdst entry (we only avoid xdst alloc/free) but we don't add a lot of extra work in case we can't reuse. xfrm_pol_dead() check is removed, reasoning is that xfrm_tmpl_resolve does all needed checks. Cc: Paul Moore <paul@paul-moore.com> Fixes: ec30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Stephen Smalley <sds@tycho.nsa.gov> Tested-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Florian Westphal <fw@strlen.de> Acked-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-11-02 23:46:01 +08:00
while (err > 0)
xfrm_state_put(xfrm[--err]);
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
return xdst;
}
old = xdst;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
dst = xfrm_bundle_create(pols[0], xfrm, bundle, err, fl, dst_orig);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (IS_ERR(dst)) {
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTBUNDLEGENERROR);
return ERR_CAST(dst);
}
xdst = (struct xfrm_dst *)dst;
xdst->num_xfrms = err;
xdst->num_pols = num_pols;
xfrm: checkpatch errors with foo * bar This patch clean up some checkpatch errors like this: ERROR: "foo * bar" should be "foo *bar" ERROR: "(foo*)" should be "(foo *)" Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:47 +08:00
memcpy(xdst->pols, pols, sizeof(struct xfrm_policy *) * num_pols);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xdst->policy_genid = atomic_read(&pols[0]->genid);
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
atomic_set(&xdst->u.dst.__refcnt, 2);
xfrm_last_dst_update(xdst, old);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
return xdst;
}
xfrm: Convert timers to use timer_setup() In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() helper to pass the timer pointer explicitly. Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: "David S. Miller" <davem@davemloft.net> Cc: netdev@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-17 08:28:56 +08:00
static void xfrm_policy_queue_process(struct timer_list *t)
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
{
struct sk_buff *skb;
struct sock *sk;
struct dst_entry *dst;
xfrm: Convert timers to use timer_setup() In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() helper to pass the timer pointer explicitly. Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: "David S. Miller" <davem@davemloft.net> Cc: netdev@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-17 08:28:56 +08:00
struct xfrm_policy *pol = from_timer(pol, t, polq.hold_timer);
xfrm: Only compute net once in xfrm_policy_queue_process Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-10-08 05:48:34 +08:00
struct net *net = xp_net(pol);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
struct xfrm_policy_queue *pq = &pol->polq;
struct flowi fl;
struct sk_buff_head list;
spin_lock(&pq->hold_queue.lock);
skb = skb_peek(&pq->hold_queue);
xfrm: check for a vaild skb in xfrm_policy_queue_process We might dreference a NULL pointer if the hold_queue is empty, so add a check to avoid this. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:51 +08:00
if (!skb) {
spin_unlock(&pq->hold_queue.lock);
goto out;
}
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
dst = skb_dst(skb);
sk = skb->sk;
xfrm_decode_session(skb, &fl, dst->ops->family);
spin_unlock(&pq->hold_queue.lock);
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
dst_hold(xfrm_dst_path(dst));
xfrm: Fix policy hold queue after flowcache removal. Now that the flowcache is removed we need to generate a new dummy bundle every time we check if the needed SAs are in place because the dummy bundle is not cached anymore. Fix it by passing the XFRM_LOOKUP_QUEUE flag to xfrm_lookup(). This makes sure that we get a dummy bundle in case the SAs are not yet in place. Fixes: 3ca28286ea80 ("xfrm_policy: bypass flow_cache_lookup") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-02-01 18:26:12 +08:00
dst = xfrm_lookup(net, xfrm_dst_path(dst), &fl, sk, XFRM_LOOKUP_QUEUE);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
if (IS_ERR(dst))
goto purge_queue;
if (dst->flags & DST_XFRM_QUEUE) {
dst_release(dst);
if (pq->timeout >= XFRM_QUEUE_TMO_MAX)
goto purge_queue;
pq->timeout = pq->timeout << 1;
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
if (!mod_timer(&pq->hold_timer, jiffies + pq->timeout))
xfrm_pol_hold(pol);
goto out;
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
}
dst_release(dst);
__skb_queue_head_init(&list);
spin_lock(&pq->hold_queue.lock);
pq->timeout = 0;
skb_queue_splice_init(&pq->hold_queue, &list);
spin_unlock(&pq->hold_queue.lock);
while (!skb_queue_empty(&list)) {
skb = __skb_dequeue(&list);
xfrm_decode_session(skb, &fl, skb_dst(skb)->ops->family);
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
dst_hold(xfrm_dst_path(skb_dst(skb)));
dst = xfrm_lookup(net, xfrm_dst_path(skb_dst(skb)), &fl, skb->sk, 0);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
if (IS_ERR(dst)) {
kfree_skb(skb);
continue;
}
nf_reset(skb);
skb_dst_drop(skb);
skb_dst_set(skb, dst);
net: Pass net into dst_output and remove dst_output_okfn Replace dst_output_okfn with dst_output Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-10-08 05:48:35 +08:00
dst_output(net, skb->sk, skb);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
}
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
out:
xfrm_pol_put(pol);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
return;
purge_queue:
pq->timeout = 0;
xfrm: remove the xfrm_queue_purge definition The task of xfrm_queue_purge is same as skb_queue_purge, so remove it Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-22 15:51:16 +08:00
skb_queue_purge(&pq->hold_queue);
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
xfrm_pol_put(pol);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
}
dst: Pass net into dst->output The network namespace is already passed into dst_output pass it into dst->output lwt->output and friends. Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-10-08 05:48:47 +08:00
static int xdst_queue_output(struct net *net, struct sock *sk, struct sk_buff *skb)
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
{
unsigned long sched_next;
struct dst_entry *dst = skb_dst(skb);
struct xfrm_dst *xdst = (struct xfrm_dst *) dst;
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
struct xfrm_policy *pol = xdst->pols[0];
struct xfrm_policy_queue *pq = &pol->polq;
xfrm: Don't queue retransmitted packets if the original is still on the host It does not make sense to queue retransmitted packets if the original packet is still in some queue of this host. So add a check to xdst_queue_output() and drop the packet if the original packet is not yet sent. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Acked-by: Eric Dumazet <edumazet@google.com>
2013-10-16 19:42:46 +08:00
net: skb_fclone_busy() needs to detect orphaned skb Some drivers are unable to perform TX completions in a bound time. They instead call skb_orphan() Problem is skb_fclone_busy() has to detect this case, otherwise we block TCP retransmits and can freeze unlucky tcp sessions on mostly idle hosts. Signed-off-by: Eric Dumazet <edumazet@google.com> Fixes: 1f3279ae0c13 ("tcp: avoid retransmits of TCP packets hanging in host queues") Signed-off-by: David S. Miller <davem@davemloft.net>
2014-10-31 01:32:34 +08:00
if (unlikely(skb_fclone_busy(sk, skb))) {
xfrm: Don't queue retransmitted packets if the original is still on the host It does not make sense to queue retransmitted packets if the original packet is still in some queue of this host. So add a check to xdst_queue_output() and drop the packet if the original packet is not yet sent. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Acked-by: Eric Dumazet <edumazet@google.com>
2013-10-16 19:42:46 +08:00
kfree_skb(skb);
return 0;
}
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
if (pq->hold_queue.qlen > XFRM_MAX_QUEUE_LEN) {
kfree_skb(skb);
return -EAGAIN;
}
skb_dst_force(skb);
spin_lock_bh(&pq->hold_queue.lock);
if (!pq->timeout)
pq->timeout = XFRM_QUEUE_TMO_MIN;
sched_next = jiffies + pq->timeout;
if (del_timer(&pq->hold_timer)) {
if (time_before(pq->hold_timer.expires, sched_next))
sched_next = pq->hold_timer.expires;
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
xfrm_pol_put(pol);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
}
__skb_queue_tail(&pq->hold_queue, skb);
xfrm: Add refcount handling to queued policies We need to ensure that policies can't go away as long as the hold timer is armed, so take a refcont when we arm the timer and drop one if we delete it. Bug was introduced with git commit a0073fe18 ("xfrm: Add a state resolution packet queue") Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-10-08 16:49:45 +08:00
if (!mod_timer(&pq->hold_timer, sched_next))
xfrm_pol_hold(pol);
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
spin_unlock_bh(&pq->hold_queue.lock);
return 0;
}
static struct xfrm_dst *xfrm_create_dummy_bundle(struct net *net,
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
struct xfrm_flo *xflo,
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
const struct flowi *fl,
int num_xfrms,
u16 family)
{
int err;
struct net_device *dev;
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
struct dst_entry *dst;
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
struct dst_entry *dst1;
struct xfrm_dst *xdst;
xdst = xfrm_alloc_dst(net, family);
if (IS_ERR(xdst))
return xdst;
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
if (!(xflo->flags & XFRM_LOOKUP_QUEUE) ||
net->xfrm.sysctl_larval_drop ||
num_xfrms <= 0)
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
return xdst;
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
dst = xflo->dst_orig;
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
dst1 = &xdst->u.dst;
dst_hold(dst);
xdst->route = dst;
dst_copy_metrics(dst1, dst);
dst1->obsolete = DST_OBSOLETE_FORCE_CHK;
dst1->flags |= DST_HOST | DST_XFRM_QUEUE;
dst1->lastuse = jiffies;
dst1->input = dst_discard;
dst1->output = xdst_queue_output;
dst_hold(dst);
ipsec: Create and use new helpers for dst child access. This will make a future change moving the dst->child pointer less invasive. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:28 +08:00
xfrm_dst_set_child(xdst, dst);
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
xdst->path = dst;
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
xfrm_init_path((struct xfrm_dst *)dst1, dst, 0);
err = -ENODEV;
dev = dst->dev;
if (!dev)
goto free_dst;
err = xfrm_fill_dst(xdst, dev, fl);
if (err)
goto free_dst;
out:
return xdst;
free_dst:
dst_release(dst1);
xdst = ERR_PTR(err);
goto out;
}
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
static struct xfrm_dst *
xfrm_policy: bypass flow_cache_lookup Instead of consulting flow cache, call the xfrm bundle/policy lookup functions directly. This pretends the flow cache had no entry. This helps to gradually remove flow cache integration, followup commit will remove the dead code that this change adds. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:21 +08:00
xfrm_bundle_lookup(struct net *net, const struct flowi *fl, u16 family, u8 dir, struct xfrm_flo *xflo)
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
{
struct xfrm_policy *pols[XFRM_POLICY_TYPE_MAX];
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
int num_pols = 0, num_xfrms = 0, err;
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
struct xfrm_dst *xdst;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
/* Resolve policies to use if we couldn't get them from
* previous cache entry */
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
num_pols = 1;
xfrm_policy: remove xfrm_policy_lookup This removes the wrapper and renames the __xfrm_policy_lookup variant to get rid of another place that used flow cache objects. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:24 +08:00
pols[0] = xfrm_policy_lookup(net, fl, family, dir);
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
err = xfrm_expand_policies(fl, family, pols,
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
&num_pols, &num_xfrms);
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
if (err < 0)
goto inc_error;
if (num_pols == 0)
return NULL;
if (num_xfrms <= 0)
goto make_dummy_bundle;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xfrm: Fix a race in the xdst pcpu cache. We need to run xfrm_resolve_and_create_bundle() with bottom halves off. Otherwise we may reuse an already released dst_enty when the xfrm lookup functions are called from process context. Fixes: c30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Darius Ski <darius.ski@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-01-10 19:14:28 +08:00
local_bh_disable();
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
xdst = xfrm_resolve_and_create_bundle(pols, num_pols, fl, family,
xfrm: Fix a race in the xdst pcpu cache. We need to run xfrm_resolve_and_create_bundle() with bottom halves off. Otherwise we may reuse an already released dst_enty when the xfrm lookup functions are called from process context. Fixes: c30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Darius Ski <darius.ski@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-01-10 19:14:28 +08:00
xflo->dst_orig);
local_bh_enable();
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
if (IS_ERR(xdst)) {
err = PTR_ERR(xdst);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (err != -EAGAIN)
goto error;
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
goto make_dummy_bundle;
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
} else if (xdst == NULL) {
xfrm: do not assume that template resolving always returns xfrms xfrm_resolve_and_create_bundle() assumed that, if policies indicated presence of xfrms, bundle template resolution would always return some xfrms. This is not true for 'use' level policies which can result in no xfrm's being applied if there is no suitable xfrm states. This fixes a crash by this incorrect assumption. Reported-by: George Spelvin <linux@horizon.com> Bisected-by: George Spelvin <linux@horizon.com> Tested-by: George Spelvin <linux@horizon.com> Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-07-13 05:29:42 +08:00
num_xfrms = 0;
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
goto make_dummy_bundle;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
}
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
return xdst;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
make_dummy_bundle:
/* We found policies, but there's no bundles to instantiate:
* either because the policy blocks, has no transformations or
* we could not build template (no xfrm_states).*/
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
xdst = xfrm_create_dummy_bundle(net, xflo, fl, num_xfrms, family);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (IS_ERR(xdst)) {
xfrm_pols_put(pols, num_pols);
return ERR_CAST(xdst);
}
xdst->num_pols = num_pols;
xdst->num_xfrms = num_xfrms;
xfrm: checkpatch errors with foo * bar This patch clean up some checkpatch errors like this: ERROR: "foo * bar" should be "foo *bar" ERROR: "(foo*)" should be "(foo *)" Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:47 +08:00
memcpy(xdst->pols, pols, sizeof(struct xfrm_policy *) * num_pols);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
return xdst;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
inc_error:
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTPOLERROR);
error:
xfrm_policy: remove always true/false branches after previous change oldflo and xdst are always NULL. These branches were already removed by gcc, this doesn't change code. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:22 +08:00
xfrm_pols_put(pols, num_pols);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
return ERR_PTR(err);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Handle blackhole route creation via afinfo. That way we don't have to potentially do this in every xfrm_lookup() caller. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-02 06:59:04 +08:00
static struct dst_entry *make_blackhole(struct net *net, u16 family,
struct dst_entry *dst_orig)
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo = xfrm_policy_get_afinfo(family);
xfrm: Handle blackhole route creation via afinfo. That way we don't have to potentially do this in every xfrm_lookup() caller. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-02 06:59:04 +08:00
struct dst_entry *ret;
if (!afinfo) {
dst_release(dst_orig);
xfrm: fix a read lock imbalance in make_blackhole if xfrm_policy_get_afinfo returns 0, it has already released the read lock, xfrm_policy_put_afinfo should not be called again. Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-09-18 06:40:10 +08:00
return ERR_PTR(-EINVAL);
xfrm: Handle blackhole route creation via afinfo. That way we don't have to potentially do this in every xfrm_lookup() caller. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-02 06:59:04 +08:00
} else {
ret = afinfo->blackhole_route(net, dst_orig);
}
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
xfrm: Handle blackhole route creation via afinfo. That way we don't have to potentially do this in every xfrm_lookup() caller. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-02 06:59:04 +08:00
return ret;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* Main function: finds/creates a bundle for given flow.
*
* At the moment we eat a raw IP route. Mostly to speed up lookups
* on interfaces with disabled IPsec.
