OpenCloudOS-Kernel/drivers/infiniband/core/cm.c

4378 lines
122 KiB
C
Raw Normal View History

/*
* Copyright (c) 2004-2007 Intel Corporation. All rights reserved.
* Copyright (c) 2004 Topspin Corporation. All rights reserved.
* Copyright (c) 2004, 2005 Voltaire Corporation. All rights reserved.
* Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
* General Public License (GPL) Version 2, available from the file
* COPYING in the main directory of this source tree, or the
* OpenIB.org BSD license below:
*
* Redistribution and use in source and binary forms, with or
* without modification, are permitted provided that the following
* conditions are met:
*
* - Redistributions of source code must retain the above
* copyright notice, this list of conditions and the following
* disclaimer.
*
* - Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials
* provided with the distribution.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include <linux/completion.h>
#include <linux/dma-mapping.h>
#include <linux/device.h>
#include <linux/module.h>
#include <linux/err.h>
#include <linux/idr.h>
#include <linux/interrupt.h>
#include <linux/random.h>
#include <linux/rbtree.h>
#include <linux/spinlock.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/slab.h>
#include <linux/sysfs.h>
#include <linux/workqueue.h>
#include <linux/kdev_t.h>
IB/core: Ethernet L2 attributes in verbs/cm structures This patch add the support for Ethernet L2 attributes in the verbs/cm/cma structures. When dealing with L2 Ethernet, we should use smac, dmac, vlan ID and priority in a similar manner that the IB L2 (and the L4 PKEY) attributes are used. Thus, those attributes were added to the following structures: * ib_ah_attr - added dmac * ib_qp_attr - added smac and vlan_id, (sl remains vlan priority) * ib_wc - added smac, vlan_id * ib_sa_path_rec - added smac, dmac, vlan_id * cm_av - added smac and vlan_id For the path record structure, extra care was taken to avoid the new fields when packing it into wire format, so we don't break the IB CM and SA wire protocol. On the active side, the CM fills. its internal structures from the path provided by the ULP. We add there taking the ETH L2 attributes and placing them into the CM Address Handle (struct cm_av). On the passive side, the CM fills its internal structures from the WC associated with the REQ message. We add there taking the ETH L2 attributes from the WC. When the HW driver provides the required ETH L2 attributes in the WC, they set the IB_WC_WITH_SMAC and IB_WC_WITH_VLAN flags. The IB core code checks for the presence of these flags, and in their absence does address resolution from the ib_init_ah_from_wc() helper function. ib_modify_qp_is_ok is also updated to consider the link layer. Some parameters are mandatory for Ethernet link layer, while they are irrelevant for IB. Vendor drivers are modified to support the new function signature. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2013-12-13 00:03:11 +08:00
#include <linux/etherdevice.h>
#include <rdma/ib_cache.h>
#include <rdma/ib_cm.h>
#include "cm_msgs.h"
MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("InfiniBand CM");
MODULE_LICENSE("Dual BSD/GPL");
static const char * const ibcm_rej_reason_strs[] = {
[IB_CM_REJ_NO_QP] = "no QP",
[IB_CM_REJ_NO_EEC] = "no EEC",
[IB_CM_REJ_NO_RESOURCES] = "no resources",
[IB_CM_REJ_TIMEOUT] = "timeout",
[IB_CM_REJ_UNSUPPORTED] = "unsupported",
[IB_CM_REJ_INVALID_COMM_ID] = "invalid comm ID",
[IB_CM_REJ_INVALID_COMM_INSTANCE] = "invalid comm instance",
[IB_CM_REJ_INVALID_SERVICE_ID] = "invalid service ID",
[IB_CM_REJ_INVALID_TRANSPORT_TYPE] = "invalid transport type",
[IB_CM_REJ_STALE_CONN] = "stale conn",
[IB_CM_REJ_RDC_NOT_EXIST] = "RDC not exist",
[IB_CM_REJ_INVALID_GID] = "invalid GID",
[IB_CM_REJ_INVALID_LID] = "invalid LID",
[IB_CM_REJ_INVALID_SL] = "invalid SL",
[IB_CM_REJ_INVALID_TRAFFIC_CLASS] = "invalid traffic class",
[IB_CM_REJ_INVALID_HOP_LIMIT] = "invalid hop limit",
[IB_CM_REJ_INVALID_PACKET_RATE] = "invalid packet rate",
[IB_CM_REJ_INVALID_ALT_GID] = "invalid alt GID",
[IB_CM_REJ_INVALID_ALT_LID] = "invalid alt LID",
[IB_CM_REJ_INVALID_ALT_SL] = "invalid alt SL",
[IB_CM_REJ_INVALID_ALT_TRAFFIC_CLASS] = "invalid alt traffic class",
[IB_CM_REJ_INVALID_ALT_HOP_LIMIT] = "invalid alt hop limit",
[IB_CM_REJ_INVALID_ALT_PACKET_RATE] = "invalid alt packet rate",
[IB_CM_REJ_PORT_CM_REDIRECT] = "port CM redirect",
[IB_CM_REJ_PORT_REDIRECT] = "port redirect",
[IB_CM_REJ_INVALID_MTU] = "invalid MTU",
[IB_CM_REJ_INSUFFICIENT_RESP_RESOURCES] = "insufficient resp resources",
[IB_CM_REJ_CONSUMER_DEFINED] = "consumer defined",
[IB_CM_REJ_INVALID_RNR_RETRY] = "invalid RNR retry",
[IB_CM_REJ_DUPLICATE_LOCAL_COMM_ID] = "duplicate local comm ID",
[IB_CM_REJ_INVALID_CLASS_VERSION] = "invalid class version",
[IB_CM_REJ_INVALID_FLOW_LABEL] = "invalid flow label",
[IB_CM_REJ_INVALID_ALT_FLOW_LABEL] = "invalid alt flow label",
};
const char *__attribute_const__ ibcm_reject_msg(int reason)
{
size_t index = reason;
if (index < ARRAY_SIZE(ibcm_rej_reason_strs) &&
ibcm_rej_reason_strs[index])
return ibcm_rej_reason_strs[index];
else
return "unrecognized reason";
}
EXPORT_SYMBOL(ibcm_reject_msg);
static void cm_add_one(struct ib_device *device);
static void cm_remove_one(struct ib_device *device, void *client_data);
static struct ib_client cm_client = {
.name = "cm",
.add = cm_add_one,
.remove = cm_remove_one
};
static struct ib_cm {
spinlock_t lock;
struct list_head device_list;
rwlock_t device_lock;
struct rb_root listen_service_table;
u64 listen_service_id;
/* struct rb_root peer_service_table; todo: fix peer to peer */
struct rb_root remote_qp_table;
struct rb_root remote_id_table;
struct rb_root remote_sidr_table;
struct idr local_id_table;
__be32 random_id_operand;
struct list_head timewait_list;
struct workqueue_struct *wq;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
/* Sync on cm change port state */
spinlock_t state_lock;
} cm;
/* Counter indexes ordered by attribute ID */
enum {
CM_REQ_COUNTER,
CM_MRA_COUNTER,
CM_REJ_COUNTER,
CM_REP_COUNTER,
CM_RTU_COUNTER,
CM_DREQ_COUNTER,
CM_DREP_COUNTER,
CM_SIDR_REQ_COUNTER,
CM_SIDR_REP_COUNTER,
CM_LAP_COUNTER,
CM_APR_COUNTER,
CM_ATTR_COUNT,
CM_ATTR_ID_OFFSET = 0x0010,
};
enum {
CM_XMIT,
CM_XMIT_RETRIES,
CM_RECV,
CM_RECV_DUPLICATES,
CM_COUNTER_GROUPS
};
static char const counter_group_names[CM_COUNTER_GROUPS]
[sizeof("cm_rx_duplicates")] = {
"cm_tx_msgs", "cm_tx_retries",
"cm_rx_msgs", "cm_rx_duplicates"
};
struct cm_counter_group {
struct kobject obj;
atomic_long_t counter[CM_ATTR_COUNT];
};
struct cm_counter_attribute {
struct attribute attr;
int index;
};
#define CM_COUNTER_ATTR(_name, _index) \
struct cm_counter_attribute cm_##_name##_counter_attr = { \
.attr = { .name = __stringify(_name), .mode = 0444 }, \
.index = _index \
}
static CM_COUNTER_ATTR(req, CM_REQ_COUNTER);
static CM_COUNTER_ATTR(mra, CM_MRA_COUNTER);
static CM_COUNTER_ATTR(rej, CM_REJ_COUNTER);
static CM_COUNTER_ATTR(rep, CM_REP_COUNTER);
static CM_COUNTER_ATTR(rtu, CM_RTU_COUNTER);
static CM_COUNTER_ATTR(dreq, CM_DREQ_COUNTER);
static CM_COUNTER_ATTR(drep, CM_DREP_COUNTER);
static CM_COUNTER_ATTR(sidr_req, CM_SIDR_REQ_COUNTER);
static CM_COUNTER_ATTR(sidr_rep, CM_SIDR_REP_COUNTER);
static CM_COUNTER_ATTR(lap, CM_LAP_COUNTER);
static CM_COUNTER_ATTR(apr, CM_APR_COUNTER);
static struct attribute *cm_counter_default_attrs[] = {
&cm_req_counter_attr.attr,
&cm_mra_counter_attr.attr,
&cm_rej_counter_attr.attr,
&cm_rep_counter_attr.attr,
&cm_rtu_counter_attr.attr,
&cm_dreq_counter_attr.attr,
&cm_drep_counter_attr.attr,
&cm_sidr_req_counter_attr.attr,
&cm_sidr_rep_counter_attr.attr,
&cm_lap_counter_attr.attr,
&cm_apr_counter_attr.attr,
NULL
};
struct cm_port {
struct cm_device *cm_dev;
struct ib_mad_agent *mad_agent;
struct kobject port_obj;
u8 port_num;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
struct list_head cm_priv_prim_list;
struct list_head cm_priv_altr_list;
struct cm_counter_group counter_group[CM_COUNTER_GROUPS];
};
struct cm_device {
struct list_head list;
struct ib_device *ib_device;
struct device *device;
u8 ack_delay;
int going_down;
struct cm_port *port[0];
};
struct cm_av {
struct cm_port *port;
union ib_gid dgid;
struct rdma_ah_attr ah_attr;
u16 pkey_index;
u8 timeout;
};
struct cm_work {
struct delayed_work work;
struct list_head list;
struct cm_port *port;
struct ib_mad_recv_wc *mad_recv_wc; /* Received MADs */
__be32 local_id; /* Established / timewait */
__be32 remote_id;
struct ib_cm_event cm_event;
struct sa_path_rec path[0];
};
struct cm_timewait_info {
struct cm_work work; /* Must be first. */
struct list_head list;
struct rb_node remote_qp_node;
struct rb_node remote_id_node;
__be64 remote_ca_guid;
__be32 remote_qpn;
u8 inserted_remote_qp;
u8 inserted_remote_id;
};
struct cm_id_private {
struct ib_cm_id id;
struct rb_node service_node;
struct rb_node sidr_id_node;
spinlock_t lock; /* Do not acquire inside cm.lock */
struct completion comp;
atomic_t refcount;
/* Number of clients sharing this ib_cm_id. Only valid for listeners.
