The SMC protocol [1] uses a rendezvous protocol to negotiate SMC
capability between peers. The current Linux implementation does not yet
use this rendezvous protocol and, thus, is not compliant to RFC7609 and
incompatible with other SMC implementations like in zOS.
This patch adds support for the SMC rendezvous protocol. It uses a new
TCP experimental option. With this option, SMC capabilities are
exchanged between the peers during the TCP three way handshake.
[1] SMC-R Informational RFC: http://www.rfc-editor.org/info/rfc7609
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Link group creation is synchronized with the smc_create_lgr_pending
lock. In smc_listen_work() this mutex is sometimes unlocked, even
though it has not been locked before. This issue will surface in
presence of the SMC rendezvous code.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Notice that in this particular case I placed the "fall through" comment
on its own line, which is what GCC is expecting to find.
Signed-off-by: Gustavo A. R. Silva <garsilva@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For RoCEs ib_query_gid() takes a reference count on the net_device.
This reference count must be decreased by the caller.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reported-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Fixes: 0cfdd8f92c ("smc: connection and link group creation")
Signed-off-by: David S. Miller <davem@davemloft.net>
SMC should not open code the function pointer get_netdev of the
IB device. Replacing ib_query_gid(..., NULL) with
ib_query_gid(..., gid_attr) allows access to the netdev.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Suggested-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use the smc_connection as first parameter with smc_cdc_get_free_slot().
This is just a small code cleanup, no functional change.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Usually socket closing is delayed if there is still data available in
the send buffer to be transmitted. If a process is killed, the delay
should be avoided.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The number of outstanding work requests is limited. If all work
requests are in use, tx processing is postponed to another scheduling
of the tx worker. Switch to a delayed worker to have a gap for tx
completion queue events before the next retry.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
An out-of-sync condition can just be detected by the client.
If the server receives a CLC DECLINE message indicating an out-of-sync
condition for the link groups, the server must clean up the out-of-sync
link group.
There is no need for an extra third parameter in smc_clc_send_decline().
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Client link group creation always follows the server linkgroup creation.
If peer creates a new server link group, client has to create a new
client link group. If peer reuses a server link group for a new
connection, client has to reuse its client link group as well. This
patch introduces a longer delay for client link group removal to make
sure this link group still exists, once the peer decides to reuse a
server link group. This avoids out-of-sync conditions for link groups.
If already scheduled, modify the delay.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The solicited flag is meaningful for the receive completion queue.
Ask for next work completion of any type on the send queue.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
smc_pnet_fill_entry() uses dev_get_by_name() adding a refcount to ndev.
The following smc_pnet_enter() has to reduce the refcount if the entry
to be added exists already in the pnet table.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
smc_netinfo_by_tcpsk() looks up the routing cache. Such a lookup requires
protection by an RCU read lock.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The SMC receive function currently lacks a timeout check under the
condition that no data were received and no data are available. This
patch adds such a check.
Signed-off-by: Hans Wippel <hwippel@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the infiniband part, SMC currently uses get_netdev which calls
dev_hold on the returned net device. However, the SMC code never calls
dev_put on that net device resulting in a wrong reference count.
This patch adds a dev_put after the usage of the net device to fix the
issue.
Signed-off-by: Hans Wippel <hwippel@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Usage of send buffer "sndbuf" is synced
(a) before filling sndbuf for cpu access
(b) after filling sndbuf for device access
Usage of receive buffer "RMB" is synced
(a) before reading RMB content for cpu access
(b) after reading RMB content for device access
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Split function __smc_buf_create() for better readability.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Creation and deletion of SMC receive and send buffers shares a high
amount of common code . This patch introduces common functions to get
rid of duplicate code.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
SMC send buffers are processed the same way as RMBs. Since RMBs have
been converted to sg-logic, do the same for send buffers.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now separate memory regions are created and registered for separate
RMBs. The unsafe_global_rkey of the protection domain is no longer
used. Thus the exposing memory warning can be removed.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A memory region created for a new RMB must be registered explicitly,
before the peer can make use of it for remote DMA transfer.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
SMC currently uses the unsafe_global_rkey of the protection domain,
which exposes all memory for remote reads and writes once a connection
is established. This patch introduces separate memory regions with
separate rkeys for every RMB. Now the unsafe_global_rkey of the
protection domain is no longer needed.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The follow-on patch makes use of ib_map_mr_sg() when introducing
separate memory regions for RMBs. This function is based on
scatterlists; thus this patch introduces scatterlists for RMBs.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Initiate the coming rework of SMC buffer handling with this
small code cleanup. No functional changes here.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If a link group for a new server connection exists already, the mutex
serializing the determination of link groups is given up early.
The coming registration of memory regions benefits from the serialization
as well, if the mutex is held till connection creation is finished.