*/
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
struct dst_entry *xfrm_lookup(struct net *net, struct dst_entry *dst_orig,
const struct flowi *fl,
inet: constify ip_route_output_flow() socket argument Very soon, TCP stack might call inet_csk_route_req(), which calls inet_csk_route_req() with an unlocked listener socket, so we need to make sure ip_route_output_flow() is not trying to change any field from its socket argument. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-25 22:39:10 +08:00
const struct sock *sk, int flags)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
struct xfrm_policy *pols[XFRM_POLICY_TYPE_MAX];
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
struct xfrm_dst *xdst;
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
struct dst_entry *dst, *route;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
u16 family = dst_orig->ops->family;
xfrm_policy: kill flow to policy dir conversion XFRM_POLICY_IN/OUT/FWD are identical to FLOW_DIR_*, so gcc already removed this function as its just returns the argument. Again, no code change. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:23 +08:00
u8 dir = XFRM_POLICY_OUT;
xfrm: potential uninitialized variable num_xfrms potential uninitialized variable num_xfrms fix compiler warning: 'num_xfrms' may be used uninitialized in this function. Signed-off-by: Changli Gao <xiaosuo@gmail.com> ---- net/xfrm/xfrm_policy.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-28 05:20:22 +08:00
int i, err, num_pols, num_xfrms = 0, drop_pols = 0;
[MLSXFRM]: Flow based matching of xfrm policy and state This implements a seemless mechanism for xfrm policy selection and state matching based on the flow sid. This also includes the necessary SELinux enforcement pieces. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-07-25 14:29:07 +08:00
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
dst = NULL;
xdst = NULL;
route = NULL;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
xfrm: take care of request sockets TCP SYNACK messages might now be attached to request sockets. XFRM needs to get back to a listener socket. Adds new helpers that might be used elsewhere : sk_to_full_sk() and sk_const_to_full_sk() Note: We also need to add RCU protection for xfrm lookups, now TCP/DCCP have lockless listener processing. This will be addressed in separate patches. Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener") Reported-by: Dave Jones <davej@codemonkey.org.uk> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 00:53:17 +08:00
sk = sk_const_to_full_sk(sk);
[XFRM] policy: Replace magic number with XFRM_POLICY_OUT Signed-off-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-08-26 04:46:55 +08:00
if (sk && sk->sk_policy[XFRM_POLICY_OUT]) {
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
num_pols = 1;
xfrm: Don't use sk_family for socket policy lookups On IPv4-mapped IPv6 addresses sk_family is AF_INET6, but the flow informations are created based on AF_INET. So the routing set up 'struct flowi4' but we try to access 'struct flowi6' what leads to an out of bounds access. Fix this by using the family we get with the dst_entry, like we do it for the standard policy lookup. Reported-by: Dmitry Vyukov <dvyukov@google.com> Tested-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-14 14:43:56 +08:00
pols[0] = xfrm_sk_policy_lookup(sk, XFRM_POLICY_OUT, fl, family);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
err = xfrm_expand_policies(fl, family, pols,
&num_pols, &num_xfrms);
if (err < 0)
[IPSEC]: Fix potential dst leak in xfrm_lookup If we get an error during the actual policy lookup we don't free the original dst while the caller expects us to always free the original dst in case of error. This patch fixes that. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-11 20:38:08 +08:00
goto dropdst;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (num_pols) {
if (num_xfrms <= 0) {
drop_pols = num_pols;
goto no_transform;
}
xfrm: Fix a race in the xdst pcpu cache. We need to run xfrm_resolve_and_create_bundle() with bottom halves off. Otherwise we may reuse an already released dst_enty when the xfrm lookup functions are called from process context. Fixes: c30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Darius Ski <darius.ski@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-01-10 19:14:28 +08:00
local_bh_disable();
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xdst = xfrm_resolve_and_create_bundle(
pols, num_pols, fl,
family, dst_orig);
xfrm: Fix a race in the xdst pcpu cache. We need to run xfrm_resolve_and_create_bundle() with bottom halves off. Otherwise we may reuse an already released dst_enty when the xfrm lookup functions are called from process context. Fixes: c30d78c14a813db39a647b6a348b428 ("xfrm: add xdst pcpu cache") Reported-by: Darius Ski <darius.ski@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-01-10 19:14:28 +08:00
local_bh_enable();
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (IS_ERR(xdst)) {
xfrm_pols_put(pols, num_pols);
err = PTR_ERR(xdst);
goto dropdst;
xfrm: do not assume that template resolving always returns xfrms xfrm_resolve_and_create_bundle() assumed that, if policies indicated presence of xfrms, bundle template resolution would always return some xfrms. This is not true for 'use' level policies which can result in no xfrm's being applied if there is no suitable xfrm states. This fixes a crash by this incorrect assumption. Reported-by: George Spelvin <linux@horizon.com> Bisected-by: George Spelvin <linux@horizon.com> Tested-by: George Spelvin <linux@horizon.com> Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-07-13 05:29:42 +08:00
} else if (xdst == NULL) {
num_xfrms = 0;
drop_pols = num_pols;
goto no_transform;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
}
route = xdst->route;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (xdst == NULL) {
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
struct xfrm_flo xflo;
xflo.dst_orig = dst_orig;
xflo.flags = flags;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* To accelerate a bit... */
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if ((dst_orig->flags & DST_NOXFRM) ||
netns xfrm: lookup in netns Pass netns to xfrm_lookup()/__xfrm_lookup(). For that pass netns to flow_cache_lookup() and resolver callback. Take it from socket or netdevice. Stub DECnet to init_net. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:35:18 +08:00
!net->xfrm.policy_count[XFRM_POLICY_OUT])
[IPSEC]: Add ICMP host relookup support RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch implements this for ICMP traffic that originates from or terminates on localhost. This is activated on outbound with the new policy flag XFRM_POLICY_ICMP, and on inbound by the new state flag XFRM_STATE_ICMP. On inbound the policy check is now performed by the ICMP protocol so that it can repeat the policy check where necessary. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:43 +08:00
goto nopol;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
xdst = xfrm_bundle_lookup(net, fl, family, dir, &xflo);
if (xdst == NULL)
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
goto nopol;
xfrm_policy: make xfrm_bundle_lookup return xfrm dst object This allows to remove flow cache object embedded in struct xfrm_dst. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:25 +08:00
if (IS_ERR(xdst)) {
err = PTR_ERR(xdst);
[IPSEC]: Fix potential dst leak in xfrm_lookup If we get an error during the actual policy lookup we don't free the original dst while the caller expects us to always free the original dst in case of error. This patch fixes that. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-11 20:38:08 +08:00
goto dropdst;
[XFRM] Statistics: Add outbound-dropping error. o Increment PolError counter when flow_cache_lookup() returns errored pointer. o Increment NoStates counter at larval-drop. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-08 13:46:15 +08:00
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
num_pols = xdst->num_pols;
num_xfrms = xdst->num_xfrms;
xfrm: checkpatch errors with foo * bar This patch clean up some checkpatch errors like this: ERROR: "foo * bar" should be "foo *bar" ERROR: "(foo*)" should be "(foo *)" Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:47 +08:00
memcpy(pols, xdst->pols, sizeof(struct xfrm_policy *) * num_pols);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
route = xdst->route;
}
dst = &xdst->u.dst;
if (route == NULL && num_xfrms > 0) {
/* The only case when xfrm_bundle_lookup() returns a
* bundle with null route, is when the template could
* not be resolved. It means policies are there, but
* bundle could not be created, since we don't yet
* have the xfrm_state's. We need to wait for KM to
* negotiate new SA's or bail out with error.*/
if (net->xfrm.sysctl_larval_drop) {
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTNOSTATES);
xfrm: release dst_orig in case of error in xfrm_lookup() dst_orig should be released on error. Function like __xfrm_route_forward() expects that behavior. Since a recent commit, xfrm_lookup() may also be called by xfrm_lookup_route(), which expects the opposite. Let's introduce a new flag (XFRM_LOOKUP_KEEP_DST_REF) to tell what should be done in case of error. Fixes: f92ee61982d("xfrm: Generate blackhole routes only from route lookup functions") Signed-off-by: huaibin Wang <huaibin.wang@6wind.com> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-02-12 01:10:36 +08:00
err = -EREMOTE;
goto error;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
}
xfrm: Remove ancient sleeping when the SA is in acquire state We now queue packets to the policy if the states are not yet resolved, this replaces the ancient sleeping code. Also the sleeping can cause indefinite task hangs if the needed state does not get resolved. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-08-27 19:43:30 +08:00
err = -EAGAIN;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTNOSTATES);
goto error;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
no_transform:
if (num_pols == 0)
[IPSEC]: Add ICMP host relookup support RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch implements this for ICMP traffic that originates from or terminates on localhost. This is activated on outbound with the new policy flag XFRM_POLICY_ICMP, and on inbound by the new state flag XFRM_STATE_ICMP. On inbound the policy check is now performed by the ICMP protocol so that it can repeat the policy check where necessary. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:43 +08:00
goto nopol;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if ((flags & XFRM_LOOKUP_ICMP) &&
!(pols[0]->flags & XFRM_POLICY_ICMP)) {
err = -ENOENT;
[IPSEC]: Add ICMP host relookup support RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch implements this for ICMP traffic that originates from or terminates on localhost. This is activated on outbound with the new policy flag XFRM_POLICY_ICMP, and on inbound by the new state flag XFRM_STATE_ICMP. On inbound the policy check is now performed by the ICMP protocol so that it can repeat the policy check where necessary. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:43 +08:00
goto error;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
}
[IPSEC]: Add ICMP host relookup support RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch implements this for ICMP traffic that originates from or terminates on localhost. This is activated on outbound with the new policy flag XFRM_POLICY_ICMP, and on inbound by the new state flag XFRM_STATE_ICMP. On inbound the policy check is now performed by the ICMP protocol so that it can repeat the policy check where necessary. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:43 +08:00
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
for (i = 0; i < num_pols; i++)
pols[i]->curlft.use_time = get_seconds();
[IPSEC]: Add ICMP host relookup support RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch implements this for ICMP traffic that originates from or terminates on localhost. This is activated on outbound with the new policy flag XFRM_POLICY_ICMP, and on inbound by the new state flag XFRM_STATE_ICMP. On inbound the policy check is now performed by the ICMP protocol so that it can repeat the policy check where necessary. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:43 +08:00
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (num_xfrms < 0) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* Prohibit the flow */
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTPOLBLOCK);
[XFRM]: Always release dst_entry on error in xfrm_lookup Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-09-09 06:11:55 +08:00
err = -EPERM;
goto error;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
} else if (num_xfrms > 0) {
/* Flow transformed */
dst_release(dst_orig);
} else {
/* Flow passes untransformed */
dst_release(dst);
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
dst = dst_orig;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
ok:
xfrm_pols_put(pols, drop_pols);
ipv6: fix incorrect ipsec fragment Since commit ad0081e43a "ipv6: Fragment locally generated tunnel-mode IPSec6 packets as needed" the fragment of packets is incorrect. because tunnel mode needs IPsec headers and trailer for all fragments, while on transport mode it is sufficient to add the headers to the first fragment and the trailer to the last. so modify mtu and maxfraglen base on ipsec mode and if fragment is first or last. with my test,it work well(every fragment's size is the mtu) and does not trigger slow fragment path. Changes from v1: though optimization, mtu_prev and maxfraglen_prev can be delete. replace xfrm mode codes with dst_entry's new frag DST_XFRM_TUNNEL. add fuction ip6_append_data_mtu to make codes clearer. Signed-off-by: Gao feng <gaofeng@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-26 09:30:53 +08:00
if (dst && dst->xfrm &&
dst->xfrm->props.mode == XFRM_MODE_TUNNEL)
dst->flags |= DST_XFRM_TUNNEL;
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
return dst;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
nopol:
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
if (!(flags & XFRM_LOOKUP_ICMP)) {
dst = dst_orig;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
goto ok;
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
}
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
err = -ENOENT;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
error:
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
dst_release(dst);
[IPSEC]: Fix potential dst leak in xfrm_lookup If we get an error during the actual policy lookup we don't free the original dst while the caller expects us to always free the original dst in case of error. This patch fixes that. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-11 20:38:08 +08:00
dropdst:
xfrm: release dst_orig in case of error in xfrm_lookup() dst_orig should be released on error. Function like __xfrm_route_forward() expects that behavior. Since a recent commit, xfrm_lookup() may also be called by xfrm_lookup_route(), which expects the opposite. Let's introduce a new flag (XFRM_LOOKUP_KEEP_DST_REF) to tell what should be done in case of error. Fixes: f92ee61982d("xfrm: Generate blackhole routes only from route lookup functions") Signed-off-by: huaibin Wang <huaibin.wang@6wind.com> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-02-12 01:10:36 +08:00
if (!(flags & XFRM_LOOKUP_KEEP_DST_REF))
dst_release(dst_orig);
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
xfrm_pols_put(pols, drop_pols);
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
return ERR_PTR(err);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(xfrm_lookup);
xfrm: Generate blackhole routes only from route lookup functions Currently we genarate a blackhole route route whenever we have matching policies but can not resolve the states. Here we assume that dst_output() is called to kill the balckholed packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating blackhole routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: 2774c131b1d ("xfrm: Handle blackhole route creation via afinfo.") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:40 +08:00
/* Callers of xfrm_lookup_route() must ensure a call to dst_output().
* Otherwise we may send out blackholed packets.
*/
struct dst_entry *xfrm_lookup_route(struct net *net, struct dst_entry *dst_orig,
const struct flowi *fl,
inet: constify ip_route_output_flow() socket argument Very soon, TCP stack might call inet_csk_route_req(), which calls inet_csk_route_req() with an unlocked listener socket, so we need to make sure ip_route_output_flow() is not trying to change any field from its socket argument. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-25 22:39:10 +08:00
const struct sock *sk, int flags)
xfrm: Generate blackhole routes only from route lookup functions Currently we genarate a blackhole route route whenever we have matching policies but can not resolve the states. Here we assume that dst_output() is called to kill the balckholed packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating blackhole routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: 2774c131b1d ("xfrm: Handle blackhole route creation via afinfo.") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:40 +08:00
{
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
struct dst_entry *dst = xfrm_lookup(net, dst_orig, fl, sk,
xfrm: release dst_orig in case of error in xfrm_lookup() dst_orig should be released on error. Function like __xfrm_route_forward() expects that behavior. Since a recent commit, xfrm_lookup() may also be called by xfrm_lookup_route(), which expects the opposite. Let's introduce a new flag (XFRM_LOOKUP_KEEP_DST_REF) to tell what should be done in case of error. Fixes: f92ee61982d("xfrm: Generate blackhole routes only from route lookup functions") Signed-off-by: huaibin Wang <huaibin.wang@6wind.com> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-02-12 01:10:36 +08:00
flags | XFRM_LOOKUP_QUEUE |
XFRM_LOOKUP_KEEP_DST_REF);
xfrm: Generate blackhole routes only from route lookup functions Currently we genarate a blackhole route route whenever we have matching policies but can not resolve the states. Here we assume that dst_output() is called to kill the balckholed packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating blackhole routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: 2774c131b1d ("xfrm: Handle blackhole route creation via afinfo.") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:40 +08:00
if (IS_ERR(dst) && PTR_ERR(dst) == -EREMOTE)
return make_blackhole(net, dst_orig->ops->family, dst_orig);
return dst;
}
EXPORT_SYMBOL(xfrm_lookup_route);
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
static inline int
xfrm: Mark flowi arg to xfrm_type->reject() const. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-23 09:59:59 +08:00
xfrm_secpath_reject(int idx, struct sk_buff *skb, const struct flowi *fl)
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
{
struct xfrm_state *x;
if (!skb->sp || idx < 0 || idx >= skb->sp->len)
return 0;
x = skb->sp->xvec[idx];
if (!x->type->reject)
return 0;
[IPSEC]: Remove bogus ref count in xfrm_secpath_reject Constructs of the form xfrm_state_hold(x); foo(x); xfrm_state_put(x); tend to be broken because foo is either synchronous where this is totally unnecessary or if foo is asynchronous then the reference count is in the wrong spot. In the case of xfrm_secpath_reject, the function is synchronous and therefore we should just kill the reference count. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-10 04:24:07 +08:00
return x->type->reject(x, skb, fl);
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* When skb is transformed back to its "native" form, we have to
* check policy restrictions. At the moment we make this in maximally
* stupid way. Shame on me. :-) Of course, connected sockets must
* have policy cached at them.
*/
static inline int
xfrm: Const'ify ptr args to xfrm_state_ok. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:43:01 +08:00
xfrm_state_ok(const struct xfrm_tmpl *tmpl, const struct xfrm_state *x,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
unsigned short family)
{
if (xfrm_state_kern(x))
[IPSEC]: Fix the address family to refer encap_family Fix the address family to refer encap_family when comparing with a kernel generated xfrm_state Signed-off-by: Kazunori MIYAZAWA <miyazawa@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-14 04:57:16 +08:00
return tmpl->optional && !xfrm_state_addr_cmp(tmpl, x, tmpl->encap_family);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return x->id.proto == tmpl->id.proto &&
(x->id.spi == tmpl->id.spi || !tmpl->id.spi) &&
(x->props.reqid == tmpl->reqid || !tmpl->reqid) &&
x->props.mode == tmpl->mode &&
[IPSEC]: Fix catch-22 with algorithm IDs above 31 As it stands it's impossible to use any authentication algorithms with an ID above 31 portably. It just happens to work on x86 but fails miserably on ppc64. The reason is that we're using a bit mask to check the algorithm ID but the mask is only 32 bits wide. After looking at how this is used in the field, I have concluded that in the long term we should phase out state matching by IDs because this is made superfluous by the reqid feature. For current applications, the best solution IMHO is to allow all algorithms when the bit masks are all ~0. The following patch does exactly that. This bug was identified by IBM when testing on the ppc64 platform using the NULL authentication algorithm which has an ID of 251. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-04-22 15:46:42 +08:00
(tmpl->allalgs || (tmpl->aalgos & (1<<x->props.aalgo)) ||
[XFRM]: Restrict authentication algorithm only when inbound transformation protocol is IPsec. For Mobile IPv6 usage, routing header or destination options header is used and it doesn't require this comparison. It is checked only for IPsec template. Based on MIPL2 kernel patch. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 09:00:48 +08:00
!(xfrm_id_proto_match(tmpl->id.proto, IPSEC_PROTO_ANY))) &&
[XFRM]: Add XFRM_MODE_xxx for future use. Transformation mode is used as either IPsec transport or tunnel. It is required to add two more items, route optimization and inbound trigger for Mobile IPv6. Based on MIPL2 kernel patch. This patch was also written by: Ville Nuorvala <vnuorval@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-23 06:05:15 +08:00
!(x->props.mode != XFRM_MODE_TRANSPORT &&
xfrm_state_addr_cmp(tmpl, x, family));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
/*
* 0 or more than 0 is returned when validation is succeeded (either bypass
* because of optional transport mode, or next index of the mathced secpath
* state with the template.
* -1 is returned when no matching template is found.
* Otherwise "-2 - errored_index" is returned.