* Protected by the cm.lock spinlock. */
int listen_sharecount;
struct ib_mad_send_buf *msg;
struct cm_timewait_info *timewait_info;
/* todo: use alternate port on send failure */
struct cm_av av;
struct cm_av alt_av;
void *private_data;
__be64 tid;
__be32 local_qpn;
__be32 remote_qpn;
enum ib_qp_type qp_type;
__be32 sq_psn;
__be32 rq_psn;
int timeout_ms;
enum ib_mtu path_mtu;
__be16 pkey;
u8 private_data_len;
u8 max_cm_retries;
u8 peer_to_peer;
u8 responder_resources;
u8 initiator_depth;
u8 retry_count;
u8 rnr_retry_count;
u8 service_timeout;
u8 target_ack_delay;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
struct list_head prim_list;
struct list_head altr_list;
/* Indicates that the send port mad is registered and av is set */
int prim_send_port_not_ready;
int altr_send_port_not_ready;
struct list_head work_list;
atomic_t work_count;
};
static void cm_work_handler(struct work_struct *work);
static inline void cm_deref_id(struct cm_id_private *cm_id_priv)
{
if (atomic_dec_and_test(&cm_id_priv->refcount))
complete(&cm_id_priv->comp);
}
static int cm_alloc_msg(struct cm_id_private *cm_id_priv,
struct ib_mad_send_buf **msg)
{
struct ib_mad_agent *mad_agent;
struct ib_mad_send_buf *m;
struct ib_ah *ah;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
struct cm_av *av;
unsigned long flags, flags2;
int ret = 0;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
/* don't let the port to be released till the agent is down */
spin_lock_irqsave(&cm.state_lock, flags2);
spin_lock_irqsave(&cm.lock, flags);
if (!cm_id_priv->prim_send_port_not_ready)
av = &cm_id_priv->av;
else if (!cm_id_priv->altr_send_port_not_ready &&
(cm_id_priv->alt_av.port))
av = &cm_id_priv->alt_av;
else {
pr_info("%s: not valid CM id\n", __func__);
ret = -ENODEV;
spin_unlock_irqrestore(&cm.lock, flags);
goto out;
}
spin_unlock_irqrestore(&cm.lock, flags);
/* Make sure the port haven't released the mad yet */
mad_agent = cm_id_priv->av.port->mad_agent;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
if (!mad_agent) {
pr_info("%s: not a valid MAD agent\n", __func__);
ret = -ENODEV;
goto out;
}
ah = rdma_create_ah(mad_agent->qp->pd, &av->ah_attr);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
if (IS_ERR(ah)) {
ret = PTR_ERR(ah);
goto out;
}
m = ib_create_send_mad(mad_agent, cm_id_priv->id.remote_cm_qpn,
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
av->pkey_index,
0, IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA,
GFP_ATOMIC,
IB_MGMT_BASE_VERSION);
if (IS_ERR(m)) {
rdma_destroy_ah(ah);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
ret = PTR_ERR(m);
goto out;
}
/* Timeout set by caller if response is expected. */
m->ah = ah;
m->retries = cm_id_priv->max_cm_retries;
atomic_inc(&cm_id_priv->refcount);
m->context[0] = cm_id_priv;
*msg = m;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
out:
spin_unlock_irqrestore(&cm.state_lock, flags2);
return ret;
}
static struct ib_mad_send_buf *cm_alloc_response_msg_no_ah(struct cm_port *port,
struct ib_mad_recv_wc *mad_recv_wc)
{
return ib_create_send_mad(port->mad_agent, 1, mad_recv_wc->wc->pkey_index,
0, IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA,
GFP_ATOMIC,
IB_MGMT_BASE_VERSION);
}
static int cm_create_response_msg_ah(struct cm_port *port,
struct ib_mad_recv_wc *mad_recv_wc,
struct ib_mad_send_buf *msg)
{
struct ib_ah *ah;
ah = ib_create_ah_from_wc(port->mad_agent->qp->pd, mad_recv_wc->wc,
mad_recv_wc->recv_buf.grh, port->port_num);
if (IS_ERR(ah))
return PTR_ERR(ah);
msg->ah = ah;
return 0;
}
static void cm_free_msg(struct ib_mad_send_buf *msg)
{
if (msg->ah)
rdma_destroy_ah(msg->ah);
if (msg->context[0])
cm_deref_id(msg->context[0]);
ib_free_send_mad(msg);
}
static int cm_alloc_response_msg(struct cm_port *port,
struct ib_mad_recv_wc *mad_recv_wc,
struct ib_mad_send_buf **msg)
{
struct ib_mad_send_buf *m;
int ret;
m = cm_alloc_response_msg_no_ah(port, mad_recv_wc);
if (IS_ERR(m))
return PTR_ERR(m);
ret = cm_create_response_msg_ah(port, mad_recv_wc, m);
if (ret) {
cm_free_msg(m);
return ret;
}
*msg = m;
return 0;
}
static void * cm_copy_private_data(const void *private_data,
u8 private_data_len)
{
void *data;
if (!private_data || !private_data_len)
return NULL;
data = kmemdup(private_data, private_data_len, GFP_KERNEL);
if (!data)
return ERR_PTR(-ENOMEM);
return data;
}
static void cm_set_private_data(struct cm_id_private *cm_id_priv,
void *private_data, u8 private_data_len)
{
if (cm_id_priv->private_data && cm_id_priv->private_data_len)
kfree(cm_id_priv->private_data);
cm_id_priv->private_data = private_data;
cm_id_priv->private_data_len = private_data_len;
}
static void cm_init_av_for_response(struct cm_port *port, struct ib_wc *wc,
struct ib_grh *grh, struct cm_av *av)
{
av->port = port;
av->pkey_index = wc->pkey_index;
ib_init_ah_from_wc(port->cm_dev->ib_device, port->port_num, wc,
grh, &av->ah_attr);
}
static int cm_init_av_by_path(struct sa_path_rec *path, struct cm_av *av,
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
struct cm_id_private *cm_id_priv)
{
struct cm_device *cm_dev;
struct cm_port *port = NULL;
unsigned long flags;
int ret;
u8 p;
struct net_device *ndev = ib_get_ndev_from_path(path);
read_lock_irqsave(&cm.device_lock, flags);
list_for_each_entry(cm_dev, &cm.device_list, list) {
if (!ib_find_cached_gid(cm_dev->ib_device, &path->sgid,
sa_conv_pathrec_to_gid_type(path),
ndev, &p, NULL)) {
port = cm_dev->port[p-1];
break;
}
}
read_unlock_irqrestore(&cm.device_lock, flags);
if (ndev)
dev_put(ndev);
if (!port)
return -EINVAL;
ret = ib_find_cached_pkey(cm_dev->ib_device, port->port_num,
be16_to_cpu(path->pkey), &av->pkey_index);
if (ret)
return ret;
av->port = port;
ib_init_ah_from_path(cm_dev->ib_device, port->port_num, path,
&av->ah_attr);
av->timeout = path->packet_life_time + 1;
IB/core: Ethernet L2 attributes in verbs/cm structures This patch add the support for Ethernet L2 attributes in the verbs/cm/cma structures. When dealing with L2 Ethernet, we should use smac, dmac, vlan ID and priority in a similar manner that the IB L2 (and the L4 PKEY) attributes are used. Thus, those attributes were added to the following structures: * ib_ah_attr - added dmac * ib_qp_attr - added smac and vlan_id, (sl remains vlan priority) * ib_wc - added smac, vlan_id * ib_sa_path_rec - added smac, dmac, vlan_id * cm_av - added smac and vlan_id For the path record structure, extra care was taken to avoid the new fields when packing it into wire format, so we don't break the IB CM and SA wire protocol. On the active side, the CM fills. its internal structures from the path provided by the ULP. We add there taking the ETH L2 attributes and placing them into the CM Address Handle (struct cm_av). On the passive side, the CM fills its internal structures from the WC associated with the REQ message. We add there taking the ETH L2 attributes from the WC. When the HW driver provides the required ETH L2 attributes in the WC, they set the IB_WC_WITH_SMAC and IB_WC_WITH_VLAN flags. The IB core code checks for the presence of these flags, and in their absence does address resolution from the ib_init_ah_from_wc() helper function. ib_modify_qp_is_ok is also updated to consider the link layer. Some parameters are mandatory for Ethernet link layer, while they are irrelevant for IB. Vendor drivers are modified to support the new function signature. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2013-12-13 00:03:11 +08:00