This patch postpones the unlocking of the link group creation mutex.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver explicitly bypasses APIs to register all memory once a
connection is made, and thus allows remote access to memory.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Acked-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, SMC enables remote access to physical memory when a user
has successfully configured and established an SMC-connection until ten
minutes after the last SMC connection is closed. Because this is considered
a security risk, drivers are supposed to use IB_PD_UNSAFE_GLOBAL_RKEY in
such a case.
This patch changes the current SMC code to use IB_PD_UNSAFE_GLOBAL_RKEY.
This improves user awareness, but does not remove the security risk itself.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull RCU updates from Ingo Molnar:
"The main changes are:
- Debloat RCU headers
- Parallelize SRCU callback handling (plus overlapping patches)
- Improve the performance of Tree SRCU on a CPU-hotplug stress test
- Documentation updates
- Miscellaneous fixes"
* 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (74 commits)
rcu: Open-code the rcu_cblist_n_lazy_cbs() function
rcu: Open-code the rcu_cblist_n_cbs() function
rcu: Open-code the rcu_cblist_empty() function
rcu: Separately compile large rcu_segcblist functions
srcu: Debloat the <linux/rcu_segcblist.h> header
srcu: Adjust default auto-expediting holdoff
srcu: Specify auto-expedite holdoff time
srcu: Expedite first synchronize_srcu() when idle
srcu: Expedited grace periods with reduced memory contention
srcu: Make rcutorture writer stalls print SRCU GP state
srcu: Exact tracking of srcu_data structures containing callbacks
srcu: Make SRCU be built by default
srcu: Fix Kconfig botch when SRCU not selected
rcu: Make non-preemptive schedule be Tasks RCU quiescent state
srcu: Expedite srcu_schedule_cbs_snp() callback invocation
srcu: Parallelize callback handling
kvm: Move srcu_struct fields to end of struct kvm
rcu: Fix typo in PER_RCU_NODE_PERIOD header comment
rcu: Use true/false in assignment to bool
rcu: Use bool value directly
...
rdma_ah_attr can now be either ib or roce allowing
core components to use one type or the other and also
to define attributes unique to a specific type. struct
ib_ah is also initialized with the type when its first
created. This ensures that calls such as modify_ah
dont modify the type of the address handle attribute.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Don Hiatt <don.hiatt@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Modify core and driver components to use accessor functions
introduced to access individual fields of rdma_ah_attr
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Don Hiatt <don.hiatt@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
A group of Linux kernel hackers reported chasing a bug that resulted
from their assumption that SLAB_DESTROY_BY_RCU provided an existence
guarantee, that is, that no block from such a slab would be reallocated
during an RCU read-side critical section. Of course, that is not the
case. Instead, SLAB_DESTROY_BY_RCU only prevents freeing of an entire
slab of blocks.
However, there is a phrase for this, namely "type safety". This commit
therefore renames SLAB_DESTROY_BY_RCU to SLAB_TYPESAFE_BY_RCU in order
to avoid future instances of this sort of confusion.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: <linux-mm@kvack.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
[ paulmck: Add comments mentioning the old name, as requested by Eric
Dumazet, in order to help people familiar with the old name find
the new one. ]
Acked-by: David Rientjes <rientjes@google.com>
smc specifies IB_SEND_INLINE for IB_WR_SEND ib_post_send calls, but
provides a mapped buffer to be sent. This is inconsistent, since
IB_SEND_INLINE works without mapped buffer. Problem has not been
detected in the past, because tests had been limited to Connect X3 cards
from Mellanox, whose mlx4 driver just ignored the IB_SEND_INLINE flag.
For now, the IB_SEND_INLINE flag is removed.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure sockets never accepted are removed cleanly.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
unhash is already called in sock_put_work. Remove the second call.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
State SMC_CLOSED should be reached only, if ConnClosed has been sent to
the peer. If ConnClosed is received from the peer, a socket with
shutdown SHUT_WR done, switches errorneously to state SMC_CLOSED, which
means the peer socket is dangling. The local SMC socket is supposed to
switch to state APPFINCLOSEWAIT to make sure smc_close_final() is called
during socket close.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Several state changes occur during SMC socket closing. Currently
state changes triggered locally occur in process context with
lock_sock() taken while state changes triggered by peer occur in
tasklet context with bh_lock_sock() taken. bh_lock_sock() does not
wait till a lock_sock(() task in process context is finished. This
may lead to races in socket state transitions resulting in dangling
SMC-sockets, or it may lead to duplicate SMC socket freeing.
This patch introduces a closing worker to run all state changes under
lock_sock().
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Wake up reading file descriptors for a closing socket as well, otherwise
some socket applications may stall.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If peer indicates write_blocked, the cursor state of the received data
should be send to the peer immediately (in smc_tx_consumer_update()).
Afterwards the write_blocked indicator is cleared.
If there is no free slot for another write request, sending is postponed
to worker smc_tx_work, and the write_blocked indicator is not cleared.
Therefore another clearing check is needed in smc_tx_work().