*/
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static inline int
xfrm: Const'ify ptr args to xfrm_policy_ok. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:43:33 +08:00
xfrm_policy_ok(const struct xfrm_tmpl *tmpl, const struct sec_path *sp, int start,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
unsigned short family)
{
int idx = start;
if (tmpl->optional) {
[XFRM]: Add XFRM_MODE_xxx for future use. Transformation mode is used as either IPsec transport or tunnel. It is required to add two more items, route optimization and inbound trigger for Mobile IPv6. Based on MIPL2 kernel patch. This patch was also written by: Ville Nuorvala <vnuorval@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-23 06:05:15 +08:00
if (tmpl->mode == XFRM_MODE_TRANSPORT)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return start;
} else
start = -1;
for (; idx < sp->len; idx++) {
[IPSEC]: Kill unused decap state structure This patch removes the *_decap_state structures which were previously used to share state between input/post_input. This is no longer needed. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-04-01 16:54:16 +08:00
if (xfrm_state_ok(tmpl, sp->xvec[idx], family))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return ++idx;
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
if (sp->xvec[idx]->props.mode != XFRM_MODE_TRANSPORT) {
if (start == -1)
start = -2-idx;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
return start;
}
[IPSEC]: Added xfrm_decode_session_reverse and xfrmX_policy_check_reverse RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch adds the functions xfrm_decode_session_reverse and xfrmX_policy_check_reverse so we can get the reverse flow to perform such a lookup. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:16 +08:00
int __xfrm_decode_session(struct sk_buff *skb, struct flowi *fl,
unsigned int family, int reverse)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: policy: make policy backend const Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:19 +08:00
const struct xfrm_policy_afinfo *afinfo = xfrm_policy_get_afinfo(family);
[MLSXFRM]: Flow based matching of xfrm policy and state This implements a seemless mechanism for xfrm policy selection and state matching based on the flow sid. This also includes the necessary SELinux enforcement pieces. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-07-25 14:29:07 +08:00
int err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (unlikely(afinfo == NULL))
return -EAFNOSUPPORT;
[IPSEC]: Added xfrm_decode_session_reverse and xfrmX_policy_check_reverse RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch adds the functions xfrm_decode_session_reverse and xfrmX_policy_check_reverse so we can get the reverse flow to perform such a lookup. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:16 +08:00
afinfo->decode_session(skb, fl, reverse);
net: Put flowi_* prefix on AF independent members of struct flowi I intend to turn struct flowi into a union of AF specific flowi structs. There will be a common structure that each variant includes first, much like struct sock_common. This is the first step to move in that direction. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-12 13:29:39 +08:00
err = security_xfrm_decode_session(skb, &fl->flowi_secid);
xfrm: policy: remove xfrm_policy_put_afinfo Alternative is to keep it an make the (unused) afinfo arg const to avoid the compiler warnings once the afinfo structs get constified. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:18 +08:00
rcu_read_unlock();
[MLSXFRM]: Flow based matching of xfrm policy and state This implements a seemless mechanism for xfrm policy selection and state matching based on the flow sid. This also includes the necessary SELinux enforcement pieces. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-07-25 14:29:07 +08:00
return err;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[IPSEC]: Added xfrm_decode_session_reverse and xfrmX_policy_check_reverse RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch adds the functions xfrm_decode_session_reverse and xfrmX_policy_check_reverse so we can get the reverse flow to perform such a lookup. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:16 +08:00
EXPORT_SYMBOL(__xfrm_decode_session);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Const'ify sec_path arg to secpath_has_nontransport. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 14:44:12 +08:00
static inline int secpath_has_nontransport(const struct sec_path *sp, int k, int *idxp)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
for (; k < sp->len; k++) {
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
if (sp->xvec[k]->props.mode != XFRM_MODE_TRANSPORT) {
[XFRM]: remove xerr_idxp from __xfrm_policy_check() It seems that during the MIPv6 respin, some code which was originally conditionally compiled around CONFIG_XFRM_ADVANCED was accidently left in after the config option was removed. This patch removes an extraneous pointer (xerr_idxp) which is no longer needed. Signed-off-by: James Morris <jmorris@namei.org> Acked-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-01 15:32:12 +08:00
*idxp = k;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 1;
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
return 0;
}
[NET] XFRM: Fix whitespace errors. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 22:25:29 +08:00
int __xfrm_policy_check(struct sock *sk, int dir, struct sk_buff *skb,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
unsigned short family)
{
netns xfrm: xfrm_policy_check in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:35:44 +08:00
struct net *net = dev_net(skb->dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct xfrm_policy *pol;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
struct xfrm_policy *pols[XFRM_POLICY_TYPE_MAX];
int npols = 0;
int xfrm_nr;
int pi;
[IPSEC]: Added xfrm_decode_session_reverse and xfrmX_policy_check_reverse RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch adds the functions xfrm_decode_session_reverse and xfrmX_policy_check_reverse so we can get the reverse flow to perform such a lookup. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:16 +08:00
int reverse;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct flowi fl;
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
int xerr_idx = -1;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IPSEC]: Added xfrm_decode_session_reverse and xfrmX_policy_check_reverse RFC 4301 requires us to relookup ICMP traffic that does not match any policies using the reverse of its payload. This patch adds the functions xfrm_decode_session_reverse and xfrmX_policy_check_reverse so we can get the reverse flow to perform such a lookup. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-13 02:44:16 +08:00
reverse = dir & ~XFRM_POLICY_MASK;
dir &= XFRM_POLICY_MASK;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (__xfrm_decode_session(skb, &fl, family, reverse) < 0) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINHDRERROR);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 0;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
[NETFILTER]: Handle NAT in IPsec policy checks Handle NAT of decapsulated IPsec packets by reconstructing the struct flowi of the original packet from the conntrack information for IPsec policy checks. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-01-07 15:06:30 +08:00
nf_nat_decode_session(skb, &fl, family);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* First, check used SA against their selectors. */
if (skb->sp) {
int i;
xfrm: checkpatch errors with space This patch cleanup some space errors. Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:46 +08:00
for (i = skb->sp->len-1; i >= 0; i--) {
[IPSEC]: Kill unused decap state structure This patch removes the *_decap_state structures which were previously used to share state between input/post_input. This is no longer needed. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-04-01 16:54:16 +08:00
struct xfrm_state *x = skb->sp->xvec[i];
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (!xfrm_selector_match(&x->sel, &fl, family)) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEMISMATCH);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 0;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
}
pol = NULL;
xfrm: take care of request sockets TCP SYNACK messages might now be attached to request sockets. XFRM needs to get back to a listener socket. Adds new helpers that might be used elsewhere : sk_to_full_sk() and sk_const_to_full_sk() Note: We also need to add RCU protection for xfrm lookups, now TCP/DCCP have lockless listener processing. This will be addressed in separate patches. Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener") Reported-by: Dave Jones <davej@codemonkey.org.uk> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-08 00:53:17 +08:00
sk = sk_to_full_sk(sk);
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
if (sk && sk->sk_policy[dir]) {
xfrm: Don't use sk_family for socket policy lookups On IPv4-mapped IPv6 addresses sk_family is AF_INET6, but the flow informations are created based on AF_INET. So the routing set up 'struct flowi4' but we try to access 'struct flowi6' what leads to an out of bounds access. Fix this by using the family we get with the dst_entry, like we do it for the standard policy lookup. Reported-by: Dmitry Vyukov <dvyukov@google.com> Tested-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-14 14:43:56 +08:00
pol = xfrm_sk_policy_lookup(sk, dir, &fl, family);
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (IS_ERR(pol)) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINPOLERROR);
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
return 0;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
IPsec: fix handling of errors for socket policies This treats the security errors encountered in the case of socket policy matching, the same as how these are treated in the case of main/sub policies, which is to return a full lookup failure. Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com> Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:35 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_policy: remove xfrm_policy_lookup This removes the wrapper and renames the __xfrm_policy_lookup variant to get rid of another place that used flow cache objects. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:24 +08:00
if (!pol)
pol = xfrm_policy_lookup(net, &fl, family, dir);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (IS_ERR(pol)) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINPOLERROR);
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
return 0;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
if (!pol) {
[XFRM]: remove xerr_idxp from __xfrm_policy_check() It seems that during the MIPv6 respin, some code which was originally conditionally compiled around CONFIG_XFRM_ADVANCED was accidently left in after the config option was removed. This patch removes an extraneous pointer (xerr_idxp) which is no longer needed. Signed-off-by: James Morris <jmorris@namei.org> Acked-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-01 15:32:12 +08:00
if (skb->sp && secpath_has_nontransport(skb->sp, 0, &xerr_idx)) {
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
xfrm_secpath_reject(xerr_idx, skb, &fl);
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINNOPOLS);
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
return 0;
}
return 1;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[NET]: Convert xtime.tv_sec to get_seconds() Where appropriate, convert references to xtime.tv_sec to the get_seconds() helper function. Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-05 08:12:44 +08:00
pol->curlft.use_time = get_seconds();
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
pols[0] = pol;
xfrm: checkpatch erros with space prohibited Fix checkpatch error "space prohibited xxx". Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:48 +08:00
npols++;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
#ifdef CONFIG_XFRM_SUB_POLICY
if (pols[0]->type != XFRM_POLICY_TYPE_MAIN) {
netns xfrm: xfrm_policy_check in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:35:44 +08:00
pols[1] = xfrm_policy_lookup_bytype(net, XFRM_POLICY_TYPE_MAIN,
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
&fl, family,
XFRM_POLICY_IN);
if (pols[1]) {
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (IS_ERR(pols[1])) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINPOLERROR);
IPsec: propagate security module errors up from flow_cache_lookup When a security module is loaded (in this case, SELinux), the security_xfrm_policy_lookup() hook can return an access denied permission (or other error). We were not handling that correctly, and in fact inverting the return logic and propagating a false "ok" back up to xfrm_lookup(), which then allowed packets to pass as if they were not associated with an xfrm policy. The way I was seeing the problem was when connecting via IPsec to a confined service on an SELinux box (vsftpd), which did not have the appropriate SELinux policy permissions to send packets via IPsec. The first SYNACK would be blocked, because of an uncached lookup via flow_cache_lookup(), which would fail to resolve an xfrm policy because the SELinux policy is checked at that point via the resolver. However, retransmitted SYNACKs would then find a cached flow entry when calling into flow_cache_lookup() with a null xfrm policy, which is interpreted by xfrm_lookup() as the packet not having any associated policy and similarly to the first case, allowing it to pass without transformation. The solution presented here is to first ensure that errno values are correctly propagated all the way back up through the various call chains from security_xfrm_policy_lookup(), and handled correctly. Then, flow_cache_lookup() is modified, so that if the policy resolver fails (typically a permission denied via the security module), the flow cache entry is killed rather than having a null policy assigned (which indicates that the packet can pass freely). This also forces any future lookups for the same flow to consult the security module (e.g. SELinux) for current security policy (rather than, say, caching the error on the flow cache entry). Signed-off-by: James Morris <jmorris@namei.org>
2006-10-06 04:42:27 +08:00
return 0;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
[NET]: Convert xtime.tv_sec to get_seconds() Where appropriate, convert references to xtime.tv_sec to the get_seconds() helper function. Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-03-05 08:12:44 +08:00
pols[1]->curlft.use_time = get_seconds();
xfrm: checkpatch erros with space prohibited Fix checkpatch error "space prohibited xxx". Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:48 +08:00
npols++;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
}
}
#endif
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (pol->action == XFRM_POLICY_ALLOW) {
struct sec_path *sp;
static struct sec_path dummy;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
struct xfrm_tmpl *tp[XFRM_MAX_DEPTH];
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
struct xfrm_tmpl *stp[XFRM_MAX_DEPTH];
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
struct xfrm_tmpl **tpp = tp;
int ti = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int i, k;
if ((sp = skb->sp) == NULL)
sp = &dummy;
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
for (pi = 0; pi < npols; pi++) {
if (pols[pi] != pol &&
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
pols[pi]->action != XFRM_POLICY_ALLOW) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINPOLBLOCK);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
goto reject;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
if (ti + pols[pi]->xfrm_nr >= XFRM_MAX_DEPTH) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINBUFFERERROR);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
goto reject_error;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
for (i = 0; i < pols[pi]->xfrm_nr; i++)
tpp[ti++] = &pols[pi]->xfrm_vec[i];
}
xfrm_nr = ti;
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
if (npols > 1) {
xfrm: Namespacify xfrm state/policy locks By semantics, xfrm layer is fully name space aware, so will the locks, e.g. xfrm_state/pocliy_lock. Ensure exclusive access into state/policy link list for different name space with one global lock is not right in terms of semantics aspect at first place, as they are indeed mutually independent with each other, but also more seriously causes scalability problem. One practical scenario is on a Open Network Stack, more than hundreds of lxc tenants acts as routers within one host, a global xfrm_state/policy_lock becomes the bottleneck. But onces those locks are decoupled in a per-namespace fashion, locks contend is just with in specific name space scope, without causing additional SPD/SAD access delay for other name space. Also this patch improve scalability while as without changing original xfrm behavior. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:50 +08:00
xfrm_tmpl_sort(stp, tpp, xfrm_nr, family, net);
[XFRM]: Add sorting interface for state and template. Under two transformation policies it is required to merge them. This is a platform to sort state for outbound and templates for inbound respectively. It will be used when Mobile IPv6 and IPsec are used at the same time. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 13:48:31 +08:00
tpp = stp;
}
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/* For each tunnel xfrm, find the first matching tmpl.
* For each tmpl before that, find corresponding xfrm.
* Order is _important_. Later we will implement
* some barriers, but at the moment barriers
* are implied between each two transformations.
*/
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
for (i = xfrm_nr-1, k = 0; i >= 0; i--) {
k = xfrm_policy_ok(tpp[i], sp, k, family);
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
if (k < 0) {
[XFRM]: remove xerr_idxp from __xfrm_policy_check() It seems that during the MIPv6 respin, some code which was originally conditionally compiled around CONFIG_XFRM_ADVANCED was accidently left in after the config option was removed. This patch removes an extraneous pointer (xerr_idxp) which is no longer needed. Signed-off-by: James Morris <jmorris@namei.org> Acked-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-09-01 15:32:12 +08:00
if (k < -1)
/* "-2 - errored_index" returned */
xerr_idx = -(2+k);
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINTMPLMISMATCH);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
goto reject;
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (secpath_has_nontransport(sp, k, &xerr_idx)) {
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINTMPLMISMATCH);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
goto reject;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
xfrm_pols_put(pols, npols);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 1;
}
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMINPOLBLOCK);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
reject:
[XFRM]: Trace which secpath state is reject factor. For Mobile IPv6 usage, it is required to trace which secpath state is reject factor in order to notify it to user space (to know the address which cannot be used route optimized communication). Based on MIPL2 kernel patch. This patch was also written by: Henrik Petander <petander@tcs.hut.fi> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 11:41:00 +08:00
xfrm_secpath_reject(xerr_idx, skb, &fl);
[XFRM] POLICY: sub policy support. Sub policy is introduced. Main and sub policy are applied the same flow. (Policy that current kernel uses is named as main.) It is required another transformation policy management to keep IPsec and Mobile IPv6 lives separate. Policy which lives shorter time in kernel should be a sub i.e. normally main is for IPsec and sub is for Mobile IPv6. (Such usage as two IPsec policies on different database can be used, too.) Limitation or TODOs: - Sub policy is not supported for per socket one (it is always inserted as main). - Current kernel makes cached outbound with flowi to skip searching database. However this patch makes it disabled only when "two policies are used and the first matched one is bypass case" because neither flowi nor bundle information knows about transformation template size. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2006-08-24 13:43:30 +08:00
reject_error:
xfrm_pols_put(pols, npols);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 0;
}
EXPORT_SYMBOL(__xfrm_policy_check);
int __xfrm_route_forward(struct sk_buff *skb, unsigned short family)
{
netns xfrm: xfrm_route_forward() in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:36:13 +08:00
struct net *net = dev_net(skb->dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct flowi fl;
net: skb->dst accessors Define three accessors to get/set dst attached to a skb struct dst_entry *skb_dst(const struct sk_buff *skb) void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst) void skb_dst_drop(struct sk_buff *skb) This one should replace occurrences of : dst_release(skb->dst) skb->dst = NULL; Delete skb->dst field Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-02 13:19:30 +08:00
struct dst_entry *dst;
xfrm: fix __xfrm_route_forward() This function should return 0 in case of error, 1 if OK commit 452edd598f60522 (xfrm: Return dst directly from xfrm_lookup()) got it wrong. Reported-and-bisected-by: Michael Smith <msmith@cbnco.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-16 06:26:43 +08:00
int res = 1;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
if (xfrm_decode_session(skb, &fl, family) < 0) {
xfrm: Introduce LINUX_MIB_XFRMFWDHDRERROR XFRMINHDRERROR counter is ambigous when validating forwarding path. It makes it tricky to debug when you have both in and fwd validation. Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-02-18 11:35:07 +08:00
XFRM_INC_STATS(net, LINUX_MIB_XFRMFWDHDRERROR);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 0;
[XFRM]: Support to increment packet dropping statistics. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:43:36 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: force a dst reference in __xfrm_route_forward() Packets going through __xfrm_route_forward() have a not refcounted dst entry, since we enabled a noref forwarding path. xfrm_lookup() might incorrectly release this dst entry. It's a bit late to make invasive changes in xfrm_lookup(), so lets force a refcount in this path. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-06-01 18:04:49 +08:00
skb_dst_force(skb);
net: skb->dst accessors Define three accessors to get/set dst attached to a skb struct dst_entry *skb_dst(const struct sk_buff *skb) void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst) void skb_dst_drop(struct sk_buff *skb) This one should replace occurrences of : dst_release(skb->dst) skb->dst = NULL; Delete skb->dst field Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-02 13:19:30 +08:00
xfrm: Generate queueing routes only from route lookup functions Currently we genarate a queueing route if we have matching policies but can not resolve the states and the sysctl xfrm_larval_drop is disabled. Here we assume that dst_output() is called to kill the queued packets. Unfortunately this assumption is not true in all cases, so it is possible that these packets leave the system unwanted. We fix this by generating queueing routes only from the route lookup functions, here we can guarantee a call to dst_output() afterwards. Fixes: a0073fe18e71 ("xfrm: Add a state resolution packet queue") Reported-by: Konstantinos Kolelis <k.kolelis@sirrix.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-09-16 16:08:49 +08:00
dst = xfrm_lookup(net, skb_dst(skb), &fl, NULL, XFRM_LOOKUP_QUEUE);
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
if (IS_ERR(dst)) {
xfrm: fix __xfrm_route_forward() This function should return 0 in case of error, 1 if OK commit 452edd598f60522 (xfrm: Return dst directly from xfrm_lookup()) got it wrong. Reported-and-bisected-by: Michael Smith <msmith@cbnco.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-16 06:26:43 +08:00
res = 0;
xfrm: Return dst directly from xfrm_lookup() Instead of on the stack. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-03-03 05:27:41 +08:00
dst = NULL;
}
net: skb->dst accessors Define three accessors to get/set dst attached to a skb struct dst_entry *skb_dst(const struct sk_buff *skb) void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst) void skb_dst_drop(struct sk_buff *skb) This one should replace occurrences of : dst_release(skb->dst) skb->dst = NULL; Delete skb->dst field Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-02 13:19:30 +08:00
skb_dst_set(skb, dst);
return res;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(__xfrm_route_forward);
[IPSEC]: Validate properly in xfrm_dst_check() If dst->obsolete is -1, this is a signal from the bundle creator that we want the XFRM dst and the dsts that it references to be validated on every use. I misunderstood this intention when I changed xfrm_dst_check() to always return NULL. Now, when we purge a dst entry, by running dst_free() on it. This will set the dst->obsolete to a positive integer, and we want to return NULL in that case so that the socket does a relookup for the route. Thus, if dst->obsolete<0, let stale_bundle() validate the state, else always return NULL. In general, we need to do things more intelligently here because we flush too much state during rule changes. Herbert Xu has some ideas wherein the key manager gives us some help in this area. We can also use smarter state management algorithms inside of the kernel as well. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-14 09:55:53 +08:00
/* Optimize later using cookies and generation ids. */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static struct dst_entry *xfrm_dst_check(struct dst_entry *dst, u32 cookie)
{
[IPSEC]: Validate properly in xfrm_dst_check() If dst->obsolete is -1, this is a signal from the bundle creator that we want the XFRM dst and the dsts that it references to be validated on every use. I misunderstood this intention when I changed xfrm_dst_check() to always return NULL. Now, when we purge a dst entry, by running dst_free() on it. This will set the dst->obsolete to a positive integer, and we want to return NULL in that case so that the socket does a relookup for the route. Thus, if dst->obsolete<0, let stale_bundle() validate the state, else always return NULL. In general, we need to do things more intelligently here because we flush too much state during rule changes. Herbert Xu has some ideas wherein the key manager gives us some help in this area. We can also use smarter state management algorithms inside of the kernel as well. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-14 09:55:53 +08:00
/* Code (such as __xfrm4_bundle_create()) sets dst->obsolete
net: Document dst->obsolete better. Add a big comment explaining how the field works, and use defines instead of magic constants for the values assigned to it. Suggested by Joe Perches. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-07-20 03:31:33 +08:00
* to DST_OBSOLETE_FORCE_CHK to force all XFRM destinations to
* get validated by dst_ops->check on every use. We do this
* because when a normal route referenced by an XFRM dst is
* obsoleted we do not go looking around for all parent
* referencing XFRM dsts so that we can invalidate them. It
* is just too much work. Instead we make the checks here on
* every use. For example:
[IPSEC]: Validate properly in xfrm_dst_check() If dst->obsolete is -1, this is a signal from the bundle creator that we want the XFRM dst and the dsts that it references to be validated on every use. I misunderstood this intention when I changed xfrm_dst_check() to always return NULL. Now, when we purge a dst entry, by running dst_free() on it. This will set the dst->obsolete to a positive integer, and we want to return NULL in that case so that the socket does a relookup for the route. Thus, if dst->obsolete<0, let stale_bundle() validate the state, else always return NULL. In general, we need to do things more intelligently here because we flush too much state during rule changes. Herbert Xu has some ideas wherein the key manager gives us some help in this area. We can also use smarter state management algorithms inside of the kernel as well. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-14 09:55:53 +08:00
*
* XFRM dst A --> IPv4 dst X
*
* X is the "xdst->route" of A (X is also the "dst->path" of A
* in this example). If X is marked obsolete, "A" will not
* notice. That's what we are validating here via the
* stale_bundle() check.