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
spin_lock_irqsave(&cm.lock, flags);
if (&cm_id_priv->av == av)
list_add_tail(&cm_id_priv->prim_list, &port->cm_priv_prim_list);
else if (&cm_id_priv->alt_av == av)
list_add_tail(&cm_id_priv->altr_list, &port->cm_priv_altr_list);
else
ret = -EINVAL;
spin_unlock_irqrestore(&cm.lock, flags);
return ret;
}
static int cm_alloc_id(struct cm_id_private *cm_id_priv)
{
unsigned long flags;
int id;
idr_preload(GFP_KERNEL);
spin_lock_irqsave(&cm.lock, flags);
id = idr_alloc_cyclic(&cm.local_id_table, cm_id_priv, 0, 0, GFP_NOWAIT);
spin_unlock_irqrestore(&cm.lock, flags);
idr_preload_end();
cm_id_priv->id.local_id = (__force __be32)id ^ cm.random_id_operand;
return id < 0 ? id : 0;
}
static void cm_free_id(__be32 local_id)
{
spin_lock_irq(&cm.lock);
idr_remove(&cm.local_id_table,
(__force int) (local_id ^ cm.random_id_operand));
spin_unlock_irq(&cm.lock);
}
static struct cm_id_private * cm_get_id(__be32 local_id, __be32 remote_id)
{
struct cm_id_private *cm_id_priv;
cm_id_priv = idr_find(&cm.local_id_table,
(__force int) (local_id ^ cm.random_id_operand));
if (cm_id_priv) {
if (cm_id_priv->id.remote_id == remote_id)
atomic_inc(&cm_id_priv->refcount);
else
cm_id_priv = NULL;
}
return cm_id_priv;
}
static struct cm_id_private * cm_acquire_id(__be32 local_id, __be32 remote_id)
{
struct cm_id_private *cm_id_priv;
spin_lock_irq(&cm.lock);
cm_id_priv = cm_get_id(local_id, remote_id);
spin_unlock_irq(&cm.lock);
return cm_id_priv;
}
/*
* Trivial helpers to strip endian annotation and compare; the
* endianness doesn't actually matter since we just need a stable
* order for the RB tree.
*/
static int be32_lt(__be32 a, __be32 b)
{
return (__force u32) a < (__force u32) b;
}
static int be32_gt(__be32 a, __be32 b)
{
return (__force u32) a > (__force u32) b;
}
static int be64_lt(__be64 a, __be64 b)
{
return (__force u64) a < (__force u64) b;
}
static int be64_gt(__be64 a, __be64 b)
{
return (__force u64) a > (__force u64) b;
}
static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
{
struct rb_node **link = &cm.listen_service_table.rb_node;
struct rb_node *parent = NULL;
struct cm_id_private *cur_cm_id_priv;
__be64 service_id = cm_id_priv->id.service_id;
__be64 service_mask = cm_id_priv->id.service_mask;
while (*link) {
parent = *link;
cur_cm_id_priv = rb_entry(parent, struct cm_id_private,
service_node);
if ((cur_cm_id_priv->id.service_mask & service_id) ==
(service_mask & cur_cm_id_priv->id.service_id) &&
(cm_id_priv->id.device == cur_cm_id_priv->id.device))
return cur_cm_id_priv;
if (cm_id_priv->id.device < cur_cm_id_priv->id.device)
link = &(*link)->rb_left;
else if (cm_id_priv->id.device > cur_cm_id_priv->id.device)
link = &(*link)->rb_right;
else if (be64_lt(service_id, cur_cm_id_priv->id.service_id))
link = &(*link)->rb_left;
else if (be64_gt(service_id, cur_cm_id_priv->id.service_id))
link = &(*link)->rb_right;
else
link = &(*link)->rb_right;
}
rb_link_node(&cm_id_priv->service_node, parent, link);
rb_insert_color(&cm_id_priv->service_node, &cm.listen_service_table);
return NULL;
}
static struct cm_id_private * cm_find_listen(struct ib_device *device,
__be64 service_id)
{
struct rb_node *node = cm.listen_service_table.rb_node;
struct cm_id_private *cm_id_priv;
while (node) {
cm_id_priv = rb_entry(node, struct cm_id_private, service_node);
if ((cm_id_priv->id.service_mask & service_id) ==
cm_id_priv->id.service_id &&
(cm_id_priv->id.device == device))
return cm_id_priv;
if (device < cm_id_priv->id.device)
node = node->rb_left;
else if (device > cm_id_priv->id.device)
node = node->rb_right;
else if (be64_lt(service_id, cm_id_priv->id.service_id))
node = node->rb_left;
else if (be64_gt(service_id, cm_id_priv->id.service_id))
node = node->rb_right;
else
node = node->rb_right;
}
return NULL;
}
static struct cm_timewait_info * cm_insert_remote_id(struct cm_timewait_info
*timewait_info)
{
struct rb_node **link = &cm.remote_id_table.rb_node;
struct rb_node *parent = NULL;
struct cm_timewait_info *cur_timewait_info;
__be64 remote_ca_guid = timewait_info->remote_ca_guid;
__be32 remote_id = timewait_info->work.remote_id;
while (*link) {
parent = *link;
cur_timewait_info = rb_entry(parent, struct cm_timewait_info,
remote_id_node);
if (be32_lt(remote_id, cur_timewait_info->work.remote_id))
link = &(*link)->rb_left;
else if (be32_gt(remote_id, cur_timewait_info->work.remote_id))
link = &(*link)->rb_right;
else if (be64_lt(remote_ca_guid, cur_timewait_info->remote_ca_guid))
link = &(*link)->rb_left;
else if (be64_gt(remote_ca_guid, cur_timewait_info->remote_ca_guid))
link = &(*link)->rb_right;
else
return cur_timewait_info;
}
timewait_info->inserted_remote_id = 1;
rb_link_node(&timewait_info->remote_id_node, parent, link);
rb_insert_color(&timewait_info->remote_id_node, &cm.remote_id_table);
return NULL;
}
static struct cm_timewait_info * cm_find_remote_id(__be64 remote_ca_guid,
__be32 remote_id)
{
struct rb_node *node = cm.remote_id_table.rb_node;
struct cm_timewait_info *timewait_info;
while (node) {
timewait_info = rb_entry(node, struct cm_timewait_info,
remote_id_node);
if (be32_lt(remote_id, timewait_info->work.remote_id))
node = node->rb_left;
else if (be32_gt(remote_id, timewait_info->work.remote_id))
node = node->rb_right;
else if (be64_lt(remote_ca_guid, timewait_info->remote_ca_guid))
node = node->rb_left;
else if (be64_gt(remote_ca_guid, timewait_info->remote_ca_guid))
node = node->rb_right;
else
return timewait_info;
}
return NULL;
}
static struct cm_timewait_info * cm_insert_remote_qpn(struct cm_timewait_info
*timewait_info)
{
struct rb_node **link = &cm.remote_qp_table.rb_node;
struct rb_node *parent = NULL;
struct cm_timewait_info *cur_timewait_info;
__be64 remote_ca_guid = timewait_info->remote_ca_guid;
__be32 remote_qpn = timewait_info->remote_qpn;
while (*link) {
parent = *link;
cur_timewait_info = rb_entry(parent, struct cm_timewait_info,
remote_qp_node);
if (be32_lt(remote_qpn, cur_timewait_info->remote_qpn))
link = &(*link)->rb_left;
else if (be32_gt(remote_qpn, cur_timewait_info->remote_qpn))
link = &(*link)->rb_right;
else if (be64_lt(remote_ca_guid, cur_timewait_info->remote_ca_guid))
link = &(*link)->rb_left;
else if (be64_gt(remote_ca_guid, cur_timewait_info->remote_ca_guid))
link = &(*link)->rb_right;