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
SMC requires an active ib port on the RoCE device.
smc_pnet_find_roce_resource() determines the matching RoCE device port
according to the configured PNET table. Do not return the found
RoCE device port, if it is not flagged active.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The global event handler is created only, if the ib_device has already
been used by at least one link group. It is guaranteed that there exists
the corresponding entry in the smc_ib_devices list. Get rid of this
superfluous check.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch removes an outdated comment.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These files all use functions declared in interrupt.h, but currently rely
on implicit inclusion of this file (via netns/xfrm.h).
That won't work anymore when the flow cache is removed so include that
header where needed.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Lockdep issues a circular dependency warning when AFS issues an operation
through AF_RXRPC from a context in which the VFS/VM holds the mmap_sem.
The theory lockdep comes up with is as follows:
(1) If the pagefault handler decides it needs to read pages from AFS, it
calls AFS with mmap_sem held and AFS begins an AF_RXRPC call, but
creating a call requires the socket lock:
mmap_sem must be taken before sk_lock-AF_RXRPC
(2) afs_open_socket() opens an AF_RXRPC socket and binds it. rxrpc_bind()
binds the underlying UDP socket whilst holding its socket lock.
inet_bind() takes its own socket lock:
sk_lock-AF_RXRPC must be taken before sk_lock-AF_INET
(3) Reading from a TCP socket into a userspace buffer might cause a fault
and thus cause the kernel to take the mmap_sem, but the TCP socket is
locked whilst doing this:
sk_lock-AF_INET must be taken before mmap_sem
However, lockdep's theory is wrong in this instance because it deals only
with lock classes and not individual locks. The AF_INET lock in (2) isn't
really equivalent to the AF_INET lock in (3) as the former deals with a
socket entirely internal to the kernel that never sees userspace. This is
a limitation in the design of lockdep.
Fix the general case by:
(1) Double up all the locking keys used in sockets so that one set are
used if the socket is created by userspace and the other set is used
if the socket is created by the kernel.
(2) Store the kern parameter passed to sk_alloc() in a variable in the
sock struct (sk_kern_sock). This informs sock_lock_init(),
sock_init_data() and sk_clone_lock() as to the lock keys to be used.
Note that the child created by sk_clone_lock() inherits the parent's
kern setting.
(3) Add a 'kern' parameter to ->accept() that is analogous to the one
passed in to ->create() that distinguishes whether kernel_accept() or
sys_accept4() was the caller and can be passed to sk_alloc().
Note that a lot of accept functions merely dequeue an already
allocated socket. I haven't touched these as the new socket already
exists before we get the parameter.
Note also that there are a couple of places where I've made the accepted
socket unconditionally kernel-based:
irda_accept()
rds_rcp_accept_one()
tcp_accept_from_sock()
because they follow a sock_create_kern() and accept off of that.
Whilst creating this, I noticed that lustre and ocfs don't create sockets
through sock_create_kern() and thus they aren't marked as for-kernel,
though they appear to be internal. I wonder if these should do that so
that they use the new set of lock keys.
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
task_struct::signal and task_struct::sighand are pointers, which would normally make it
straightforward to not define those types in sched.h.
That is not so, because the types are accompanied by a myriad of APIs (macros and inline
functions) that dereference them.
Split the types and the APIs out of sched.h and move them into a new header, <linux/sched/signal.h>.
With this change sched.h does not know about 'struct signal' and 'struct sighand' anymore,
trying to put accessors into sched.h as a test fails the following way:
./include/linux/sched.h: In function ‘test_signal_types’:
./include/linux/sched.h:2461:18: error: dereferencing pointer to incomplete type ‘struct signal_struct’
^
This reduces the size and complexity of sched.h significantly.
Update all headers and .c code that relied on getting the signal handling
functionality from <linux/sched.h> to include <linux/sched/signal.h>.
The list of affected files in the preparatory patch was partly generated by
grepping for the APIs, and partly by doing coverage build testing, both
all[yes|mod|def|no]config builds on 64-bit and 32-bit x86, and an array of
cross-architecture builds.
Nevertheless some (trivial) build breakage is still expected related to rare
Kconfig combinations and in-flight patches to various kernel code, but most
of it should be handled by this patch.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Say we got really unlucky and these failed on the last iteration, then
it could lead to a use after free bug.
Fixes: cd6851f303 ("smc: remote memory buffers (RMBs)")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When creating an SMC connection, there is a CLC (connection layer control)
handshake to prepare for RDMA traffic. The corresponding code is part of
commit 0cfdd8f92c ("smc: connection and link group creation").
Mac addresses to be exchanged in the handshake are copied with a wrong
length of 12 instead of 6 bytes. Following code overwrites the wrongly
copied code, but nevertheless the correct length should already be used for
the preceding mac address copying. Use ETH_ALEN for the memcpy length with
mac addresses.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Fixes: 0cfdd8f92c ("smc: connection and link group creation")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Support for SMC socket monitoring via netlink sockets of protocol
NETLINK_SOCK_DIAG.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
smc_shutdown() and smc_release() handling
delayed linkgroup cleanup for linkgroups without connections
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ursula Braun