*
xfrm: take refcnt of dst when creating struct xfrm_dst bundle During the creation of xfrm_dst bundle, always take ref count when allocating the dst. This way, xfrm_bundle_create() will form a linked list of dst with dst->child pointing to a ref counted dst child. And the returned dst pointer is also ref counted. This makes the link from the flow cache to this dst now ref counted properly. As the dst is always ref counted properly, we can safely mark DST_NOGC flag so dst_release() will release dst based on refcnt only. And dst gc is no longer needed and all dst_free() and its related function calls should be replaced with dst_release() or dst_release_immediate(). The special handling logic for dst->child in dst_destroy() can be replaced with a simple dst_release_immediate() call on the child to release the whole list linked by dst->child pointer. Previously used DST_NOHASH flag is not needed anymore as well. The reason that DST_NOHASH is used in the existing code is mainly to prevent the dst inserted in the fib tree to be wrongly destroyed during the deletion of the xfrm_dst bundle. So in the existing code, DST_NOHASH flag is marked in all the dst children except the one which is in the fib tree. However, with this patch series to remove dst gc logic and release dst only based on ref count, it is safe to release all the children from a xfrm_dst bundle as long as the dst children are all ref counted properly which is already the case in the existing code. So, this patch removes the use of DST_NOHASH flag. Signed-off-by: Wei Wang <weiwan@google.com> Acked-by: Martin KaFai Lau <kafai@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-06-18 01:42:38 +08:00
* When a dst is removed from the fib tree, DST_OBSOLETE_DEAD will
* be marked on it.
xfrm: remove flow cache After rcu conversions performance degradation in forward tests isn't that noticeable anymore. See next patch for some numbers. A followup patcg could then also remove genid from the policies as we do not cache bundles anymore. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:26 +08:00
* This will force stale_bundle() to fail on any xdst bundle with
xfrm: take refcnt of dst when creating struct xfrm_dst bundle During the creation of xfrm_dst bundle, always take ref count when allocating the dst. This way, xfrm_bundle_create() will form a linked list of dst with dst->child pointing to a ref counted dst child. And the returned dst pointer is also ref counted. This makes the link from the flow cache to this dst now ref counted properly. As the dst is always ref counted properly, we can safely mark DST_NOGC flag so dst_release() will release dst based on refcnt only. And dst gc is no longer needed and all dst_free() and its related function calls should be replaced with dst_release() or dst_release_immediate(). The special handling logic for dst->child in dst_destroy() can be replaced with a simple dst_release_immediate() call on the child to release the whole list linked by dst->child pointer. Previously used DST_NOHASH flag is not needed anymore as well. The reason that DST_NOHASH is used in the existing code is mainly to prevent the dst inserted in the fib tree to be wrongly destroyed during the deletion of the xfrm_dst bundle. So in the existing code, DST_NOHASH flag is marked in all the dst children except the one which is in the fib tree. However, with this patch series to remove dst gc logic and release dst only based on ref count, it is safe to release all the children from a xfrm_dst bundle as long as the dst children are all ref counted properly which is already the case in the existing code. So, this patch removes the use of DST_NOHASH flag. Signed-off-by: Wei Wang <weiwan@google.com> Acked-by: Martin KaFai Lau <kafai@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-06-18 01:42:38 +08:00
* this dst linked in it.
[IPSEC]: Perform SA switchover immediately. When we insert a new xfrm_state which potentially subsumes an existing one, make sure all cached bundles are flushed so that the new SA is used immediately. Signed-off-by: David S. Miller <davem@davemloft.net>
2005-12-20 06:23:23 +08:00
*/
[IPSEC]: Validate properly in xfrm_dst_check() If dst->obsolete is -1, this is a signal from the bundle creator that we want the XFRM dst and the dsts that it references to be validated on every use. I misunderstood this intention when I changed xfrm_dst_check() to always return NULL. Now, when we purge a dst entry, by running dst_free() on it. This will set the dst->obsolete to a positive integer, and we want to return NULL in that case so that the socket does a relookup for the route. Thus, if dst->obsolete<0, let stale_bundle() validate the state, else always return NULL. In general, we need to do things more intelligently here because we flush too much state during rule changes. Herbert Xu has some ideas wherein the key manager gives us some help in this area. We can also use smarter state management algorithms inside of the kernel as well. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-14 09:55:53 +08:00
if (dst->obsolete < 0 && !stale_bundle(dst))
return dst;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return NULL;
}
static int stale_bundle(struct dst_entry *dst)
{
xfrm: Remove family arg from xfrm_bundle_ok The family arg is not used any more, so remove it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-06-30 07:18:20 +08:00
return !xfrm_bundle_ok((struct xfrm_dst *)dst);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[IPSEC]: Store idev entries I found a bug that stopped IPsec/IPv6 from working. About a month ago IPv6 started using rt6i_idev->dev on the cached socket dst entries. If the cached socket dst entry is IPsec, then rt6i_idev will be NULL. Since we want to look at the rt6i_idev of the original route in this case, the easiest fix is to store rt6i_idev in the IPsec dst entry just as we do for a number of other IPv6 route attributes. Unfortunately this means that we need some new code to handle the references to rt6i_idev. That's why this patch is bigger than it would otherwise be. I've also done the same thing for IPv4 since it is conceivable that once these idev attributes start getting used for accounting, we probably need to dereference them for IPv4 IPsec entries too. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-05-04 07:27:10 +08:00
void xfrm_dst_ifdown(struct dst_entry *dst, struct net_device *dev)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
net: Create and use new helper xfrm_dst_child(). Only IPSEC routes have a non-NULL dst->child pointer. And IPSEC routes are identified by a non-NULL dst->xfrm pointer. Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-29 04:40:22 +08:00
while ((dst = xfrm_dst_child(dst)) && dst->xfrm && dst->dev == dev) {
[NET] NETNS: Omit net_device->nd_net without CONFIG_NET_NS. Introduce per-net_device inlines: dev_net(), dev_net_set(). Without CONFIG_NET_NS, no namespace other than &init_net exists. Let's explicitly define them to help compiler optimizations. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
2008-03-25 20:47:49 +08:00
dst->dev = dev_net(dev)->loopback_dev;
[NET]: Dynamically allocate the loopback device, part 1. This patch replaces all occurences to the static variable loopback_dev to a pointer loopback_dev. That provides the mindless, trivial, uninteressting change part for the dynamic allocation for the loopback. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com> Acked-By: Kirill Korotaev <dev@sw.ru> Acked-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-26 10:16:28 +08:00
dev_hold(dst->dev);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
dev_put(dev);
}
}
[IPSEC]: Store idev entries I found a bug that stopped IPsec/IPv6 from working. About a month ago IPv6 started using rt6i_idev->dev on the cached socket dst entries. If the cached socket dst entry is IPsec, then rt6i_idev will be NULL. Since we want to look at the rt6i_idev of the original route in this case, the easiest fix is to store rt6i_idev in the IPsec dst entry just as we do for a number of other IPv6 route attributes. Unfortunately this means that we need some new code to handle the references to rt6i_idev. That's why this patch is bigger than it would otherwise be. I've also done the same thing for IPv4 since it is conceivable that once these idev attributes start getting used for accounting, we probably need to dereference them for IPv4 IPsec entries too. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-05-04 07:27:10 +08:00
EXPORT_SYMBOL(xfrm_dst_ifdown);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static void xfrm_link_failure(struct sk_buff *skb)
{
/* Impossible. Such dst must be popped before reaches point of failure. */
}
static struct dst_entry *xfrm_negative_advice(struct dst_entry *dst)
{
if (dst) {
if (dst->obsolete) {
dst_release(dst);
dst = NULL;
}
}
return dst;
}
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
static void xfrm_init_pmtu(struct xfrm_dst **bundle, int nr)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
while (nr--) {
struct xfrm_dst *xdst = bundle[nr];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
u32 pmtu, route_mtu_cached;
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
struct dst_entry *dst;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
dst = &xdst->u.dst;
net: Create and use new helper xfrm_dst_child(). Only IPSEC routes have a non-NULL dst->child pointer. And IPSEC routes are identified by a non-NULL dst->xfrm pointer. Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-29 04:40:22 +08:00
pmtu = dst_mtu(xfrm_dst_child(dst));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xdst->child_mtu_cached = pmtu;
pmtu = xfrm_state_mtu(dst->xfrm, pmtu);
route_mtu_cached = dst_mtu(xdst->route);
xdst->route_mtu_cached = route_mtu_cached;
if (pmtu > route_mtu_cached)
pmtu = route_mtu_cached;
net: Abstract away all dst_entry metrics accesses. Use helper functions to hide all direct accesses, especially writes, to dst_entry metrics values. This will allow us to: 1) More easily change how the metrics are stored. 2) Implement COW for metrics. In particular this will help us put metrics into the inetpeer cache if that is what we end up doing. We can make the _metrics member a pointer instead of an array, initially have it point at the read-only metrics in the FIB, and then on the first set grab an inetpeer entry and point the _metrics member there. Signed-off-by: David S. Miller <davem@davemloft.net> Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
2010-12-09 13:16:57 +08:00
dst_metric_set(dst, RTAX_MTU, pmtu);
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
/* Check that the bundle accepts the flow and its components are
* still valid.
*/
xfrm: Remove family arg from xfrm_bundle_ok The family arg is not used any more, so remove it. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-06-30 07:18:20 +08:00
static int xfrm_bundle_ok(struct xfrm_dst *first)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
struct xfrm_dst *bundle[XFRM_MAX_DEPTH];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct dst_entry *dst = &first->u.dst;
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
struct xfrm_dst *xdst;
int start_from, nr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
u32 mtu;
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
if (!dst_check(xfrm_dst_path(dst), ((struct xfrm_dst *)dst)->path_cookie) ||
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
(dst->dev && !netif_running(dst->dev)))
return 0;
xfrm: Add a state resolution packet queue As the default, we blackhole packets until the key manager resolves the states. This patch implements a packet queue where IPsec packets are queued until the states are resolved. We generate a dummy xfrm bundle, the output routine of the returned route enqueues the packet to a per policy queue and arms a timer that checks for state resolution when dst_output() is called. Once the states are resolved, the packets are sent out of the queue. If the states are not resolved after some time, the queue is flushed. This patch keeps the defaut behaviour to blackhole packets as long as we have no states. To enable the packet queue the sysctl xfrm_larval_drop must be switched off. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-02-05 19:52:55 +08:00
if (dst->flags & DST_XFRM_QUEUE)
return 1;
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
start_from = nr = 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
do {
struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
if (dst->xfrm->km.state != XFRM_STATE_VALID)
return 0;
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
if (xdst->xfrm_genid != dst->xfrm->genid)
return 0;
xfrm: check bundle policy existance before dereferencing it Fix the bundle validation code to not assume having a valid policy. When we have multiple transformations for a xfrm policy, the bundle instance will be a chain of bundles with only the first one having the policy reference. When policy_genid is bumped it will expire the first bundle in the chain which is equivalent of expiring the whole chain. Reported-bisected-and-tested-by: Justin P. Mattock <justinmattock@gmail.com> Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-06-25 05:35:00 +08:00
if (xdst->num_pols > 0 &&
xdst->policy_genid != atomic_read(&xdst->pols[0]->genid))
[XFRM]: Add generation count to xfrm_state and xfrm_dst. Each xfrm_state inserted gets a new generation counter value. When a bundle is created, the xfrm_dst objects get the current generation counter of the xfrm_state they will attach to at dst->xfrm. xfrm_bundle_ok() will return false if it sees an xfrm_dst with a generation count different from the generation count of the xfrm_state that dst points to. This provides a facility by which to passively and cheaply invalidate cached IPSEC routes during SA database changes. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 18:18:09 +08:00
return 0;
[XFRM] IPV6: Restrict bundle reusing For outbound transformation, bundle is checked whether it is suitable for current flow to be reused or not. In such IPv6 case as below, transformation may apply incorrect bundle for the flow instead of creating another bundle: - The policy selector has destination prefix length < 128 (Two or more addresses can be matched it) - Its bundle holds dst entry of default route whose prefix length < 128 (Previous traffic was used such route as next hop) - The policy and the bundle were used a transport mode state and this time flow address is not matched the bundled state. This issue is found by Mobile IPv6 usage to protect mobility signaling by IPsec, but it is not a Mobile IPv6 specific. This patch adds strict check to xfrm_bundle_ok() for each state mode and address when prefix length is less than 128. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 10:12:01 +08:00
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
bundle[nr++] = xdst;
net: Create and use new helper xfrm_dst_child(). Only IPSEC routes have a non-NULL dst->child pointer. And IPSEC routes are identified by a non-NULL dst->xfrm pointer. Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-29 04:40:22 +08:00
mtu = dst_mtu(xfrm_dst_child(dst));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (xdst->child_mtu_cached != mtu) {
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
start_from = nr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xdst->child_mtu_cached = mtu;
}
From: Kazunori Miyazawa <kazunori@miyazawa.org> [XFRM] Call dst_check() with appropriate cookie This fixes infinite loop issue with IPv6 tunnel mode. Signed-off-by: Kazunori Miyazawa <kazunori@miyazawa.org> Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2005-05-27 03:58:04 +08:00
if (!dst_check(xdst->route, xdst->route_cookie))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 0;
mtu = dst_mtu(xdst->route);
if (xdst->route_mtu_cached != mtu) {
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
start_from = nr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xdst->route_mtu_cached = mtu;
}
net: Create and use new helper xfrm_dst_child(). Only IPSEC routes have a non-NULL dst->child pointer. And IPSEC routes are identified by a non-NULL dst->xfrm pointer. Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-29 04:40:22 +08:00
dst = xfrm_dst_child(dst);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
} while (dst->xfrm);
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
if (likely(!start_from))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return 1;
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
xdst = bundle[start_from - 1];
mtu = xdst->child_mtu_cached;
while (start_from--) {
dst = &xdst->u.dst;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
mtu = xfrm_state_mtu(dst->xfrm, mtu);
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
if (mtu > xdst->route_mtu_cached)
mtu = xdst->route_mtu_cached;
net: Abstract away all dst_entry metrics accesses. Use helper functions to hide all direct accesses, especially writes, to dst_entry metrics values. This will allow us to: 1) More easily change how the metrics are stored. 2) Implement COW for metrics. In particular this will help us put metrics into the inetpeer cache if that is what we end up doing. We can make the _metrics member a pointer instead of an array, initially have it point at the read-only metrics in the FIB, and then on the first set grab an inetpeer entry and point the _metrics member there. Signed-off-by: David S. Miller <davem@davemloft.net> Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
2010-12-09 13:16:57 +08:00
dst_metric_set(dst, RTAX_MTU, mtu);
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
if (!start_from)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
break;
xfrm: Stop using dst->next in bundle construction. While building ipsec bundles, blocks of xfrm dsts are linked together using dst->next from bottom to the top. The only thing this is used for is initializing the pmtu values of the xfrm stack, and for updating the mtu values at xfrm_bundle_ok() time. The bundle pmtu entries must be processed in this order so that pmtu values lower in the stack of routes can propagate up to the higher ones. Avoid using dst->next by simply maintaining an array of dst pointers as we already do for the xfrm_state objects when building the bundle. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:41:01 +08:00
xdst = bundle[start_from - 1];
xdst->child_mtu_cached = mtu;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
return 1;
}
net: Abstract default ADVMSS behind an accessor. Make all RTAX_ADVMSS metric accesses go through a new helper function, dst_metric_advmss(). Leave the actual default metric as "zero" in the real metric slot, and compute the actual default value dynamically via a new dst_ops AF specific callback. For stacked IPSEC routes, we use the advmss of the path which preserves existing behavior. Unlike ipv4/ipv6, DecNET ties the advmss to the mtu and thus updates advmss on pmtu updates. This inconsistency in advmss handling results in more raw metric accesses than I wish we ended up with. Signed-off-by: David S. Miller <davem@davemloft.net>
2010-12-14 04:52:14 +08:00