else
return cur_timewait_info;
}
timewait_info->inserted_remote_qp = 1;
rb_link_node(&timewait_info->remote_qp_node, parent, link);
rb_insert_color(&timewait_info->remote_qp_node, &cm.remote_qp_table);
return NULL;
}
static struct cm_id_private * cm_insert_remote_sidr(struct cm_id_private
*cm_id_priv)
{
struct rb_node **link = &cm.remote_sidr_table.rb_node;
struct rb_node *parent = NULL;
struct cm_id_private *cur_cm_id_priv;
union ib_gid *port_gid = &cm_id_priv->av.dgid;
__be32 remote_id = cm_id_priv->id.remote_id;
while (*link) {
parent = *link;
cur_cm_id_priv = rb_entry(parent, struct cm_id_private,
sidr_id_node);
if (be32_lt(remote_id, cur_cm_id_priv->id.remote_id))
link = &(*link)->rb_left;
else if (be32_gt(remote_id, cur_cm_id_priv->id.remote_id))
link = &(*link)->rb_right;
else {
int cmp;
cmp = memcmp(port_gid, &cur_cm_id_priv->av.dgid,
sizeof *port_gid);
if (cmp < 0)
link = &(*link)->rb_left;
else if (cmp > 0)
link = &(*link)->rb_right;
else
return cur_cm_id_priv;
}
}
rb_link_node(&cm_id_priv->sidr_id_node, parent, link);
rb_insert_color(&cm_id_priv->sidr_id_node, &cm.remote_sidr_table);
return NULL;
}
static void cm_reject_sidr_req(struct cm_id_private *cm_id_priv,
enum ib_cm_sidr_status status)
{
struct ib_cm_sidr_rep_param param;
memset(&param, 0, sizeof param);
param.status = status;
ib_send_cm_sidr_rep(&cm_id_priv->id, &param);
}
struct ib_cm_id *ib_create_cm_id(struct ib_device *device,
ib_cm_handler cm_handler,
void *context)
{
struct cm_id_private *cm_id_priv;
int ret;
cm_id_priv = kzalloc(sizeof *cm_id_priv, GFP_KERNEL);
if (!cm_id_priv)
return ERR_PTR(-ENOMEM);
cm_id_priv->id.state = IB_CM_IDLE;
cm_id_priv->id.device = device;
cm_id_priv->id.cm_handler = cm_handler;
cm_id_priv->id.context = context;
cm_id_priv->id.remote_cm_qpn = 1;
ret = cm_alloc_id(cm_id_priv);
if (ret)
goto error;
spin_lock_init(&cm_id_priv->lock);
init_completion(&cm_id_priv->comp);
INIT_LIST_HEAD(&cm_id_priv->work_list);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
INIT_LIST_HEAD(&cm_id_priv->prim_list);
INIT_LIST_HEAD(&cm_id_priv->altr_list);
atomic_set(&cm_id_priv->work_count, -1);
atomic_set(&cm_id_priv->refcount, 1);
return &cm_id_priv->id;
error:
kfree(cm_id_priv);
return ERR_PTR(-ENOMEM);
}
EXPORT_SYMBOL(ib_create_cm_id);
static struct cm_work * cm_dequeue_work(struct cm_id_private *cm_id_priv)
{
struct cm_work *work;
if (list_empty(&cm_id_priv->work_list))
return NULL;
work = list_entry(cm_id_priv->work_list.next, struct cm_work, list);
list_del(&work->list);
return work;
}
static void cm_free_work(struct cm_work *work)
{
if (work->mad_recv_wc)
ib_free_recv_mad(work->mad_recv_wc);
kfree(work);
}
static inline int cm_convert_to_ms(int iba_time)
{
/* approximate conversion to ms from 4.096us x 2^iba_time */
return 1 << max(iba_time - 8, 0);
}
/*
* calculate: 4.096x2^ack_timeout = 4.096x2^ack_delay + 2x4.096x2^life_time
* Because of how ack_timeout is stored, adding one doubles the timeout.
* To avoid large timeouts, select the max(ack_delay, life_time + 1), and
* increment it (round up) only if the other is within 50%.
*/
static u8 cm_ack_timeout(u8 ca_ack_delay, u8 packet_life_time)
{
int ack_timeout = packet_life_time + 1;
if (ack_timeout >= ca_ack_delay)
ack_timeout += (ca_ack_delay >= (ack_timeout - 1));
else
ack_timeout = ca_ack_delay +
(ack_timeout >= (ca_ack_delay - 1));
return min(31, ack_timeout);
}
static void cm_cleanup_timewait(struct cm_timewait_info *timewait_info)
{
if (timewait_info->inserted_remote_id) {
rb_erase(&timewait_info->remote_id_node, &cm.remote_id_table);
timewait_info->inserted_remote_id = 0;
}
if (timewait_info->inserted_remote_qp) {
rb_erase(&timewait_info->remote_qp_node, &cm.remote_qp_table);
timewait_info->inserted_remote_qp = 0;
}
}
static struct cm_timewait_info * cm_create_timewait_info(__be32 local_id)
{
struct cm_timewait_info *timewait_info;
timewait_info = kzalloc(sizeof *timewait_info, GFP_KERNEL);
if (!timewait_info)
return ERR_PTR(-ENOMEM);
timewait_info->work.local_id = local_id;
INIT_DELAYED_WORK(&timewait_info->work.work, cm_work_handler);
timewait_info->work.cm_event.event = IB_CM_TIMEWAIT_EXIT;
return timewait_info;
}
static void cm_enter_timewait(struct cm_id_private *cm_id_priv)
{
int wait_time;
unsigned long flags;
struct cm_device *cm_dev;
cm_dev = ib_get_client_data(cm_id_priv->id.device, &cm_client);
if (!cm_dev)
return;
spin_lock_irqsave(&cm.lock, flags);
cm_cleanup_timewait(cm_id_priv->timewait_info);
list_add_tail(&cm_id_priv->timewait_info->list, &cm.timewait_list);
spin_unlock_irqrestore(&cm.lock, flags);
/*
* The cm_id could be destroyed by the user before we exit timewait.
* To protect against this, we search for the cm_id after exiting
* timewait before notifying the user that we've exited timewait.
*/
cm_id_priv->id.state = IB_CM_TIMEWAIT;
wait_time = cm_convert_to_ms(cm_id_priv->av.timeout);
/* Check if the device started its remove_one */
IB/cm: Fix a recently introduced deadlock ib_send_cm_drep() calls cm_enter_timewait() while holding a spinlock that can be locked from inside an interrupt handler. Hence do not enable interrupts inside cm_enter_timewait() if called with interrupts disabled. This patch fixes e.g. the following deadlock: Acked-by: Erez Shitrit <erezsh@mellanox.com> ================================= [ INFO: inconsistent lock state ] 4.4.0-rc7+ #1 Tainted: G E --------------------------------- inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage. swapper/8/0 [HC1[1]:SC0[0]:HE0:SE1] takes: (&(&cm_id_priv->lock)->rlock){?.+...}, at: [<ffffffffa036eec4>] cm_establish+0x 74/0x1b0 [ib_cm] {HARDIRQ-ON-W} state was registered at: [<ffffffff810a3c11>] mark_held_locks+0x71/0x90 [<ffffffff810a3e87>] trace_hardirqs_on_caller+0xa7/0x1c0 [<ffffffff810a3fad>] trace_hardirqs_on+0xd/0x10 [<ffffffff8151c40b>] _raw_spin_unlock_irq+0x2b/0x40 [<ffffffffa036ea8e>] cm_enter_timewait+0xae/0x100 [ib_cm] [<ffffffffa036ff76>] ib_send_cm_drep+0xb6/0x190 [ib_cm] [<ffffffffa052ed08>] srp_cm_handler+0x128/0x1a0 [ib_srp] [<ffffffffa0370340>] cm_process_work+0x20/0xf0 [ib_cm] [<ffffffffa0371335>] cm_dreq_handler+0x135/0x2c0 [ib_cm] [<ffffffffa03733c5>] cm_work_handler+0x75/0xd0 [ib_cm] [<ffffffff8107184d>] process_one_work+0x1bd/0x460 [<ffffffff81073148>] worker_thread+0x118/0x420 [<ffffffff81078454>] kthread+0xe4/0x100 [<ffffffff8151cbbf>] ret_from_fork+0x3f/0x70 irq event stamp: 1672286 hardirqs last enabled at (1672283): [<ffffffff81408ec0>] poll_idle+0x10/0x80 hardirqs last disabled at (1672284): [<ffffffff8151d304>] common_interrupt+0x84/0x89 softirqs last enabled at (1672286): [<ffffffff8105b4dc>] _local_bh_enable+0x1c/0x50 softirqs last disabled at (1672285): [<ffffffff8105b697>] irq_enter+0x47/0x70 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(&(&cm_id_priv->lock)->rlock); <Interrupt> lock(&(&cm_id_priv->lock)->rlock); *** DEADLOCK *** no locks