static unsigned int xfrm_default_advmss(const struct dst_entry *dst)
{
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
return dst_metric_advmss(xfrm_dst_path(dst));
net: Abstract default ADVMSS behind an accessor. Make all RTAX_ADVMSS metric accesses go through a new helper function, dst_metric_advmss(). Leave the actual default metric as "zero" in the real metric slot, and compute the actual default value dynamically via a new dst_ops AF specific callback. For stacked IPSEC routes, we use the advmss of the path which preserves existing behavior. Unlike ipv4/ipv6, DecNET ties the advmss to the mtu and thus updates advmss on pmtu updates. This inconsistency in advmss handling results in more raw metric accesses than I wish we ended up with. Signed-off-by: David S. Miller <davem@davemloft.net>
2010-12-14 04:52:14 +08:00
}
net: Rename the dst_opt default_mtu method to mtu We plan to invoke the dst_opt->default_mtu() method unconditioally from dst_mtu(). So rename the method to dst_opt->mtu() to match the name with the new meaning. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-23 10:12:51 +08:00
static unsigned int xfrm_mtu(const struct dst_entry *dst)
net: Abstract default MTU metric calculation behind an accessor. Like RTAX_ADVMSS, make the default calculation go through a dst_ops method rather than caching the computation in the routing cache entries. Now dst metrics are pretty much left as-is when new entries are created, thus optimizing metric sharing becomes a real possibility. Signed-off-by: David S. Miller <davem@davemloft.net>
2010-12-15 05:01:14 +08:00
{
net: Move mtu handling down to the protocol depended handlers We move all mtu handling from dst_mtu() down to the protocol layer. So each protocol can implement the mtu handling in a different manner. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-23 10:13:31 +08:00
unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
return mtu ? : dst_mtu(xfrm_dst_path(dst));
net: Abstract default MTU metric calculation behind an accessor. Like RTAX_ADVMSS, make the default calculation go through a dst_ops method rather than caching the computation in the routing cache entries. Now dst metrics are pretty much left as-is when new entries are created, thus optimizing metric sharing becomes a real possibility. Signed-off-by: David S. Miller <davem@davemloft.net>
2010-12-15 05:01:14 +08:00
}
xfrm: provide correct dst in xfrm_neigh_lookup Fix xfrm_neigh_lookup to provide dst->path to the neigh_lookup dst_ops method. When skb is provided, the IP address in packet should already match the dst->path address family. But for the non-skb case, we should consider the last tunnel address as nexthop address. Fixes: f894cbf847c9 ("net: Add optional SKB arg to dst_ops->neigh_lookup().") Signed-off-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-25 23:57:43 +08:00
static const void *xfrm_get_dst_nexthop(const struct dst_entry *dst,
const void *daddr)
net: add confirm_neigh method to dst_ops Add confirm_neigh method to dst_ops and use it from IPv4 and IPv6 to lookup and confirm the neighbour. Its usage via the new helper dst_confirm_neigh() should be restricted to MSG_PROBE users for performance reasons. For XFRM prefer the last tunnel address, if present. With help from Steffen Klassert. Signed-off-by: Julian Anastasov <ja@ssi.bg> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-07 05:14:15 +08:00
{
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
while (dst->xfrm) {
net: add confirm_neigh method to dst_ops Add confirm_neigh method to dst_ops and use it from IPv4 and IPv6 to lookup and confirm the neighbour. Its usage via the new helper dst_confirm_neigh() should be restricted to MSG_PROBE users for performance reasons. For XFRM prefer the last tunnel address, if present. With help from Steffen Klassert. Signed-off-by: Julian Anastasov <ja@ssi.bg> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-07 05:14:15 +08:00
const struct xfrm_state *xfrm = dst->xfrm;
xfrm: Fix infinite loop in xfrm_get_dst_nexthop with transport mode. On transport mode we forget to fetch the child dst_entry before we continue the while loop, this leads to an infinite loop. Fix this by fetching the child dst_entry before we continue the while loop. Fixes: 0f6c480f23f4 ("xfrm: Move dst->path into struct xfrm_dst") Reported-by: syzbot+7d03c810e50aaedef98a@syzkaller.appspotmail.com Tested-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2018-02-19 14:44:07 +08:00
dst = xfrm_dst_child(dst);
net: add confirm_neigh method to dst_ops Add confirm_neigh method to dst_ops and use it from IPv4 and IPv6 to lookup and confirm the neighbour. Its usage via the new helper dst_confirm_neigh() should be restricted to MSG_PROBE users for performance reasons. For XFRM prefer the last tunnel address, if present. With help from Steffen Klassert. Signed-off-by: Julian Anastasov <ja@ssi.bg> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-07 05:14:15 +08:00
if (xfrm->props.mode == XFRM_MODE_TRANSPORT)
continue;
if (xfrm->type->flags & XFRM_TYPE_REMOTE_COADDR)
daddr = xfrm->coaddr;
else if (!(xfrm->type->flags & XFRM_TYPE_LOCAL_COADDR))
daddr = &xfrm->id.daddr;
}
xfrm: provide correct dst in xfrm_neigh_lookup Fix xfrm_neigh_lookup to provide dst->path to the neigh_lookup dst_ops method. When skb is provided, the IP address in packet should already match the dst->path address family. But for the non-skb case, we should consider the last tunnel address as nexthop address. Fixes: f894cbf847c9 ("net: Add optional SKB arg to dst_ops->neigh_lookup().") Signed-off-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-25 23:57:43 +08:00
return daddr;
}
static struct neighbour *xfrm_neigh_lookup(const struct dst_entry *dst,
struct sk_buff *skb,
const void *daddr)
{
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
const struct dst_entry *path = xfrm_dst_path(dst);
xfrm: provide correct dst in xfrm_neigh_lookup Fix xfrm_neigh_lookup to provide dst->path to the neigh_lookup dst_ops method. When skb is provided, the IP address in packet should already match the dst->path address family. But for the non-skb case, we should consider the last tunnel address as nexthop address. Fixes: f894cbf847c9 ("net: Add optional SKB arg to dst_ops->neigh_lookup().") Signed-off-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-25 23:57:43 +08:00
if (!skb)
daddr = xfrm_get_dst_nexthop(dst, daddr);
return path->ops->neigh_lookup(path, skb, daddr);
}
static void xfrm_confirm_neigh(const struct dst_entry *dst, const void *daddr)
{
xfrm: Move dst->path into struct xfrm_dst The first member of an IPSEC route bundle chain sets it's dst->path to the underlying ipv4/ipv6 route that carries the bundle. Stated another way, if one were to follow the xfrm_dst->child chain of the bundle, the final non-NULL pointer would be the path and point to either an ipv4 or an ipv6 route. This is largely used to make sure that PMTU events propagate down to the correct ipv4 or ipv6 route. When we don't have the top of an IPSEC bundle 'dst->path == dst'. Move it down into xfrm_dst and key off of dst->xfrm. Signed-off-by: David S. Miller <davem@davemloft.net> Reviewed-by: Eric Dumazet <edumazet@google.com>
2017-11-29 04:40:46 +08:00
const struct dst_entry *path = xfrm_dst_path(dst);
xfrm: provide correct dst in xfrm_neigh_lookup Fix xfrm_neigh_lookup to provide dst->path to the neigh_lookup dst_ops method. When skb is provided, the IP address in packet should already match the dst->path address family. But for the non-skb case, we should consider the last tunnel address as nexthop address. Fixes: f894cbf847c9 ("net: Add optional SKB arg to dst_ops->neigh_lookup().") Signed-off-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-25 23:57:43 +08:00
daddr = xfrm_get_dst_nexthop(dst, daddr);
net: add confirm_neigh method to dst_ops Add confirm_neigh method to dst_ops and use it from IPv4 and IPv6 to lookup and confirm the neighbour. Its usage via the new helper dst_confirm_neigh() should be restricted to MSG_PROBE users for performance reasons. For XFRM prefer the last tunnel address, if present. With help from Steffen Klassert. Signed-off-by: Julian Anastasov <ja@ssi.bg> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-07 05:14:15 +08:00
path->ops->confirm_neigh(path, daddr);
}
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
int xfrm_policy_register_afinfo(const struct xfrm_policy_afinfo *afinfo, int family)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
int err = 0;
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
if (WARN_ON(family >= ARRAY_SIZE(xfrm_policy_afinfo)))
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return -EAFNOSUPPORT;
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
spin_lock(&xfrm_policy_afinfo_lock);
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
if (unlikely(xfrm_policy_afinfo[family] != NULL))
xfrm: fix the return code when xfrm_*_register_afinfo failed If xfrm_*_register_afinfo failed since xfrm_*_afinfo[afinfo->family] had the value, return the -EEXIST, not -ENOBUFS Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-23 11:06:53 +08:00
err = -EEXIST;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
else {
struct dst_ops *dst_ops = afinfo->dst_ops;
if (likely(dst_ops->kmem_cachep == NULL))
dst_ops->kmem_cachep = xfrm_dst_cache;
if (likely(dst_ops->check == NULL))
dst_ops->check = xfrm_dst_check;
net: Abstract default ADVMSS behind an accessor. Make all RTAX_ADVMSS metric accesses go through a new helper function, dst_metric_advmss(). Leave the actual default metric as "zero" in the real metric slot, and compute the actual default value dynamically via a new dst_ops AF specific callback. For stacked IPSEC routes, we use the advmss of the path which preserves existing behavior. Unlike ipv4/ipv6, DecNET ties the advmss to the mtu and thus updates advmss on pmtu updates. This inconsistency in advmss handling results in more raw metric accesses than I wish we ended up with. Signed-off-by: David S. Miller <davem@davemloft.net>
2010-12-14 04:52:14 +08:00
if (likely(dst_ops->default_advmss == NULL))
dst_ops->default_advmss = xfrm_default_advmss;
net: Rename the dst_opt default_mtu method to mtu We plan to invoke the dst_opt->default_mtu() method unconditioally from dst_mtu(). So rename the method to dst_opt->mtu() to match the name with the new meaning. Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-23 10:12:51 +08:00
if (likely(dst_ops->mtu == NULL))
dst_ops->mtu = xfrm_mtu;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (likely(dst_ops->negative_advice == NULL))
dst_ops->negative_advice = xfrm_negative_advice;
if (likely(dst_ops->link_failure == NULL))
dst_ops->link_failure = xfrm_link_failure;
net: Add ->neigh_lookup() operation to dst_ops In the future dst entries will be neigh-less. In that environment we need to have an easy transition point for current users of dst->neighbour outside of the packet output fast path. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-07-18 15:40:17 +08:00
if (likely(dst_ops->neigh_lookup == NULL))
dst_ops->neigh_lookup = xfrm_neigh_lookup;
net: add confirm_neigh method to dst_ops Add confirm_neigh method to dst_ops and use it from IPv4 and IPv6 to lookup and confirm the neighbour. Its usage via the new helper dst_confirm_neigh() should be restricted to MSG_PROBE users for performance reasons. For XFRM prefer the last tunnel address, if present. With help from Steffen Klassert. Signed-off-by: Julian Anastasov <ja@ssi.bg> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-07 05:14:15 +08:00
if (likely(!dst_ops->confirm_neigh))
dst_ops->confirm_neigh = xfrm_confirm_neigh;
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
rcu_assign_pointer(xfrm_policy_afinfo[family], afinfo);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
spin_unlock(&xfrm_policy_afinfo_lock);
netns xfrm: deal with dst entries in netns GC is non-existent in netns, so after you hit GC threshold, no new dst entries will be created until someone triggers cleanup in init_net. Make xfrm4_dst_ops and xfrm6_dst_ops per-netns. This is not done in a generic way, because it woule waste (AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns. Reorder GC threshold initialization so it'd be done before registering XFRM policies. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-25 14:47:53 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return err;
}
EXPORT_SYMBOL(xfrm_policy_register_afinfo);
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
void xfrm_policy_unregister_afinfo(const struct xfrm_policy_afinfo *afinfo)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
xfrm: policy: xfrm_policy_unregister_afinfo can return void Nothing checks the return value. Also, the errors returned on unregister are impossible (we only support INET and INET6, so no way xfrm_policy_afinfo[afinfo->family] can be anything other than 'afinfo' itself). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:15 +08:00
struct dst_ops *dst_ops = afinfo->dst_ops;
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
int i;
xfrm: policy: xfrm_policy_unregister_afinfo can return void Nothing checks the return value. Also, the errors returned on unregister are impossible (we only support INET and INET6, so no way xfrm_policy_afinfo[afinfo->family] can be anything other than 'afinfo' itself). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:15 +08:00
xfrm: policy: remove family field Only needed it to register the policy backend at init time. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:17 +08:00
for (i = 0; i < ARRAY_SIZE(xfrm_policy_afinfo); i++) {
if (xfrm_policy_afinfo[i] != afinfo)
continue;
RCU_INIT_POINTER(xfrm_policy_afinfo[i], NULL);
break;
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
}
xfrm: policy: xfrm_policy_unregister_afinfo can return void Nothing checks the return value. Also, the errors returned on unregister are impossible (we only support INET and INET6, so no way xfrm_policy_afinfo[afinfo->family] can be anything other than 'afinfo' itself). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:15 +08:00
synchronize_rcu();
xfrm: fix RCU bugs This patch reverts commit 56892261ed1a (xfrm: Use rcu_dereference_bh to deference pointer protected by rcu_read_lock_bh), and fixes bugs introduced in commit 418a99ac6ad ( Replace rwlock on xfrm_policy_afinfo with rcu ) 1) We properly use RCU variant in this file, not a mix of RCU/RCU_BH 2) We must defer some writes after the synchronize_rcu() call or a reader can crash dereferencing NULL pointer. 3) Now we use the xfrm_policy_afinfo_lock spinlock only from process context, we no longer need to block BH in xfrm_policy_register_afinfo() and xfrm_policy_unregister_afinfo() 4) Can use RCU_INIT_POINTER() instead of rcu_assign_pointer() in xfrm_policy_unregister_afinfo() 5) Remove a forward inline declaration (xfrm_policy_put_afinfo()), and also move xfrm_policy_get_afinfo() declaration. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Fan Du <fan.du@windriver.com> Cc: Priyanka Jain <Priyanka.Jain@freescale.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2012-08-19 18:31:48 +08:00
xfrm: policy: xfrm_policy_unregister_afinfo can return void Nothing checks the return value. Also, the errors returned on unregister are impossible (we only support INET and INET6, so no way xfrm_policy_afinfo[afinfo->family] can be anything other than 'afinfo' itself). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-07 22:00:15 +08:00
dst_ops->kmem_cachep = NULL;
dst_ops->check = NULL;
dst_ops->negative_advice = NULL;
dst_ops->link_failure = NULL;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(xfrm_policy_unregister_afinfo);
[XFRM]: Define packet dropping statistics. This statistics is shown factor dropped by transformation at /proc/net/xfrm_stat for developer. It is a counter designed from current transformation source code and defined as linux private MIB. See Documentation/networking/xfrm_proc.txt for the detail. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:42:57 +08:00
#ifdef CONFIG_XFRM_STATISTICS
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
static int __net_init xfrm_statistics_init(struct net *net)
[XFRM]: Define packet dropping statistics. This statistics is shown factor dropped by transformation at /proc/net/xfrm_stat for developer. It is a counter designed from current transformation source code and defined as linux private MIB. See Documentation/networking/xfrm_proc.txt for the detail. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:42:57 +08:00
{
netns xfrm: /proc/net/xfrm_stat in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:14 +08:00
int rv;
net: clean up snmp stats code commit 8f0ea0fe3a036a47767f9c80e (snmp: reduce percpu needs by 50%) reduced snmp array size to 1, so technically it doesn't have to be an array any more. What's more, after the following commit: commit 933393f58fef9963eac61db8093689544e29a600 Date: Thu Dec 22 11:58:51 2011 -0600 percpu: Remove irqsafe_cpu_xxx variants We simply say that regular this_cpu use must be safe regardless of preemption and interrupt state. That has no material change for x86 and s390 implementations of this_cpu operations. However, arches that do not provide their own implementation for this_cpu operations will now get code generated that disables interrupts instead of preemption. probably no arch wants to have SNMP_ARRAY_SZ == 2. At least after almost 3 years, no one complains. So, just convert the array to a single pointer and remove snmp_mib_init() and snmp_mib_free() as well. Cc: Christoph Lameter <cl@linux.com> Cc: Eric Dumazet <eric.dumazet@gmail.com> Cc: David S. Miller <davem@davemloft.net> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-06 06:55:55 +08:00
net->mib.xfrm_statistics = alloc_percpu(struct linux_xfrm_mib);
if (!net->mib.xfrm_statistics)
[XFRM]: Define packet dropping statistics. This statistics is shown factor dropped by transformation at /proc/net/xfrm_stat for developer. It is a counter designed from current transformation source code and defined as linux private MIB. See Documentation/networking/xfrm_proc.txt for the detail. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:42:57 +08:00
return -ENOMEM;
netns xfrm: /proc/net/xfrm_stat in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:14 +08:00
rv = xfrm_proc_init(net);
if (rv < 0)
net: clean up snmp stats code commit 8f0ea0fe3a036a47767f9c80e (snmp: reduce percpu needs by 50%) reduced snmp array size to 1, so technically it doesn't have to be an array any more. What's more, after the following commit: commit 933393f58fef9963eac61db8093689544e29a600 Date: Thu Dec 22 11:58:51 2011 -0600 percpu: Remove irqsafe_cpu_xxx variants We simply say that regular this_cpu use must be safe regardless of preemption and interrupt state. That has no material change for x86 and s390 implementations of this_cpu operations. However, arches that do not provide their own implementation for this_cpu operations will now get code generated that disables interrupts instead of preemption. probably no arch wants to have SNMP_ARRAY_SZ == 2. At least after almost 3 years, no one complains. So, just convert the array to a single pointer and remove snmp_mib_init() and snmp_mib_free() as well. Cc: Christoph Lameter <cl@linux.com> Cc: Eric Dumazet <eric.dumazet@gmail.com> Cc: David S. Miller <davem@davemloft.net> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-06 06:55:55 +08:00