held by swapper/8/0. stack backtrace: CPU: 8 PID: 0 Comm: swapper/8 Tainted: G E 4.4.0-rc7+ #1 Hardware name: Dell Inc. PowerEdge R430/03XKDV, BIOS 1.0.2 11/17/2014 ffff88045af5e950 ffff88046e503a88 ffffffff81251c1b 0000000000000007 0000000000000006 0000000000000003 ffff88045af5ddc0 ffff88046e503ad8 ffffffff810a32f4 0000000000000000 0000000000000000 0000000000000001 Call Trace: <IRQ> [<ffffffff81251c1b>] dump_stack+0x4f/0x74 [<ffffffff810a32f4>] print_usage_bug+0x184/0x190 [<ffffffff810a36e2>] mark_lock_irq+0xf2/0x290 [<ffffffff810a3995>] mark_lock+0x115/0x1b0 [<ffffffff810a3b8c>] mark_irqflags+0x15c/0x170 [<ffffffff810a4fef>] __lock_acquire+0x1ef/0x560 [<ffffffff810a53c2>] lock_acquire+0x62/0x80 [<ffffffff8151bd33>] _raw_spin_lock_irqsave+0x43/0x60 [<ffffffffa036eec4>] cm_establish+0x74/0x1b0 [ib_cm] [<ffffffffa036f031>] ib_cm_notify+0x31/0x100 [ib_cm] [<ffffffffa0637f24>] srpt_qp_event+0x54/0xd0 [ib_srpt] [<ffffffffa0196052>] mlx4_ib_qp_event+0x72/0xc0 [mlx4_ib] [<ffffffffa00775b9>] mlx4_qp_event+0x69/0xd0 [mlx4_core] [<ffffffffa006000e>] mlx4_eq_int+0x51e/0xd50 [mlx4_core] [<ffffffffa006084f>] mlx4_msi_x_interrupt+0xf/0x20 [mlx4_core] [<ffffffff810b67b0>] handle_irq_event_percpu+0x40/0x110 [<ffffffff810b68bf>] handle_irq_event+0x3f/0x70 [<ffffffff810ba7f9>] handle_edge_irq+0x79/0x120 [<ffffffff81007f3d>] handle_irq+0x5d/0x130 [<ffffffff810071fd>] do_IRQ+0x6d/0x130 [<ffffffff8151d309>] common_interrupt+0x89/0x89 <EOI> [<ffffffff8140895f>] cpuidle_enter_state+0xcf/0x200 [<ffffffff81408aa2>] cpuidle_enter+0x12/0x20 [<ffffffff810990d6>] call_cpuidle+0x36/0x60 [<ffffffff81099163>] cpuidle_idle_call+0x63/0x110 [<ffffffff8109930a>] cpu_idle_loop+0xfa/0x130 [<ffffffff8109934e>] cpu_startup_entry+0xe/0x10 [<ffffffff8103c443>] start_secondary+0x83/0x90 Fixes: commit be4b499323bf ("IB/cm: Do not queue work to a device that's going away") Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Erez Shitrit <erezsh@mellanox.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-01 20:17:46 +08:00
spin_lock_irqsave(&cm.lock, flags);
if (!cm_dev->going_down)
queue_delayed_work(cm.wq, &cm_id_priv->timewait_info->work.work,
msecs_to_jiffies(wait_time));
IB/cm: Fix a recently introduced deadlock ib_send_cm_drep() calls cm_enter_timewait() while holding a spinlock that can be locked from inside an interrupt handler. Hence do not enable interrupts inside cm_enter_timewait() if called with interrupts disabled. This patch fixes e.g. the following deadlock: Acked-by: Erez Shitrit <erezsh@mellanox.com> ================================= [ INFO: inconsistent lock state ] 4.4.0-rc7+ #1 Tainted: G E --------------------------------- inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage. swapper/8/0 [HC1[1]:SC0[0]:HE0:SE1] takes: (&(&cm_id_priv->lock)->rlock){?.+...}, at: [<ffffffffa036eec4>] cm_establish+0x 74/0x1b0 [ib_cm] {HARDIRQ-ON-W} state was registered at: [<ffffffff810a3c11>] mark_held_locks+0x71/0x90 [<ffffffff810a3e87>] trace_hardirqs_on_caller+0xa7/0x1c0 [<ffffffff810a3fad>] trace_hardirqs_on+0xd/0x10 [<ffffffff8151c40b>] _raw_spin_unlock_irq+0x2b/0x40 [<ffffffffa036ea8e>] cm_enter_timewait+0xae/0x100 [ib_cm] [<ffffffffa036ff76>] ib_send_cm_drep+0xb6/0x190 [ib_cm] [<ffffffffa052ed08>] srp_cm_handler+0x128/0x1a0 [ib_srp] [<ffffffffa0370340>] cm_process_work+0x20/0xf0 [ib_cm] [<ffffffffa0371335>] cm_dreq_handler+0x135/0x2c0 [ib_cm] [<ffffffffa03733c5>] cm_work_handler+0x75/0xd0 [ib_cm] [<ffffffff8107184d>] process_one_work+0x1bd/0x460 [<ffffffff81073148>] worker_thread+0x118/0x420 [<ffffffff81078454>] kthread+0xe4/0x100 [<ffffffff8151cbbf>] ret_from_fork+0x3f/0x70 irq event stamp: 1672286 hardirqs last enabled at (1672283): [<ffffffff81408ec0>] poll_idle+0x10/0x80 hardirqs last disabled at (1672284): [<ffffffff8151d304>] common_interrupt+0x84/0x89 softirqs last enabled at (1672286): [<ffffffff8105b4dc>] _local_bh_enable+0x1c/0x50 softirqs last disabled at (1672285): [<ffffffff8105b697>] irq_enter+0x47/0x70 other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(&(&cm_id_priv->lock)->rlock); <Interrupt> lock(&(&cm_id_priv->lock)->rlock); *** DEADLOCK *** no locks held by swapper/8/0. stack backtrace: CPU: 8 PID: 0 Comm: swapper/8 Tainted: G E 4.4.0-rc7+ #1 Hardware name: Dell Inc. PowerEdge R430/03XKDV, BIOS 1.0.2 11/17/2014 ffff88045af5e950 ffff88046e503a88 ffffffff81251c1b 0000000000000007 0000000000000006 0000000000000003 ffff88045af5ddc0 ffff88046e503ad8 ffffffff810a32f4 0000000000000000 0000000000000000 0000000000000001 Call Trace: <IRQ> [<ffffffff81251c1b>] dump_stack+0x4f/0x74 [<ffffffff810a32f4>] print_usage_bug+0x184/0x190 [<ffffffff810a36e2>] mark_lock_irq+0xf2/0x290 [<ffffffff810a3995>] mark_lock+0x115/0x1b0 [<ffffffff810a3b8c>] mark_irqflags+0x15c/0x170 [<ffffffff810a4fef>] __lock_acquire+0x1ef/0x560 [<ffffffff810a53c2>] lock_acquire+0x62/0x80 [<ffffffff8151bd33>] _raw_spin_lock_irqsave+0x43/0x60 [<ffffffffa036eec4>] cm_establish+0x74/0x1b0 [ib_cm] [<ffffffffa036f031>] ib_cm_notify+0x31/0x100 [ib_cm] [<ffffffffa0637f24>] srpt_qp_event+0x54/0xd0 [ib_srpt] [<ffffffffa0196052>] mlx4_ib_qp_event+0x72/0xc0 [mlx4_ib] [<ffffffffa00775b9>] mlx4_qp_event+0x69/0xd0 [mlx4_core] [<ffffffffa006000e>] mlx4_eq_int+0x51e/0xd50 [mlx4_core] [<ffffffffa006084f>] mlx4_msi_x_interrupt+0xf/0x20 [mlx4_core] [<ffffffff810b67b0>] handle_irq_event_percpu+0x40/0x110 [<ffffffff810b68bf>] handle_irq_event+0x3f/0x70 [<ffffffff810ba7f9>] handle_edge_irq+0x79/0x120 [<ffffffff81007f3d>] handle_irq+0x5d/0x130 [<ffffffff810071fd>] do_IRQ+0x6d/0x130 [<ffffffff8151d309>] common_interrupt+0x89/0x89 <EOI> [<ffffffff8140895f>] cpuidle_enter_state+0xcf/0x200 [<ffffffff81408aa2>] cpuidle_enter+0x12/0x20 [<ffffffff810990d6>] call_cpuidle+0x36/0x60 [<ffffffff81099163>] cpuidle_idle_call+0x63/0x110 [<ffffffff8109930a>] cpu_idle_loop+0xfa/0x130 [<ffffffff8109934e>] cpu_startup_entry+0xe/0x10 [<ffffffff8103c443>] start_secondary+0x83/0x90 Fixes: commit be4b499323bf ("IB/cm: Do not queue work to a device that's going away") Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Erez Shitrit <erezsh@mellanox.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-01 20:17:46 +08:00
spin_unlock_irqrestore(&cm.lock, flags);
cm_id_priv->timewait_info = NULL;
}
static void cm_reset_to_idle(struct cm_id_private *cm_id_priv)
{
unsigned long flags;
cm_id_priv->id.state = IB_CM_IDLE;
if (cm_id_priv->timewait_info) {
spin_lock_irqsave(&cm.lock, flags);
cm_cleanup_timewait(cm_id_priv->timewait_info);
spin_unlock_irqrestore(&cm.lock, flags);
kfree(cm_id_priv->timewait_info);
cm_id_priv->timewait_info = NULL;
}
}
static void cm_destroy_id(struct ib_cm_id *cm_id, int err)
{
struct cm_id_private *cm_id_priv;
struct cm_work *work;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
retest:
spin_lock_irq(&cm_id_priv->lock);
switch (cm_id->state) {
case IB_CM_LISTEN:
spin_unlock_irq(&cm_id_priv->lock);