free_percpu(net->mib.xfrm_statistics);
netns xfrm: /proc/net/xfrm_stat in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:14 +08:00
return rv;
[XFRM]: Define packet dropping statistics. This statistics is shown factor dropped by transformation at /proc/net/xfrm_stat for developer. It is a counter designed from current transformation source code and defined as linux private MIB. See Documentation/networking/xfrm_proc.txt for the detail. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:42:57 +08:00
}
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
static void xfrm_statistics_fini(struct net *net)
{
netns xfrm: /proc/net/xfrm_stat in netns Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:14 +08:00
xfrm_proc_fini(net);
net: clean up snmp stats code commit 8f0ea0fe3a036a47767f9c80e (snmp: reduce percpu needs by 50%) reduced snmp array size to 1, so technically it doesn't have to be an array any more. What's more, after the following commit: commit 933393f58fef9963eac61db8093689544e29a600 Date: Thu Dec 22 11:58:51 2011 -0600 percpu: Remove irqsafe_cpu_xxx variants We simply say that regular this_cpu use must be safe regardless of preemption and interrupt state. That has no material change for x86 and s390 implementations of this_cpu operations. However, arches that do not provide their own implementation for this_cpu operations will now get code generated that disables interrupts instead of preemption. probably no arch wants to have SNMP_ARRAY_SZ == 2. At least after almost 3 years, no one complains. So, just convert the array to a single pointer and remove snmp_mib_init() and snmp_mib_free() as well. Cc: Christoph Lameter <cl@linux.com> Cc: Eric Dumazet <eric.dumazet@gmail.com> Cc: David S. Miller <davem@davemloft.net> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-06 06:55:55 +08:00
free_percpu(net->mib.xfrm_statistics);
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
}
#else
static int __net_init xfrm_statistics_init(struct net *net)
{
return 0;
}
static void xfrm_statistics_fini(struct net *net)
{
}
[XFRM]: Define packet dropping statistics. This statistics is shown factor dropped by transformation at /proc/net/xfrm_stat for developer. It is a counter designed from current transformation source code and defined as linux private MIB. See Documentation/networking/xfrm_proc.txt for the detail. Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:42:57 +08:00
#endif
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
static int __net_init xfrm_policy_init(struct net *net)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
unsigned int hmask, sz;
int dir;
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
if (net_eq(net, &init_net))
xfrm_dst_cache = kmem_cache_create("xfrm_dst_cache",
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
sizeof(struct xfrm_dst),
[NET]: Use SLAB_PANIC Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-27 10:25:52 +08:00
0, SLAB_HWCACHE_ALIGN|SLAB_PANIC,
mm: Remove slab destructors from kmem_cache_create(). Slab destructors were no longer supported after Christoph's c59def9f222d44bb7e2f0a559f2906191a0862d7 change. They've been BUGs for both slab and slub, and slob never supported them either. This rips out support for the dtor pointer from kmem_cache_create() completely and fixes up every single callsite in the kernel (there were about 224, not including the slab allocator definitions themselves, or the documentation references). Signed-off-by: Paul Mundt <lethal@linux-sh.org>
2007-07-20 09:11:58 +08:00
NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
hmask = 8 - 1;
sz = (hmask+1) * sizeof(struct hlist_head);
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
net->xfrm.policy_byidx = xfrm_hash_alloc(sz);
if (!net->xfrm.policy_byidx)
goto out_byidx;
netns xfrm: per-netns xfrm_policy_byidx hashmask Per-netns hashes are independently resizeable. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:58 +08:00
net->xfrm.policy_idx_hmask = hmask;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
for (dir = 0; dir < XFRM_POLICY_MAX; dir++) {
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
struct xfrm_policy_hash *htab;
netns xfrm: per-netns policy counts Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:24:15 +08:00
net->xfrm.policy_count[dir] = 0;
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
net->xfrm.policy_count[XFRM_POLICY_MAX + dir] = 0;
netns xfrm: per-netns inexact policies Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:26 +08:00
INIT_HLIST_HEAD(&net->xfrm.policy_inexact[dir]);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
netns xfrm: per-netns xfrm_policy_bydst hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:48 +08:00
htab = &net->xfrm.policy_bydst[dir];
[XFRM]: Extract common hashing code into xfrm_hash.[ch] Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:50:50 +08:00
htab->table = xfrm_hash_alloc(sz);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
if (!htab->table)
netns xfrm: per-netns xfrm_policy_bydst hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:48 +08:00
goto out_bydst;
htab->hmask = hmask;
xfrm: hash prefixed policies based on preflen thresholds The idea is an extension of the current policy hashing. Today only non-prefixed policies are stored in a hash table. This patch relaxes the constraints, and hashes policies whose prefix lengths are greater or equal to a configurable threshold. Each hash table (one per direction) maintains its own set of IPv4 and IPv6 thresholds (dbits4, sbits4, dbits6, sbits6), by default (32, 32, 128, 128). Example, if the output hash table is configured with values (16, 24, 56, 64): ip xfrm policy add dir out src 10.22.0.0/20 dst 10.24.1.0/24 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.1.1/32 ... => hashed ip xfrm policy add dir out src 10.22.0.0/16 dst 10.24.0.0/16 ... => unhashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/60 dst 3ffe:304:124:2401::/64 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2401::2/128 ... => hashed ip xfrm policy add dir out \ src 3ffe:304:124:2200::/56 dst 3ffe:304:124:2400::/56 ... => unhashed The high order bits of the addresses (up to the threshold) are used to compute the hash key. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:04 +08:00
htab->dbits4 = 32;
htab->sbits4 = 32;
htab->dbits6 = 128;
htab->sbits6 = 128;
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
}
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
net->xfrm.policy_hthresh.lbits4 = 32;
net->xfrm.policy_hthresh.rbits4 = 32;
net->xfrm.policy_hthresh.lbits6 = 128;
net->xfrm.policy_hthresh.rbits6 = 128;
seqlock_init(&net->xfrm.policy_hthresh.lock);
[XFRM]: Hash policies when non-prefixed. This idea is from Alexey Kuznetsov. It is common for policies to be non-prefixed. And for that case we can optimize lookups, insert, etc. quite a bit. For each direction, we have a dynamically sized policy hash table for non-prefixed policies. We also have a hash table on policy->index. For prefixed policies, we have a list per-direction which we will consult on lookups when a non-prefix hashtable lookup fails. This still isn't as efficient as I would like it. There are four immediate problems: 1) Lots of excessive refcounting, which can be fixed just like xfrm_state was 2) We do 2 hash probes on insert, one to look for dups and one to allocate a unique policy->index. Althought I wonder how much this matters since xfrm_state inserts do up to 3 hash probes and that seems to perform fine. 3) xfrm_policy_insert() is very complex because of the priority ordering and entry replacement logic. 4) Lots of counter bumping, in addition to policy refcounts, in the form of xfrm_policy_count[]. This is merely used to let code path(s) know that some IPSEC rules exist. So this count is indexed per-direction, maybe that is overkill. Signed-off-by: David S. Miller <davem@davemloft.net>
2006-08-24 19:45:07 +08:00
netns xfrm: per-netns policy list Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:11 +08:00
INIT_LIST_HEAD(&net->xfrm.policy_all);
netns xfrm: per-netns policy hash resizing work Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:28:57 +08:00
INIT_WORK(&net->xfrm.policy_hash_work, xfrm_hash_resize);
xfrm: configure policy hash table thresholds by netlink Enable to specify local and remote prefix length thresholds for the policy hash table via a netlink XFRM_MSG_NEWSPDINFO message. prefix length thresholds are specified by XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH optional attributes (struct xfrmu_spdhthresh). example: struct xfrmu_spdhthresh thresh4 = { .lbits = 0; .rbits = 24; }; struct xfrmu_spdhthresh thresh6 = { .lbits = 0; .rbits = 56; }; struct nlmsghdr *hdr; struct nl_msg *msg; msg = nlmsg_alloc(); hdr = nlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, XFRMA_SPD_IPV4_HTHRESH, sizeof(__u32), NLM_F_REQUEST); nla_put(msg, XFRMA_SPD_IPV4_HTHRESH, sizeof(thresh4), &thresh4); nla_put(msg, XFRMA_SPD_IPV6_HTHRESH, sizeof(thresh6), &thresh6); nla_send_auto(sk, msg); The numbers are the policy selector minimum prefix lengths to put a policy in the hash table. - lbits is the local threshold (source address for out policies, destination address for in and fwd policies). - rbits is the remote threshold (destination address for out policies, source address for in and fwd policies). The default values are: XFRMA_SPD_IPV4_HTHRESH: 32 32 XFRMA_SPD_IPV6_HTHRESH: 128 128 Dynamic re-building of the SPD is performed when the thresholds values are changed. The current thresholds can be read via a XFRM_MSG_GETSPDINFO request: the kernel replies to XFRM_MSG_GETSPDINFO requests by an XFRM_MSG_NEWSPDINFO message, with both attributes XFRMA_SPD_IPV4_HTHRESH and XFRMA_SPD_IPV6_HTHRESH. Signed-off-by: Christophe Gouault <christophe.gouault@6wind.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-08-29 22:16:05 +08:00
INIT_WORK(&net->xfrm.policy_hthresh.work, xfrm_hash_rebuild);
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
return 0;
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
netns xfrm: per-netns xfrm_policy_bydst hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:48 +08:00
out_bydst:
for (dir--; dir >= 0; dir--) {
struct xfrm_policy_hash *htab;
htab = &net->xfrm.policy_bydst[dir];
xfrm_hash_free(htab->table, sz);
}
xfrm_hash_free(net->xfrm.policy_byidx, sz);
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
out_byidx:
return -ENOMEM;
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
}
static void xfrm_policy_fini(struct net *net)
{
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
unsigned int sz;
netns xfrm: per-netns inexact policies Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:26 +08:00
int dir;
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
netns xfrm: flush SA/SPDs on netns stop SA/SPD doesn't pin netns (and it shouldn't), so get rid of them by hand. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:57:44 +08:00
flush_work(&net->xfrm.policy_hash_work);
#ifdef CONFIG_XFRM_SUB_POLICY
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_policy_flush(net, XFRM_POLICY_TYPE_SUB, false);
netns xfrm: flush SA/SPDs on netns stop SA/SPD doesn't pin netns (and it shouldn't), so get rid of them by hand. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:57:44 +08:00
#endif
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_policy_flush(net, XFRM_POLICY_TYPE_MAIN, false);
netns xfrm: flush SA/SPDs on netns stop SA/SPD doesn't pin netns (and it shouldn't), so get rid of them by hand. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:57:44 +08:00
netns xfrm: per-netns policy list Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:11 +08:00
WARN_ON(!list_empty(&net->xfrm.policy_all));
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
xfrm: Do not hash socket policies Back in 2003 when I added policy expiration, I half-heartedly did a clean-up and renamed xfrm_sk_policy_link/xfrm_sk_policy_unlink to __xfrm_policy_link/__xfrm_policy_unlink, because the latter could be reused for all policies. I never actually got around to using __xfrm_policy_link for non-socket policies. Later on hashing was added to all xfrm policies, including socket policies. In fact, we don't need hashing on socket policies at all since they're always looked up via a linked list. This patch restores xfrm_sk_policy_link/xfrm_sk_policy_unlink as wrappers around __xfrm_policy_link/__xfrm_policy_unlink so that it's obvious we're dealing with socket policies. This patch also removes hashing from __xfrm_policy_link as for now it's only used by socket policies which do not need to be hashed. Ironically this will in fact allow us to use this helper for non-socket policies which I shall do later. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-11-13 17:09:49 +08:00
for (dir = 0; dir < XFRM_POLICY_MAX; dir++) {
netns xfrm: per-netns xfrm_policy_bydst hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:48 +08:00
struct xfrm_policy_hash *htab;
netns xfrm: per-netns inexact policies Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:26 +08:00
WARN_ON(!hlist_empty(&net->xfrm.policy_inexact[dir]));
netns xfrm: per-netns xfrm_policy_bydst hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:48 +08:00
htab = &net->xfrm.policy_bydst[dir];
xfrm: fix freed block size calculation in xfrm_policy_fini() Missing multiplication of block size by sizeof(struct hlist_head) can cause xfrm_hash_free() to be called with wrong second argument so that kfree() is called on a block allocated with vzalloc() or __get_free_pages() or free_pages() is called with wrong order when a namespace with enough policies is removed. Bug introduced by commit a35f6c5d, i.e. versions >= 2.6.29 are affected. Signed-off-by: Michal Kubecek <mkubecek@suse.cz> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-01-18 23:03:48 +08:00
sz = (htab->hmask + 1) * sizeof(struct hlist_head);
netns xfrm: per-netns xfrm_policy_bydst hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:48 +08:00
WARN_ON(!hlist_empty(htab->table));
xfrm_hash_free(htab->table, sz);
netns xfrm: per-netns inexact policies Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:23:26 +08:00
}
netns xfrm: per-netns xfrm_policy_byidx hashmask Per-netns hashes are independently resizeable. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:58 +08:00
sz = (net->xfrm.policy_idx_hmask + 1) * sizeof(struct hlist_head);
netns xfrm: per-netns xfrm_policy_byidx hash Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:22:35 +08:00
WARN_ON(!hlist_empty(net->xfrm.policy_byidx));
xfrm_hash_free(net->xfrm.policy_byidx, sz);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
static int __net_init xfrm_net_init(struct net *net)
{
int rv;
xfrm: policy: init locks early Dmitry reports following splat: INFO: trying to register non-static key. the code is fine but needs lockdep annotation. turning off the locking correctness validator. CPU: 0 PID: 13059 Comm: syz-executor1 Not tainted 4.10.0-rc7-next-20170207 #1 [..] spin_lock_bh include/linux/spinlock.h:304 [inline] xfrm_policy_flush+0x32/0x470 net/xfrm/xfrm_policy.c:963 xfrm_policy_fini+0xbf/0x560 net/xfrm/xfrm_policy.c:3041 xfrm_net_init+0x79f/0x9e0 net/xfrm/xfrm_policy.c:3091 ops_init+0x10a/0x530 net/core/net_namespace.c:115 setup_net+0x2ed/0x690 net/core/net_namespace.c:291 copy_net_ns+0x26c/0x530 net/core/net_namespace.c:396 create_new_namespaces+0x409/0x860 kernel/nsproxy.c:106 unshare_nsproxy_namespaces+0xae/0x1e0 kernel/nsproxy.c:205 SYSC_unshare kernel/fork.c:2281 [inline] Problem is that when we get error during xfrm_net_init we will call xfrm_policy_fini which will acquire xfrm_policy_lock before it was initialized. Just move it around so locks get set up first. Reported-by: Dmitry Vyukov <dvyukov@google.com> Fixes: 283bc9f35bbbcb0e9 ("xfrm: Namespacify xfrm state/policy locks") Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-02-08 18:52:29 +08:00
/* Initialize the per-net locks here */
spin_lock_init(&net->xfrm.xfrm_state_lock);
spin_lock_init(&net->xfrm.xfrm_policy_lock);
mutex_init(&net->xfrm.xfrm_cfg_mutex);
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
rv = xfrm_statistics_init(net);
if (rv < 0)
goto out_statistics;
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
rv = xfrm_state_init(net);
if (rv < 0)
goto out_state;
rv = xfrm_policy_init(net);
if (rv < 0)
goto out_policy;
netns xfrm: per-netns sysctls Make net.core.xfrm_aevent_etime net.core.xfrm_acq_expires net.core.xfrm_aevent_rseqth net.core.xfrm_larval_drop sysctls per-netns. For that make net_core_path[] global, register it to prevent two /proc/net/core antries and change initcall position -- xfrm_init() is called from fs_initcall, so this one should be fs_initcall at least. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:48 +08:00
rv = xfrm_sysctl_init(net);
if (rv < 0)
goto out_sysctl;
xfrm: Namespacify xfrm state/policy locks By semantics, xfrm layer is fully name space aware, so will the locks, e.g. xfrm_state/pocliy_lock. Ensure exclusive access into state/policy link list for different name space with one global lock is not right in terms of semantics aspect at first place, as they are indeed mutually independent with each other, but also more seriously causes scalability problem. One practical scenario is on a Open Network Stack, more than hundreds of lxc tenants acts as routers within one host, a global xfrm_state/policy_lock becomes the bottleneck. But onces those locks are decoupled in a per-namespace fashion, locks contend is just with in specific name space scope, without causing additional SPD/SAD access delay for other name space. Also this patch improve scalability while as without changing original xfrm behavior. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:50 +08:00
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
return 0;
netns xfrm: per-netns sysctls Make net.core.xfrm_aevent_etime net.core.xfrm_acq_expires net.core.xfrm_aevent_rseqth net.core.xfrm_larval_drop sysctls per-netns. For that make net_core_path[] global, register it to prevent two /proc/net/core antries and change initcall position -- xfrm_init() is called from fs_initcall, so this one should be fs_initcall at least. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:48 +08:00
out_sysctl:
xfrm_policy_fini(net);
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
out_policy:
xfrm_state_fini(net);
out_state:
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
xfrm_statistics_fini(net);
out_statistics:
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
return rv;
}
static void __net_exit xfrm_net_exit(struct net *net)
{
netns xfrm: per-netns sysctls Make net.core.xfrm_aevent_etime net.core.xfrm_acq_expires net.core.xfrm_aevent_rseqth net.core.xfrm_larval_drop sysctls per-netns. For that make net_core_path[] global, register it to prevent two /proc/net/core antries and change initcall position -- xfrm_init() is called from fs_initcall, so this one should be fs_initcall at least. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 10:00:48 +08:00
xfrm_sysctl_fini(net);
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
xfrm_policy_fini(net);
xfrm_state_fini(net);
netns xfrm: per-netns MIBs Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:59:52 +08:00
xfrm_statistics_fini(net);
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
}
static struct pernet_operations __net_initdata xfrm_net_ops = {
.init = xfrm_net_init,
.exit = xfrm_net_exit,
};
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void __init xfrm_init(void)
{
xfrm: add xdst pcpu cache retain last used xfrm_dst in a pcpu cache. On next request, reuse this dst if the policies are the same. The cache will not help with strict RR workloads as there is no hit. The cache packet-path part is reasonably small, the notifier part is needed so we do not add long hangs when a device is dismantled but some pcpu xdst still holds a reference, there are also calls to the flush operation when userspace deletes SAs so modules can be removed (there is no hit. We need to run the dst_release on the correct cpu to avoid races with packet path. This is done by adding a work_struct for each cpu and then doing the actual test/release on each affected cpu via schedule_work_on(). Test results using 4 network namespaces and null encryption: ns1 ns2 -> ns3 -> ns4 netperf -> xfrm/null enc -> xfrm/null dec -> netserver what TCP_STREAM UDP_STREAM UDP_RR Flow cache: 14644.61 294.35 327231.64 No flow cache: 14349.81 242.64 202301.72 Pcpu cache: 14629.70 292.21 205595.22 UDP tests used 64byte packets, tests ran for one minute each, value is average over ten iterations. 'Flow cache' is 'net-next', 'No flow cache' is net-next plus this series but without this patch. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-07-17 19:57:27 +08:00
int i;
xfrm_pcpu_work = kmalloc_array(NR_CPUS, sizeof(*xfrm_pcpu_work),
GFP_KERNEL);
BUG_ON(!xfrm_pcpu_work);
for (i = 0; i < NR_CPUS; i++)
INIT_WORK(&xfrm_pcpu_work[i], xfrm_pcpu_work_fn);
netns xfrm: add netns boilerplate Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 09:14:31 +08:00
register_pernet_subsys(&xfrm_net_ops);
xfrm: Register xfrm_dev_notifier in appropriate place Currently, driver registers it from pernet_operations::init method, and this breaks modularity, because initialization of net namespace and netdevice notifiers are orthogonal actions. We don't have per-namespace netdevice notifiers; all of them are global for all devices in all namespaces. Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-03-29 22:03:25 +08:00
xfrm_dev_init();
xfrm: policy: add sequence count to sync with hash resize Once xfrm_policy_lookup_bytype doesn't grab xfrm_policy_lock anymore its possible for a hash resize to occur in parallel. Use sequence counter to block lookup in case a resize is in progress and to also re-lookup in case hash table was altered in the mean time (might cause use to not find the best-match). Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:54 +08:00
seqcount_init(&xfrm_policy_hash_generation);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
xfrm_input_init();
}
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
#ifdef CONFIG_AUDITSYSCALL
[XFRM] xfrm_policy: kill some bloat net/xfrm/xfrm_policy.c: xfrm_audit_policy_delete | -692 xfrm_audit_policy_add | -692 2 functions changed, 1384 bytes removed, diff: -1384 net/xfrm/xfrm_policy.c: xfrm_audit_common_policyinfo | +704 1 function changed, 704 bytes added, diff: +704 net/xfrm/xfrm_policy.o: 3 functions changed, 704 bytes added, 1384 bytes removed, diff: -680 Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-12 19:20:03 +08:00
static void xfrm_audit_common_policyinfo(struct xfrm_policy *xp,
struct audit_buffer *audit_buf)
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
{
[IPSEC]: SPD auditing fix to include the netmask/prefix-length Currently the netmask/prefix-length of an IPsec SPD entry is not included in any of the SPD related audit messages. This can cause a problem when the audit log is examined as the netmask/prefix-length is vital in determining what network traffic is affected by a particular SPD entry. This patch fixes this problem by adding two additional fields, "src_prefixlen" and "dst_prefixlen", to the SPD audit messages to indicate the source and destination netmasks. These new fields are only included in the audit message when the netmask/prefix-length is less than the address length, i.e. the SPD entry applies to a network address and not a host address. Example audit message: type=UNKNOWN[1415] msg=audit(1196105849.752:25): auid=0 \ subj=root:system_r:unconfined_t:s0-s0:c0.c1023 op=SPD-add res=1 \ src=192.168.0.0 src_prefixlen=24 dst=192.168.1.0 dst_prefixlen=24 In addition, this patch also fixes a few other things in the xfrm_audit_common_policyinfo() function. The IPv4 string formatting was converted to use the standard NIPQUAD_FMT constant, the memcpy() was removed from the IPv6 code path and replaced with a typecast (the memcpy() was acting as a slow, implicit typecast anyway), and two local variables were created to make referencing the XFRM security context and selector information cleaner. Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-01 20:27:18 +08:00
struct xfrm_sec_ctx *ctx = xp->security;
struct xfrm_selector *sel = &xp->selector;
if (ctx)
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
audit_log_format(audit_buf, " sec_alg=%u sec_doi=%u sec_obj=%s",
[IPSEC]: SPD auditing fix to include the netmask/prefix-length Currently the netmask/prefix-length of an IPsec SPD entry is not included in any of the SPD related audit messages. This can cause a problem when the audit log is examined as the netmask/prefix-length is vital in determining what network traffic is affected by a particular SPD entry. This patch fixes this problem by adding two additional fields, "src_prefixlen" and "dst_prefixlen", to the SPD audit messages to indicate the source and destination netmasks. These new fields are only included in the audit message when the netmask/prefix-length is less than the address length, i.e. the SPD entry applies to a network address and not a host address. Example audit message: type=UNKNOWN[1415] msg=audit(1196105849.752:25): auid=0 \ subj=root:system_r:unconfined_t:s0-s0:c0.c1023 op=SPD-add res=1 \ src=192.168.0.0 src_prefixlen=24 dst=192.168.1.0 dst_prefixlen=24 In addition, this patch also fixes a few other things in the xfrm_audit_common_policyinfo() function. The IPv4 string formatting was converted to use the standard NIPQUAD_FMT constant, the memcpy() was removed from the IPv6 code path and replaced with a typecast (the memcpy() was acting as a slow, implicit typecast anyway), and two local variables were created to make referencing the XFRM security context and selector information cleaner. Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-01 20:27:18 +08:00
ctx->ctx_alg, ctx->ctx_doi, ctx->ctx_str);
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
xfrm: checkpatch errors with space This patch cleanup some space errors. Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:46 +08:00
switch (sel->family) {
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
case AF_INET:
net: replace NIPQUAD() in net/*/ Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u can be replaced with %pI4 Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-31 15:54:56 +08:00
audit_log_format(audit_buf, " src=%pI4", &sel->saddr.a4);
[IPSEC]: SPD auditing fix to include the netmask/prefix-length Currently the netmask/prefix-length of an IPsec SPD entry is not included in any of the SPD related audit messages. This can cause a problem when the audit log is examined as the netmask/prefix-length is vital in determining what network traffic is affected by a particular SPD entry. This patch fixes this problem by adding two additional fields, "src_prefixlen" and "dst_prefixlen", to the SPD audit messages to indicate the source and destination netmasks. These new fields are only included in the audit message when the netmask/prefix-length is less than the address length, i.e. the SPD entry applies to a network address and not a host address. Example audit message: type=UNKNOWN[1415] msg=audit(1196105849.752:25): auid=0 \ subj=root:system_r:unconfined_t:s0-s0:c0.c1023 op=SPD-add res=1 \ src=192.168.0.0 src_prefixlen=24 dst=192.168.1.0 dst_prefixlen=24 In addition, this patch also fixes a few other things in the xfrm_audit_common_policyinfo() function. The IPv4 string formatting was converted to use the standard NIPQUAD_FMT constant, the memcpy() was removed from the IPv6 code path and replaced with a typecast (the memcpy() was acting as a slow, implicit typecast anyway), and two local variables were created to make referencing the XFRM security context and selector information cleaner. Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-01 20:27:18 +08:00
if (sel->prefixlen_s != 32)
audit_log_format(audit_buf, " src_prefixlen=%d",
sel->prefixlen_s);
net: replace NIPQUAD() in net/*/ Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u can be replaced with %pI4 Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-31 15:54:56 +08:00
audit_log_format(audit_buf, " dst=%pI4", &sel->daddr.a4);
[IPSEC]: SPD auditing fix to include the netmask/prefix-length Currently the netmask/prefix-length of an IPsec SPD entry is not included in any of the SPD related audit messages. This can cause a problem when the audit log is examined as the netmask/prefix-length is vital in determining what network traffic is affected by a particular SPD entry. This patch fixes this problem by adding two additional fields, "src_prefixlen" and "dst_prefixlen", to the SPD audit messages to indicate the source and destination netmasks. These new fields are only included in the audit message when the netmask/prefix-length is less than the address length, i.e. the SPD entry applies to a network address and not a host address. Example audit message: type=UNKNOWN[1415] msg=audit(1196105849.752:25): auid=0 \ subj=root:system_r:unconfined_t:s0-s0:c0.c1023 op=SPD-add res=1 \ src=192.168.0.0 src_prefixlen=24 dst=192.168.1.0 dst_prefixlen=24 In addition, this patch also fixes a few other things in the xfrm_audit_common_policyinfo() function. The IPv4 string formatting was converted to use the standard NIPQUAD_FMT constant, the memcpy() was removed from the IPv6 code path and replaced with a typecast (the memcpy() was acting as a slow, implicit typecast anyway), and two local variables were created to make referencing the XFRM security context and selector information cleaner. Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-01 20:27:18 +08:00
if (sel->prefixlen_d != 32)
audit_log_format(audit_buf, " dst_prefixlen=%d",
sel->prefixlen_d);
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
break;
case AF_INET6:
net: replace %p6 with %pI6 Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-30 03:52:50 +08:00
audit_log_format(audit_buf, " src=%pI6", sel->saddr.a6);
[IPSEC]: SPD auditing fix to include the netmask/prefix-length Currently the netmask/prefix-length of an IPsec SPD entry is not included in any of the SPD related audit messages. This can cause a problem when the audit log is examined as the netmask/prefix-length is vital in determining what network traffic is affected by a particular SPD entry. This patch fixes this problem by adding two additional fields, "src_prefixlen" and "dst_prefixlen", to the SPD audit messages to indicate the source and destination netmasks. These new fields are only included in the audit message when the netmask/prefix-length is less than the address length, i.e. the SPD entry applies to a network address and not a host address. Example audit message: type=UNKNOWN[1415] msg=audit(1196105849.752:25): auid=0 \ subj=root:system_r:unconfined_t:s0-s0:c0.c1023 op=SPD-add res=1 \ src=192.168.0.0 src_prefixlen=24 dst=192.168.1.0 dst_prefixlen=24 In addition, this patch also fixes a few other things in the xfrm_audit_common_policyinfo() function. The IPv4 string formatting was converted to use the standard NIPQUAD_FMT constant, the memcpy() was removed from the IPv6 code path and replaced with a typecast (the memcpy() was acting as a slow, implicit typecast anyway), and two local variables were created to make referencing the XFRM security context and selector information cleaner. Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-01 20:27:18 +08:00
if (sel->prefixlen_s != 128)
audit_log_format(audit_buf, " src_prefixlen=%d",
sel->prefixlen_s);
net: replace %p6 with %pI6 Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-30 03:52:50 +08:00
audit_log_format(audit_buf, " dst=%pI6", sel->daddr.a6);
[IPSEC]: SPD auditing fix to include the netmask/prefix-length Currently the netmask/prefix-length of an IPsec SPD entry is not included in any of the SPD related audit messages. This can cause a problem when the audit log is examined as the netmask/prefix-length is vital in determining what network traffic is affected by a particular SPD entry. This patch fixes this problem by adding two additional fields, "src_prefixlen" and "dst_prefixlen", to the SPD audit messages to indicate the source and destination netmasks. These new fields are only included in the audit message when the netmask/prefix-length is less than the address length, i.e. the SPD entry applies to a network address and not a host address. Example audit message: type=UNKNOWN[1415] msg=audit(1196105849.752:25): auid=0 \ subj=root:system_r:unconfined_t:s0-s0:c0.c1023 op=SPD-add res=1 \ src=192.168.0.0 src_prefixlen=24 dst=192.168.1.0 dst_prefixlen=24 In addition, this patch also fixes a few other things in the xfrm_audit_common_policyinfo() function. The IPv4 string formatting was converted to use the standard NIPQUAD_FMT constant, the memcpy() was removed from the IPv6 code path and replaced with a typecast (the memcpy() was acting as a slow, implicit typecast anyway), and two local variables were created to make referencing the XFRM security context and selector information cleaner. Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-01 20:27:18 +08:00
if (sel->prefixlen_d != 128)
audit_log_format(audit_buf, " dst_prefixlen=%d",
sel->prefixlen_d);
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
break;
}
}
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
void xfrm_audit_policy_add(struct xfrm_policy *xp, int result, bool task_valid)
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
{
struct audit_buffer *audit_buf;
[XFRM]: RFC4303 compliant auditing This patch adds a number of new IPsec audit events to meet the auditing requirements of RFC4303. This includes audit hooks for the following events: * Could not find a valid SA [sections 2.1, 3.4.2] . xfrm_audit_state_notfound() . xfrm_audit_state_notfound_simple() * Sequence number overflow [section 3.3.3] . xfrm_audit_state_replay_overflow() * Replayed packet [section 3.4.3] . xfrm_audit_state_replay() * Integrity check failure [sections 3.4.4.1, 3.4.4.2] . xfrm_audit_state_icvfail() While RFC4304 deals only with ESP most of the changes in this patch apply to IPsec in general, i.e. both AH and ESP. The one case, integrity check failure, where ESP specific code had to be modified the same was done to the AH code for the sake of consistency. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-22 06:58:11 +08:00
audit_buf = xfrm_audit_start("SPD-add");
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
if (audit_buf == NULL)
return;
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_audit_helper_usrinfo(task_valid, audit_buf);
[XFRM]: RFC4303 compliant auditing This patch adds a number of new IPsec audit events to meet the auditing requirements of RFC4303. This includes audit hooks for the following events: * Could not find a valid SA [sections 2.1, 3.4.2] . xfrm_audit_state_notfound() . xfrm_audit_state_notfound_simple() * Sequence number overflow [section 3.3.3] . xfrm_audit_state_replay_overflow() * Replayed packet [section 3.4.3] . xfrm_audit_state_replay() * Integrity check failure [sections 3.4.4.1, 3.4.4.2] . xfrm_audit_state_icvfail() While RFC4304 deals only with ESP most of the changes in this patch apply to IPsec in general, i.e. both AH and ESP. The one case, integrity check failure, where ESP specific code had to be modified the same was done to the AH code for the sake of consistency. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-22 06:58:11 +08:00
audit_log_format(audit_buf, " res=%u", result);
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
xfrm_audit_common_policyinfo(xp, audit_buf);
audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_policy_add);
[XFRM]: Assorted IPsec fixups This patch fixes a number of small but potentially troublesome things in the XFRM/IPsec code: * Use the 'audit_enabled' variable already in include/linux/audit.h Removed the need for extern declarations local to each XFRM audit fuction * Convert 'sid' to 'secid' everywhere we can The 'sid' name is specific to SELinux, 'secid' is the common naming convention used by the kernel when refering to tokenized LSM labels, unfortunately we have to leave 'ctx_sid' in 'struct xfrm_sec_ctx' otherwise we risk breaking userspace * Convert address display to use standard NIP* macros Similar to what was recently done with the SPD audit code, this also also includes the removal of some unnecessary memcpy() calls * Move common code to xfrm_audit_common_stateinfo() Code consolidation from the "less is more" book on software development * Proper spacing around commas in function arguments Minor style tweak since I was already touching the code Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-21 12:49:33 +08:00
void xfrm_audit_policy_delete(struct xfrm_policy *xp, int result,
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
bool task_valid)
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
{
struct audit_buffer *audit_buf;
[XFRM]: RFC4303 compliant auditing This patch adds a number of new IPsec audit events to meet the auditing requirements of RFC4303. This includes audit hooks for the following events: * Could not find a valid SA [sections 2.1, 3.4.2] . xfrm_audit_state_notfound() . xfrm_audit_state_notfound_simple() * Sequence number overflow [section 3.3.3] . xfrm_audit_state_replay_overflow() * Replayed packet [section 3.4.3] . xfrm_audit_state_replay() * Integrity check failure [sections 3.4.4.1, 3.4.4.2] . xfrm_audit_state_icvfail() While RFC4304 deals only with ESP most of the changes in this patch apply to IPsec in general, i.e. both AH and ESP. The one case, integrity check failure, where ESP specific code had to be modified the same was done to the AH code for the sake of consistency. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-22 06:58:11 +08:00
audit_buf = xfrm_audit_start("SPD-delete");
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
if (audit_buf == NULL)
return;
xfrm: Remove useless xfrm_audit struct. Commit f1370cc4 "xfrm: Remove useless secid field from xfrm_audit." changed "struct xfrm_audit" to have either { audit_get_loginuid(current) / audit_get_sessionid(current) } or { INVALID_UID / -1 } pair. This means that we can represent "struct xfrm_audit" as "bool". This patch replaces "struct xfrm_audit" argument with "bool". Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2014-04-22 20:48:30 +08:00