spin_lock_irq(&cm.lock);
if (--cm_id_priv->listen_sharecount > 0) {
/* The id is still shared. */
cm_deref_id(cm_id_priv);
spin_unlock_irq(&cm.lock);
return;
}
rb_erase(&cm_id_priv->service_node, &cm.listen_service_table);
spin_unlock_irq(&cm.lock);
break;
case IB_CM_SIDR_REQ_SENT:
cm_id->state = IB_CM_IDLE;
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
spin_unlock_irq(&cm_id_priv->lock);
break;
case IB_CM_SIDR_REQ_RCVD:
spin_unlock_irq(&cm_id_priv->lock);
cm_reject_sidr_req(cm_id_priv, IB_SIDR_REJECT);
spin_lock_irq(&cm.lock);
if (!RB_EMPTY_NODE(&cm_id_priv->sidr_id_node))
rb_erase(&cm_id_priv->sidr_id_node,
&cm.remote_sidr_table);
spin_unlock_irq(&cm.lock);
break;
case IB_CM_REQ_SENT:
ib/cm: Change reject message type when destroying cm_id Problem reported by: Ted Kim <ted.h.kim@oracle.com>: We have a case where a Linux system and a non-Linux system are trying to interoperate. The Linux host is the active side and starts the connection establishment, but later decides to not go through with the connection setup and does rdma_destroy_id(). The rdma_destroy_id() eventually works its way down to cm_destroy_id() in core/cm.c, where a REJ is sent. The non-Linux system has some trouble recognizing the REJ because of: A. CM states which can't receive the REJ B. Some issues about REJ formatting (missing comm ID) ISSUE A: That part of the spec says, a Consumer Reject REJ can be sent for a connection abort, but it goes further and says: can send a REJ message with a "Consumer Reject" Reason code if they are in a CM state (i.e. REP Rcvd, MRA(REP) Sent, REQ Rcvd, MRA Sent) that allows a REJ to be sent (lines 35-38). Of the states listed there in that sentence, it would seem to limit the active side to using the Consumer Reject (for the abort case) in just the REP-Rcvd and MRA-REP-Sent states. That is basically only after the active side sees a REP (or alternatively goes down the state transitions to timeout in which case a Timeout REJ is sent). As a fix, in cm-destroy-id() move the IB-CM-MRA-REQ-RCVD case to the same as REQ-SENT. Essentially, make a REJ sent after getting an MRA on active side a timeout rather than Consumer- Reject, which is arguably more correct with the CM state diagrams previous to getting a REP. Signed-off-by: Ted Kim <ted.h.kim@oracle.com> Signed-off-by: Sean Hefty <sean.hefty@intel.com>
2015-05-15 03:49:01 +08:00
case IB_CM_MRA_REQ_RCVD:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
spin_unlock_irq(&cm_id_priv->lock);
ib_send_cm_rej(cm_id, IB_CM_REJ_TIMEOUT,
&cm_id_priv->id.device->node_guid,
sizeof cm_id_priv->id.device->node_guid,
NULL, 0);
break;
case IB_CM_REQ_RCVD:
if (err == -ENOMEM) {
/* Do not reject to allow future retries. */
cm_reset_to_idle(cm_id_priv);
spin_unlock_irq(&cm_id_priv->lock);
} else {
spin_unlock_irq(&cm_id_priv->lock);
ib_send_cm_rej(cm_id, IB_CM_REJ_CONSUMER_DEFINED,
NULL, 0, NULL, 0);
}
break;
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
/* Fall through */
case IB_CM_MRA_REQ_SENT:
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
spin_unlock_irq(&cm_id_priv->lock);
ib_send_cm_rej(cm_id, IB_CM_REJ_CONSUMER_DEFINED,
NULL, 0, NULL, 0);
break;
case IB_CM_ESTABLISHED:
spin_unlock_irq(&cm_id_priv->lock);
if (cm_id_priv->qp_type == IB_QPT_XRC_TGT)
break;
ib_send_cm_dreq(cm_id, NULL, 0);
goto retest;
case IB_CM_DREQ_SENT:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
cm_enter_timewait(cm_id_priv);
spin_unlock_irq(&cm_id_priv->lock);
break;
case IB_CM_DREQ_RCVD:
spin_unlock_irq(&cm_id_priv->lock);
ib_send_cm_drep(cm_id, NULL, 0);
break;
default:
spin_unlock_irq(&cm_id_priv->lock);
break;
}
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
spin_lock_irq(&cm.lock);
if (!list_empty(&cm_id_priv->altr_list) &&
(!cm_id_priv->altr_send_port_not_ready))
list_del(&cm_id_priv->altr_list);
if (!list_empty(&cm_id_priv->prim_list) &&
(!cm_id_priv->prim_send_port_not_ready))
list_del(&cm_id_priv->prim_list);
spin_unlock_irq(&cm.lock);
cm_free_id(cm_id->local_id);
cm_deref_id(cm_id_priv);
wait_for_completion(&cm_id_priv->comp);
while ((work = cm_dequeue_work(cm_id_priv)) != NULL)
cm_free_work(work);
kfree(cm_id_priv->private_data);
kfree(cm_id_priv);
}
void ib_destroy_cm_id(struct ib_cm_id *cm_id)
{
cm_destroy_id(cm_id, 0);
}
EXPORT_SYMBOL(ib_destroy_cm_id);
/**
* __ib_cm_listen - Initiates listening on the specified service ID for
* connection and service ID resolution requests.
* @cm_id: Connection identifier associated with the listen request.
* @service_id: Service identifier matched against incoming connection
* and service ID resolution requests. The service ID should be specified
* network-byte order. If set to IB_CM_ASSIGN_SERVICE_ID, the CM will
* assign a service ID to the caller.
* @service_mask: Mask applied to service ID used to listen across a
* range of service IDs. If set to 0, the service ID is matched
* exactly. This parameter is ignored if %service_id is set to
* IB_CM_ASSIGN_SERVICE_ID.
*/
static int __ib_cm_listen(struct ib_cm_id *cm_id, __be64 service_id,
__be64 service_mask)
{
struct cm_id_private *cm_id_priv, *cur_cm_id_priv;
int ret = 0;
service_mask = service_mask ? service_mask : ~cpu_to_be64(0);
service_id &= service_mask;
if ((service_id & IB_SERVICE_ID_AGN_MASK) == IB_CM_ASSIGN_SERVICE_ID &&
(service_id != IB_CM_ASSIGN_SERVICE_ID))
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
if (cm_id->state != IB_CM_IDLE)
return -EINVAL;
cm_id->state = IB_CM_LISTEN;
++cm_id_priv->listen_sharecount;
if (service_id == IB_CM_ASSIGN_SERVICE_ID) {
cm_id->service_id = cpu_to_be64(cm.listen_service_id++);
cm_id->service_mask = ~cpu_to_be64(0);
} else {
cm_id->service_id = service_id;
cm_id->service_mask = service_mask;
}
cur_cm_id_priv = cm_insert_listen(cm_id_priv);
if (cur_cm_id_priv) {
cm_id->state = IB_CM_IDLE;
--cm_id_priv->listen_sharecount;
ret = -EBUSY;
}
return ret;
}
int ib_cm_listen(struct ib_cm_id *cm_id, __be64 service_id, __be64 service_mask)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&cm.lock, flags);
ret = __ib_cm_listen(cm_id, service_id, service_mask);
spin_unlock_irqrestore(&cm.lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_cm_listen);
/**
* Create a new listening ib_cm_id and listen on the given service ID.
*
* If there's an existing ID listening on that same device and service ID,
* return it.
*
* @device: Device associated with the cm_id. All related communication will
* be associated with the specified device.
* @cm_handler: Callback invoked to notify the user of CM events.
* @service_id: Service identifier matched against incoming connection
* and service ID resolution requests. The service ID should be specified
* network-byte order. If set to IB_CM_ASSIGN_SERVICE_ID, the CM will
* assign a service ID to the caller.
*
* Callers should call ib_destroy_cm_id when done with the listener ID.