xfrm_audit_helper_usrinfo(task_valid, audit_buf);
[XFRM]: RFC4303 compliant auditing This patch adds a number of new IPsec audit events to meet the auditing requirements of RFC4303. This includes audit hooks for the following events: * Could not find a valid SA [sections 2.1, 3.4.2] . xfrm_audit_state_notfound() . xfrm_audit_state_notfound_simple() * Sequence number overflow [section 3.3.3] . xfrm_audit_state_replay_overflow() * Replayed packet [section 3.4.3] . xfrm_audit_state_replay() * Integrity check failure [sections 3.4.4.1, 3.4.4.2] . xfrm_audit_state_icvfail() While RFC4304 deals only with ESP most of the changes in this patch apply to IPsec in general, i.e. both AH and ESP. The one case, integrity check failure, where ESP specific code had to be modified the same was done to the AH code for the sake of consistency. Signed-off-by: Paul Moore <paul.moore@hp.com> Acked-by: James Morris <jmorris@namei.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-22 06:58:11 +08:00
audit_log_format(audit_buf, " res=%u", result);
[XFRM]: xfrm audit calls This patch modifies the current ipsec audit layer by breaking it up into purpose driven audit calls. So far, the only audit calls made are when add/delete an SA/policy. It had been discussed to give each key manager it's own calls to do this, but I found there to be much redundnacy since they did the exact same things, except for how they got auid and sid, so I combined them. The below audit calls can be made by any key manager. Hopefully, this is ok. Signed-off-by: Joy Latten <latten@austin.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:51:22 +08:00
xfrm_audit_common_policyinfo(xp, audit_buf);
audit_log_end(audit_buf);
}
EXPORT_SYMBOL_GPL(xfrm_audit_policy_delete);
#endif
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
#ifdef CONFIG_XFRM_MIGRATE
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
static bool xfrm_migrate_selector_match(const struct xfrm_selector *sel_cmp,
const struct xfrm_selector *sel_tgt)
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
{
if (sel_cmp->proto == IPSEC_ULPROTO_ANY) {
if (sel_tgt->family == sel_cmp->family &&
xfrm: Convert xfrm_addr_cmp() to boolean xfrm_addr_equal(). All users of xfrm_addr_cmp() use its result as boolean. Introduce xfrm_addr_equal() (which is equal to !xfrm_addr_cmp()) and convert all users. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-01-29 20:48:50 +08:00
xfrm_addr_equal(&sel_tgt->daddr, &sel_cmp->daddr,
sel_cmp->family) &&
xfrm_addr_equal(&sel_tgt->saddr, &sel_cmp->saddr,
sel_cmp->family) &&
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
sel_tgt->prefixlen_d == sel_cmp->prefixlen_d &&
sel_tgt->prefixlen_s == sel_cmp->prefixlen_s) {
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
return true;
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
}
} else {
if (memcmp(sel_tgt, sel_cmp, sizeof(*sel_tgt)) == 0) {
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
return true;
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
}
}
xfrm: Convert several xfrm policy match functions to bool. xfrm_selector_match xfrm_sec_ctx_match __xfrm4_selector_match __xfrm6_selector_match Signed-off-by: David S. Miller <davem@davemloft.net>
2012-05-16 03:04:57 +08:00
return false;
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
}
xfrm: checkpatch errors with foo * bar This patch clean up some checkpatch errors like this: ERROR: "foo * bar" should be "foo *bar" ERROR: "(foo*)" should be "(foo *)" Signed-off-by: Weilong Chen <chenweilong@huawei.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-12-24 09:43:47 +08:00
static struct xfrm_policy *xfrm_migrate_policy_find(const struct xfrm_selector *sel,
u8 dir, u8 type, struct net *net)
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
{
struct xfrm_policy *pol, *ret = NULL;
struct hlist_head *chain;
u32 priority = ~0U;
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_lock_bh(&net->xfrm.xfrm_policy_lock);
xfrm: Using the right namespace to migrate key info because the home agent could surely be run on a different net namespace other than init_net. The original behavior could lead into inconsistent of key info. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:49 +08:00
chain = policy_hash_direct(net, &sel->daddr, &sel->saddr, sel->family, dir);
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol, chain, bydst) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
if (xfrm_migrate_selector_match(sel, &pol->selector) &&
pol->type == type) {
ret = pol;
priority = ret->priority;
break;
}
}
xfrm: Using the right namespace to migrate key info because the home agent could surely be run on a different net namespace other than init_net. The original behavior could lead into inconsistent of key info. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:49 +08:00
chain = &net->xfrm.policy_inexact[dir];
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 09:06:00 +08:00
hlist_for_each_entry(pol, chain, bydst) {
xfrm: optimise to search the inexact policy list The policies are organized into list by priority ascent of policy, so it is unnecessary to continue to loop the policy if the priority of current looped police is larger than or equal priority which is from the policy_bydst list. This allows to match policy with ~0U priority in inexact list too. Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-05-14 11:16:59 +08:00
if ((pol->priority >= priority) && ret)
break;
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
if (xfrm_migrate_selector_match(sel, &pol->selector) &&
xfrm: optimise to search the inexact policy list The policies are organized into list by priority ascent of policy, so it is unnecessary to continue to loop the policy if the priority of current looped police is larger than or equal priority which is from the policy_bydst list. This allows to match policy with ~0U priority in inexact list too. Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-05-14 11:16:59 +08:00
pol->type == type) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
ret = pol;
break;
}
}
xfrm: remove the unnecessary checking before call xfrm_pol_hold xfrm_pol_hold will check its input with NULL Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2015-04-30 17:13:41 +08:00
xfrm_pol_hold(ret);
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
xfrm: policy: convert policy_lock to spinlock After earlier patches conversions all spots acquire the writer lock and we can now convert this to a normal spinlock. Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2016-08-11 21:17:59 +08:00
spin_unlock_bh(&net->xfrm.xfrm_policy_lock);
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
return ret;
}
xfrm: Const'ify pointer args to migrate_tmpl_match and xfrm_migrate_check Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:21:08 +08:00
static int migrate_tmpl_match(const struct xfrm_migrate *m, const struct xfrm_tmpl *t)
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
{
int match = 0;
if (t->mode == m->mode && t->id.proto == m->proto &&
(m->reqid == 0 || t->reqid == m->reqid)) {
switch (t->mode) {
case XFRM_MODE_TUNNEL:
case XFRM_MODE_BEET:
xfrm: Convert xfrm_addr_cmp() to boolean xfrm_addr_equal(). All users of xfrm_addr_cmp() use its result as boolean. Introduce xfrm_addr_equal() (which is equal to !xfrm_addr_cmp()) and convert all users. Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-01-29 20:48:50 +08:00
if (xfrm_addr_equal(&t->id.daddr, &m->old_daddr,
m->old_family) &&
xfrm_addr_equal(&t->saddr, &m->old_saddr,
m->old_family)) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
match = 1;
}
break;
case XFRM_MODE_TRANSPORT:
/* in case of transport mode, template does not store
any IP addresses, hence we just compare mode and
protocol */
match = 1;
break;
default:
break;
}
}
return match;
}
/* update endpoint address(es) of template(s) */
static int xfrm_policy_migrate(struct xfrm_policy *pol,
struct xfrm_migrate *m, int num_migrate)
{
struct xfrm_migrate *mp;
int i, j, n = 0;
write_lock_bh(&pol->lock);
ipsec: Put dumpers on the dump list Herbert Xu came up with the idea and the original patch to make xfrm_state dump list contain also dumpers: As it is we go to extraordinary lengths to ensure that states don't go away while dumpers go to sleep. It's much easier if we just put the dumpers themselves on the list since they can't go away while they're going. I've also changed the order of addition on new states to prevent a never-ending dump. Timo Teräs improved the patch to apply cleanly to latest tree, modified iteration code to be more readable by using a common struct for entries in the list, implemented the same idea for xfrm_policy dumping and moved the af_key specific "last" entry caching to af_key. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-01 22:03:24 +08:00
if (unlikely(pol->walk.dead)) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
/* target policy has been deleted */
write_unlock_bh(&pol->lock);
return -ENOENT;
}
for (i = 0; i < pol->xfrm_nr; i++) {
for (j = 0, mp = m; j < num_migrate; j++, mp++) {
if (!migrate_tmpl_match(mp, &pol->xfrm_vec[i]))
continue;
n++;
[IPSEC]: Add missing BEET checks Currently BEET mode does not reinject the packet back into the stack like tunnel mode does. Since BEET should behave just like tunnel mode this is incorrect. This patch fixes this by introducing a flags field to xfrm_mode that tells the IPsec code whether it should terminate and reinject the packet back into the stack. It then sets the flag for BEET and tunnel mode. I've also added a number of missing BEET checks elsewhere where we check whether a given mode is a tunnel or not. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-18 12:31:50 +08:00
if (pol->xfrm_vec[i].mode != XFRM_MODE_TUNNEL &&
pol->xfrm_vec[i].mode != XFRM_MODE_BEET)
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
continue;
/* update endpoints */
memcpy(&pol->xfrm_vec[i].id.daddr, &mp->new_daddr,
sizeof(pol->xfrm_vec[i].id.daddr));
memcpy(&pol->xfrm_vec[i].saddr, &mp->new_saddr,
sizeof(pol->xfrm_vec[i].saddr));
pol->xfrm_vec[i].encap_family = mp->new_family;
/* flush bundles */
xfrm: cache bundles instead of policies for outgoing flows __xfrm_lookup() is called for each packet transmitted out of system. The xfrm_find_bundle() does a linear search which can kill system performance depending on how many bundles are required per policy. This modifies __xfrm_lookup() to store bundles directly in the flow cache. If we did not get a hit, we just create a new bundle instead of doing slow search. This means that we can now get multiple xfrm_dst's for same flow (on per-cpu basis). Signed-off-by: Timo Teras <timo.teras@iki.fi> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-07 08:30:05 +08:00
atomic_inc(&pol->genid);
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
}
}
write_unlock_bh(&pol->lock);
if (!n)
return -ENODATA;
return 0;
}
xfrm: Const'ify pointer args to migrate_tmpl_match and xfrm_migrate_check Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:21:08 +08:00
static int xfrm_migrate_check(const struct xfrm_migrate *m, int num_migrate)
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
{
int i, j;
if (num_migrate < 1 || num_migrate > XFRM_MAX_DEPTH)
return -EINVAL;
for (i = 0; i < num_migrate; i++) {
if (xfrm_addr_any(&m[i].new_daddr, m[i].new_family) ||
xfrm_addr_any(&m[i].new_saddr, m[i].new_family))
return -EINVAL;
/* check if there is any duplicated entry */
for (j = i + 1; j < num_migrate; j++) {
if (!memcmp(&m[i].old_daddr, &m[j].old_daddr,
sizeof(m[i].old_daddr)) &&
!memcmp(&m[i].old_saddr, &m[j].old_saddr,
sizeof(m[i].old_saddr)) &&
m[i].proto == m[j].proto &&
m[i].mode == m[j].mode &&
m[i].reqid == m[j].reqid &&
m[i].old_family == m[j].old_family)
return -EINVAL;
}
}
return 0;
}
xfrm: Const'ify selector args in xfrm_migrate paths. Signed-off-by: David S. Miller <davem@davemloft.net>
2011-02-24 13:35:06 +08:00
int xfrm_migrate(const struct xfrm_selector *sel, u8 dir, u8 type,
xfrm: MIGRATE enhancements (draft-ebalard-mext-pfkey-enhanced-migrate) Provides implementation of the enhancements of XFRM/PF_KEY MIGRATE mechanism specified in draft-ebalard-mext-pfkey-enhanced-migrate-00. Defines associated PF_KEY SADB_X_EXT_KMADDRESS extension and XFRM/netlink XFRMA_KMADDRESS attribute. Signed-off-by: Arnaud Ebalard <arno@natisbad.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-06 04:33:42 +08:00
struct xfrm_migrate *m, int num_migrate,
xfrm: extend MIGRATE with UDP encapsulation port Add UDP encapsulation port to XFRM_MSG_MIGRATE using an optional netlink attribute XFRMA_ENCAP. The devices that support IKE MOBIKE extension (RFC-4555 Section 3.8) could go to sleep for a few minutes and wake up. When it wake up the NAT mapping could have expired, the device send a MOBIKE UPDATE_SA message to migrate the IPsec SA. The change could be a change UDP encapsulation port, IP address, or both. Reported-by: Paul Wouters <pwouters@redhat.com> Signed-off-by: Antony Antony <antony@phenome.org> Reviewed-by: Richard Guy Briggs <rgb@tricolour.ca> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-06-06 18:12:13 +08:00
struct xfrm_kmaddress *k, struct net *net,
struct xfrm_encap_tmpl *encap)
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
{
int i, err, nx_cur = 0, nx_new = 0;
struct xfrm_policy *pol = NULL;
struct xfrm_state *x, *xc;
struct xfrm_state *x_cur[XFRM_MAX_DEPTH];
struct xfrm_state *x_new[XFRM_MAX_DEPTH];
struct xfrm_migrate *mp;
xfrm: policy: check policy direction value The 'dir' parameter in xfrm_migrate() is a user-controlled byte which is used as an array index. This can lead to an out-of-bound access, kernel lockup and DoS. Add a check for the 'dir' value. This fixes CVE-2017-11600. References: https://bugzilla.redhat.com/show_bug.cgi?id=1474928 Fixes: 80c9abaabf42 ("[XFRM]: Extension for dynamic update of endpoint address(es)") Cc: <stable@vger.kernel.org> # v2.6.21-rc1 Reported-by: "bo Zhang" <zhangbo5891001@gmail.com> Signed-off-by: Vladis Dronov <vdronov@redhat.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-03 01:50:14 +08:00
/* Stage 0 - sanity checks */
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
if ((err = xfrm_migrate_check(m, num_migrate)) < 0)
goto out;
xfrm: policy: check policy direction value The 'dir' parameter in xfrm_migrate() is a user-controlled byte which is used as an array index. This can lead to an out-of-bound access, kernel lockup and DoS. Add a check for the 'dir' value. This fixes CVE-2017-11600. References: https://bugzilla.redhat.com/show_bug.cgi?id=1474928 Fixes: 80c9abaabf42 ("[XFRM]: Extension for dynamic update of endpoint address(es)") Cc: <stable@vger.kernel.org> # v2.6.21-rc1 Reported-by: "bo Zhang" <zhangbo5891001@gmail.com> Signed-off-by: Vladis Dronov <vdronov@redhat.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-08-03 01:50:14 +08:00
if (dir >= XFRM_POLICY_MAX) {
err = -EINVAL;
goto out;
}
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
/* Stage 1 - find policy */
xfrm: Using the right namespace to migrate key info because the home agent could surely be run on a different net namespace other than init_net. The original behavior could lead into inconsistent of key info. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:49 +08:00
if ((pol = xfrm_migrate_policy_find(sel, dir, type, net)) == NULL) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
err = -ENOENT;
goto out;
}
/* Stage 2 - find and update state(s) */
for (i = 0, mp = m; i < num_migrate; i++, mp++) {
xfrm: Namespacify xfrm state/policy locks By semantics, xfrm layer is fully name space aware, so will the locks, e.g. xfrm_state/pocliy_lock. Ensure exclusive access into state/policy link list for different name space with one global lock is not right in terms of semantics aspect at first place, as they are indeed mutually independent with each other, but also more seriously causes scalability problem. One practical scenario is on a Open Network Stack, more than hundreds of lxc tenants acts as routers within one host, a global xfrm_state/policy_lock becomes the bottleneck. But onces those locks are decoupled in a per-namespace fashion, locks contend is just with in specific name space scope, without causing additional SPD/SAD access delay for other name space. Also this patch improve scalability while as without changing original xfrm behavior. Signed-off-by: Fan Du <fan.du@windriver.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2013-11-07 17:47:50 +08:00
if ((x = xfrm_migrate_state_find(mp, net))) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
x_cur[nx_cur] = x;
nx_cur++;
xfrm: extend MIGRATE with UDP encapsulation port Add UDP encapsulation port to XFRM_MSG_MIGRATE using an optional netlink attribute XFRMA_ENCAP. The devices that support IKE MOBIKE extension (RFC-4555 Section 3.8) could go to sleep for a few minutes and wake up. When it wake up the NAT mapping could have expired, the device send a MOBIKE UPDATE_SA message to migrate the IPsec SA. The change could be a change UDP encapsulation port, IP address, or both. Reported-by: Paul Wouters <pwouters@redhat.com> Signed-off-by: Antony Antony <antony@phenome.org> Reviewed-by: Richard Guy Briggs <rgb@tricolour.ca> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-06-06 18:12:13 +08:00
xc = xfrm_state_migrate(x, mp, encap);
if (xc) {
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
x_new[nx_new] = xc;
nx_new++;
} else {
err = -ENODATA;
goto restore_state;
}
}
}
/* Stage 3 - update policy */
if ((err = xfrm_policy_migrate(pol, m, num_migrate)) < 0)
goto restore_state;
/* Stage 4 - delete old state(s) */
if (nx_cur) {
xfrm_states_put(x_cur, nx_cur);
xfrm_states_delete(x_cur, nx_cur);
}
/* Stage 5 - announce */
xfrm: add UDP encapsulation port in migrate message Add XFRMA_ENCAP, UDP encapsulation port, to km_migrate announcement to userland. Only add if XFRMA_ENCAP was in user migrate request. Signed-off-by: Antony Antony <antony@phenome.org> Reviewed-by: Richard Guy Briggs <rgb@tricolour.ca> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
2017-06-06 18:12:14 +08:00
km_migrate(sel, dir, type, m, num_migrate, k, encap);
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
xfrm_pol_put(pol);
return 0;
out:
return err;
restore_state:
if (pol)
xfrm_pol_put(pol);
if (nx_cur)
xfrm_states_put(x_cur, nx_cur);
if (nx_new)
xfrm_states_delete(x_new, nx_new);
return err;
}
[XFRM]: xfrm_migrate() needs exporting to modules. Needed by xfrm_user and af_key. Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:29:15 +08:00
EXPORT_SYMBOL(xfrm_migrate);
[XFRM]: Extension for dynamic update of endpoint address(es) Extend the XFRM framework so that endpoint address(es) in the XFRM databases could be dynamically updated according to a request (MIGRATE message) from user application. Target XFRM policy is first identified by the selector in the MIGRATE message. Next, the endpoint addresses of the matching templates and XFRM states are updated according to the MIGRATE message. Signed-off-by: Shinta Sugimoto <shinta.sugimoto@ericsson.com> Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-02-09 05:11:42 +08:00
#endif