*/
struct ib_cm_id *ib_cm_insert_listen(struct ib_device *device,
ib_cm_handler cm_handler,
__be64 service_id)
{
struct cm_id_private *cm_id_priv;
struct ib_cm_id *cm_id;
unsigned long flags;
int err = 0;
/* Create an ID in advance, since the creation may sleep */
cm_id = ib_create_cm_id(device, cm_handler, NULL);
if (IS_ERR(cm_id))
return cm_id;
spin_lock_irqsave(&cm.lock, flags);
if (service_id == IB_CM_ASSIGN_SERVICE_ID)
goto new_id;
/* Find an existing ID */
cm_id_priv = cm_find_listen(device, service_id);
if (cm_id_priv) {
if (cm_id->cm_handler != cm_handler || cm_id->context) {
/* Sharing an ib_cm_id with different handlers is not
* supported */
spin_unlock_irqrestore(&cm.lock, flags);
return ERR_PTR(-EINVAL);
}
atomic_inc(&cm_id_priv->refcount);
++cm_id_priv->listen_sharecount;
spin_unlock_irqrestore(&cm.lock, flags);
ib_destroy_cm_id(cm_id);
cm_id = &cm_id_priv->id;
return cm_id;
}
new_id:
/* Use newly created ID */
err = __ib_cm_listen(cm_id, service_id, 0);
spin_unlock_irqrestore(&cm.lock, flags);
if (err) {
ib_destroy_cm_id(cm_id);
return ERR_PTR(err);
}
return cm_id;
}
EXPORT_SYMBOL(ib_cm_insert_listen);
static __be64 cm_form_tid(struct cm_id_private *cm_id_priv,
enum cm_msg_sequence msg_seq)
{
u64 hi_tid, low_tid;
hi_tid = ((u64) cm_id_priv->av.port->mad_agent->hi_tid) << 32;
low_tid = (u64) ((__force u32)cm_id_priv->id.local_id |
(msg_seq << 30));
return cpu_to_be64(hi_tid | low_tid);
}
static void cm_format_mad_hdr(struct ib_mad_hdr *hdr,
__be16 attr_id, __be64 tid)
{
hdr->base_version = IB_MGMT_BASE_VERSION;
hdr->mgmt_class = IB_MGMT_CLASS_CM;
hdr->class_version = IB_CM_CLASS_VERSION;
hdr->method = IB_MGMT_METHOD_SEND;
hdr->attr_id = attr_id;
hdr->tid = tid;
}
static void cm_format_req(struct cm_req_msg *req_msg,
struct cm_id_private *cm_id_priv,
struct ib_cm_req_param *param)
{
struct sa_path_rec *pri_path = param->primary_path;
struct sa_path_rec *alt_path = param->alternate_path;
bool pri_ext = false;
if (pri_path->rec_type == SA_PATH_REC_TYPE_OPA)
pri_ext = opa_is_extended_lid(pri_path->opa.dlid,
pri_path->opa.slid);
cm_format_mad_hdr(&req_msg->hdr, CM_REQ_ATTR_ID,
cm_form_tid(cm_id_priv, CM_MSG_SEQUENCE_REQ));
req_msg->local_comm_id = cm_id_priv->id.local_id;
req_msg->service_id = param->service_id;
req_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
cm_req_set_local_qpn(req_msg, cpu_to_be32(param->qp_num));
cm_req_set_init_depth(req_msg, param->initiator_depth);
cm_req_set_remote_resp_timeout(req_msg,
param->remote_cm_response_timeout);
cm_req_set_qp_type(req_msg, param->qp_type);
cm_req_set_flow_ctrl(req_msg, param->flow_control);
cm_req_set_starting_psn(req_msg, cpu_to_be32(param->starting_psn));
cm_req_set_local_resp_timeout(req_msg,
param->local_cm_response_timeout);
req_msg->pkey = param->primary_path->pkey;
cm_req_set_path_mtu(req_msg, param->primary_path->mtu);
cm_req_set_max_cm_retries(req_msg, param->max_cm_retries);
if (param->qp_type != IB_QPT_XRC_INI) {
cm_req_set_resp_res(req_msg, param->responder_resources);
cm_req_set_retry_count(req_msg, param->retry_count);
cm_req_set_rnr_retry_count(req_msg, param->rnr_retry_count);
cm_req_set_srq(req_msg, param->srq);
}
req_msg->primary_local_gid = pri_path->sgid;
req_msg->primary_remote_gid = pri_path->dgid;
if (pri_ext) {
req_msg->primary_local_gid.global.interface_id
= OPA_MAKE_ID(be32_to_cpu(pri_path->opa.slid));
req_msg->primary_remote_gid.global.interface_id
= OPA_MAKE_ID(be32_to_cpu(pri_path->opa.dlid));
}
if (pri_path->hop_limit <= 1) {
req_msg->primary_local_lid = pri_ext ? 0 :
htons(ntohl(sa_path_get_slid(pri_path)));
req_msg->primary_remote_lid = pri_ext ? 0 :
htons(ntohl(sa_path_get_dlid(pri_path)));
} else {
/* Work-around until there's a way to obtain remote LID info */
req_msg->primary_local_lid = IB_LID_PERMISSIVE;
req_msg->primary_remote_lid = IB_LID_PERMISSIVE;
}
cm_req_set_primary_flow_label(req_msg, pri_path->flow_label);
cm_req_set_primary_packet_rate(req_msg, pri_path->rate);
req_msg->primary_traffic_class = pri_path->traffic_class;
req_msg->primary_hop_limit = pri_path->hop_limit;
cm_req_set_primary_sl(req_msg, pri_path->sl);
cm_req_set_primary_subnet_local(req_msg, (pri_path->hop_limit <= 1));
cm_req_set_primary_local_ack_timeout(req_msg,
cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay,
pri_path->packet_life_time));
if (alt_path) {
bool alt_ext = false;
if (alt_path->rec_type == SA_PATH_REC_TYPE_OPA)
alt_ext = opa_is_extended_lid(alt_path->opa.dlid,
alt_path->opa.slid);
req_msg->alt_local_gid = alt_path->sgid;
req_msg->alt_remote_gid = alt_path->dgid;
if (alt_ext) {
req_msg->alt_local_gid.global.interface_id
= OPA_MAKE_ID(be32_to_cpu(alt_path->opa.slid));
req_msg->alt_remote_gid.global.interface_id
= OPA_MAKE_ID(be32_to_cpu(alt_path->opa.dlid));
}
if (alt_path->hop_limit <= 1) {
req_msg->alt_local_lid = alt_ext ? 0 :
htons(ntohl(sa_path_get_slid(alt_path)));
req_msg->alt_remote_lid = alt_ext ? 0 :
htons(ntohl(sa_path_get_dlid(alt_path)));
} else {
req_msg->alt_local_lid = IB_LID_PERMISSIVE;
req_msg->alt_remote_lid = IB_LID_PERMISSIVE;
}
cm_req_set_alt_flow_label(req_msg,
alt_path->flow_label);
cm_req_set_alt_packet_rate(req_msg, alt_path->rate);
req_msg->alt_traffic_class = alt_path->traffic_class;
req_msg->alt_hop_limit = alt_path->hop_limit;
cm_req_set_alt_sl(req_msg, alt_path->sl);
cm_req_set_alt_subnet_local(req_msg, (alt_path->hop_limit <= 1));
cm_req_set_alt_local_ack_timeout(req_msg,
cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay,
alt_path->packet_life_time));
}
if (param->private_data && param->private_data_len)
memcpy(req_msg->private_data, param->private_data,
param->private_data_len);
}
static int cm_validate_req_param(struct ib_cm_req_param *param)
{
/* peer-to-peer not supported */
if (param->peer_to_peer)
return -EINVAL;
if (!param->primary_path)
return -EINVAL;
if (param->qp_type != IB_QPT_RC && param->qp_type != IB_QPT_UC &&
param->qp_type != IB_QPT_XRC_INI)
return -EINVAL;
if (param->private_data &&
param->private_data_len > IB_CM_REQ_PRIVATE_DATA_SIZE)
return -EINVAL;
if (param->alternate_path &&
(param->alternate_path->pkey != param->primary_path->pkey ||
param->alternate_path->mtu != param->primary_path->mtu))
return -EINVAL;
return 0;
}
int ib_send_cm_req(struct ib_cm_id *cm_id,
struct ib_cm_req_param *param)
{
struct cm_id_private *cm_id_priv;
struct cm_req_msg *req_msg;
unsigned long flags;
int ret;
ret = cm_validate_req_param(param);
if (ret)
return ret;
/* Verify that we're not in timewait. */
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_IDLE) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
ret = -EINVAL;
goto out;
}
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_id_priv->timewait_info = cm_create_timewait_info(cm_id_priv->
id.local_id);
if (IS_ERR(cm_id_priv->timewait_info)) {
ret = PTR_ERR(cm_id_priv->timewait_info);
goto out;
}
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
ret = cm_init_av_by_path(param->primary_path, &cm_id_priv->av,
cm_id_priv);
if (ret)
goto error1;
if (param->alternate_path) {
ret = cm_init_av_by_path(param->alternate_path,
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
&cm_id_priv->alt_av, cm_id_priv);
if (ret)
goto error1;
}
cm_id->service_id = param->service_id;
cm_id->service_mask = ~cpu_to_be64(0);
cm_id_priv->timeout_ms = cm_convert_to_ms(
param->primary_path->packet_life_time) * 2 +
cm_convert_to_ms(
param->remote_cm_response_timeout);
cm_id_priv->max_cm_retries = param->max_cm_retries;
cm_id_priv->initiator_depth = param->initiator_depth;
cm_id_priv->responder_resources = param->responder_resources;
cm_id_priv->retry_count = param->retry_count;
cm_id_priv->path_mtu = param->primary_path->mtu;
cm_id_priv->pkey = param->primary_path->pkey;
cm_id_priv->qp_type = param->qp_type;
ret = cm_alloc_msg(cm_id_priv, &cm_id_priv->msg);
if (ret)
goto error1;
req_msg = (struct cm_req_msg *) cm_id_priv->msg->mad;
cm_format_req(req_msg, cm_id_priv, param);
cm_id_priv->tid = req_msg->hdr.tid;
cm_id_priv->msg->timeout_ms = cm_id_priv->timeout_ms;
cm_id_priv->msg->context[1] = (void *) (unsigned long) IB_CM_REQ_SENT;
cm_id_priv->local_qpn = cm_req_get_local_qpn(req_msg);
cm_id_priv->rq_psn = cm_req_get_starting_psn(req_msg);
spin_lock_irqsave(&cm_id_priv->lock, flags);
ret = ib_post_send_mad(cm_id_priv->msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
goto error2;
}
BUG_ON(cm_id->state != IB_CM_IDLE);
cm_id->state = IB_CM_REQ_SENT;
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return 0;
error2: cm_free_msg(cm_id_priv->msg);
error1: kfree(cm_id_priv->timewait_info);
out: return ret;
}
EXPORT_SYMBOL(ib_send_cm_req);
static int cm_issue_rej(struct cm_port *port,
struct ib_mad_recv_wc *mad_recv_wc,
enum ib_cm_rej_reason reason,
enum cm_msg_response msg_rejected,
void *ari, u8 ari_length)
{
struct ib_mad_send_buf *msg = NULL;
struct cm_rej_msg *rej_msg, *rcv_msg;
int ret;
ret = cm_alloc_response_msg(port, mad_recv_wc, &msg);
if (ret)
return ret;
/* We just need common CM header information. Cast to any message. */
rcv_msg = (struct cm_rej_msg *) mad_recv_wc->recv_buf.mad;
rej_msg = (struct cm_rej_msg *) msg->mad;
cm_format_mad_hdr(&rej_msg->hdr, CM_REJ_ATTR_ID, rcv_msg->hdr.tid);
rej_msg->remote_comm_id = rcv_msg->local_comm_id;
rej_msg->local_comm_id = rcv_msg->remote_comm_id;
cm_rej_set_msg_rejected(rej_msg, msg_rejected);
rej_msg->reason = cpu_to_be16(reason);
if (ari && ari_length) {
cm_rej_set_reject_info_len(rej_msg, ari_length);
memcpy(rej_msg->ari, ari, ari_length);
}
ret = ib_post_send_mad(msg, NULL);
if (ret)
cm_free_msg(msg);
return ret;
}
static inline int cm_is_active_peer(__be64 local_ca_guid, __be64 remote_ca_guid,
__be32 local_qpn, __be32 remote_qpn)
{
return (be64_to_cpu(local_ca_guid) > be64_to_cpu(remote_ca_guid) ||
((local_ca_guid == remote_ca_guid) &&
(be32_to_cpu(local_qpn) > be32_to_cpu(remote_qpn))));
}
static bool cm_req_has_alt_path(struct cm_req_msg *req_msg)
{
return ((req_msg->alt_local_lid) ||
(ib_is_opa_gid(&req_msg->alt_local_gid)));
}
static void cm_path_set_rec_type(struct ib_device *ib_device, u8 port_num,
struct sa_path_rec *path, union ib_gid *gid)
{
if (ib_is_opa_gid(gid) && rdma_cap_opa_ah(ib_device, port_num))
path->rec_type = SA_PATH_REC_TYPE_OPA;
else
path->rec_type = SA_PATH_REC_TYPE_IB;
}
static void cm_format_path_lid_from_req(struct cm_req_msg *req_msg,
struct sa_path_rec *primary_path,
struct sa_path_rec *alt_path)
{
u32 lid;
if (primary_path->rec_type != SA_PATH_REC_TYPE_OPA) {
sa_path_set_dlid(primary_path,
ntohs(req_msg->primary_local_lid));
sa_path_set_slid(primary_path,
ntohs(req_msg->primary_remote_lid));
} else {
lid = opa_get_lid_from_gid(&req_msg->primary_local_gid);
sa_path_set_dlid(primary_path, lid);
lid = opa_get_lid_from_gid(&req_msg->primary_remote_gid);
sa_path_set_slid(primary_path, lid);
}
if (!cm_req_has_alt_path(req_msg))
return;
if (alt_path->rec_type != SA_PATH_REC_TYPE_OPA) {
sa_path_set_dlid(alt_path, ntohs(req_msg->alt_local_lid));
sa_path_set_slid(alt_path, ntohs(req_msg->alt_remote_lid));
} else {
lid = opa_get_lid_from_gid(&req_msg->alt_local_gid);
sa_path_set_dlid(alt_path, lid);
lid = opa_get_lid_from_gid(&req_msg->alt_remote_gid);
sa_path_set_slid(alt_path, lid);
}
}
static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
struct sa_path_rec *primary_path,
struct sa_path_rec *alt_path)
{
primary_path->dgid = req_msg->primary_local_gid;
primary_path->sgid = req_msg->primary_remote_gid;
primary_path->flow_label = cm_req_get_primary_flow_label(req_msg);
primary_path->hop_limit = req_msg->primary_hop_limit;
primary_path->traffic_class = req_msg->primary_traffic_class;
primary_path->reversible = 1;
primary_path->pkey = req_msg->pkey;
primary_path->sl = cm_req_get_primary_sl(req_msg);
primary_path->mtu_selector = IB_SA_EQ;
primary_path->mtu = cm_req_get_path_mtu(req_msg);
primary_path->rate_selector = IB_SA_EQ;
primary_path->rate = cm_req_get_primary_packet_rate(req_msg);
primary_path->packet_life_time_selector = IB_SA_EQ;
primary_path->packet_life_time =
cm_req_get_primary_local_ack_timeout(req_msg);
primary_path->packet_life_time -= (primary_path->packet_life_time > 0);
RDMA/SA: Fix kernel panic in CMA request handler flow Commit 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields) moved the service_id to be specific attribute for IB and OPA SA Path Record, and thus wasn't assigned for RoCE. This caused to the following kernel panic in the CMA request handler flow: [ 27.074594] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 27.074731] IP: __radix_tree_lookup+0x1d/0xe0 ... [ 27.075356] Workqueue: ib_cm cm_work_handler [ib_cm] [ 27.075401] task: ffff88022e3b8000 task.stack: ffffc90001298000 [ 27.075449] RIP: 0010:__radix_tree_lookup+0x1d/0xe0 ... [ 27.075979] Call Trace: [ 27.076015] radix_tree_lookup+0xd/0x10 [ 27.076055] cma_ps_find+0x59/0x70 [rdma_cm] [ 27.076097] cma_id_from_event+0xd2/0x470 [rdma_cm] [ 27.076144] ? ib_init_ah_from_path+0x39a/0x590 [ib_core] [ 27.076193] cma_req_handler+0x25/0x480 [rdma_cm] [ 27.076237] cm_process_work+0x25/0x120 [ib_cm] [ 27.076280] ? cm_get_bth_pkey.isra.62+0x3c/0xa0 [ib_cm] [ 27.076350] cm_req_handler+0xb03/0xd40 [ib_cm] [ 27.076430] ? sched_clock_cpu+0x11/0xb0 [ 27.076478] cm_work_handler+0x194/0x1588 [ib_cm] [ 27.076525] process_one_work+0x160/0x410 [ 27.076565] worker_thread+0x137/0x4a0 [ 27.076614] kthread+0x112/0x150 [ 27.076684] ? max_active_store+0x60/0x60 [ 27.077642] ? kthread_park+0x90/0x90 [ 27.078530] ret_from_fork+0x2c/0x40 This patch moves it back to the common SA Path Record structure and removes the redundant setter and getter. Tested on Connect-IB and Connect-X4 in Infiniband and RoCE respectively. Fixes: 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB ands ROCE specific fields) Signed-off-by: Majd Dibbiny <majd@mellanox.com> Reviewed-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-05-22 00:09:54 +08:00
primary_path->service_id = req_msg->service_id;
if (cm_req_has_alt_path(req_msg)) {
alt_path->dgid = req_msg->alt_local_gid;
alt_path->sgid = req_msg->alt_remote_gid;
alt_path->flow_label = cm_req_get_alt_flow_label(req_msg);
alt_path->hop_limit = req_msg->alt_hop_limit;
alt_path->traffic_class = req_msg->alt_traffic_class;
alt_path->reversible = 1;
alt_path->pkey = req_msg->pkey;
alt_path->sl = cm_req_get_alt_sl(req_msg);
alt_path->mtu_selector = IB_SA_EQ;
alt_path->mtu = cm_req_get_path_mtu(req_msg);
alt_path->rate_selector = IB_SA_EQ;
alt_path->rate = cm_req_get_alt_packet_rate(req_msg);
alt_path->packet_life_time_selector = IB_SA_EQ;
alt_path->packet_life_time =
cm_req_get_alt_local_ack_timeout(req_msg);
alt_path->packet_life_time -= (alt_path->packet_life_time > 0);
RDMA/SA: Fix kernel panic in CMA request handler flow Commit 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields) moved the service_id to be specific attribute for IB and OPA SA Path Record, and thus wasn't assigned for RoCE. This caused to the following kernel panic in the CMA request handler flow: [ 27.074594] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 27.074731] IP: __radix_tree_lookup+0x1d/0xe0 ... [ 27.075356] Workqueue: ib_cm cm_work_handler [ib_cm] [ 27.075401] task: ffff88022e3b8000 task.stack: ffffc90001298000 [ 27.075449] RIP: 0010:__radix_tree_lookup+0x1d/0xe0 ... [ 27.075979] Call Trace: [ 27.076015] radix_tree_lookup+0xd/0x10 [ 27.076055] cma_ps_find+0x59/0x70 [rdma_cm] [ 27.076097] cma_id_from_event+0xd2/0x470 [rdma_cm] [ 27.076144] ? ib_init_ah_from_path+0x39a/0x590 [ib_core] [ 27.076193] cma_req_handler+0x25/0x480 [rdma_cm] [ 27.076237] cm_process_work+0x25/0x120 [ib_cm] [ 27.076280] ? cm_get_bth_pkey.isra.62+0x3c/0xa0 [ib_cm] [ 27.076350] cm_req_handler+0xb03/0xd40 [ib_cm] [ 27.076430] ? sched_clock_cpu+0x11/0xb0 [ 27.076478] cm_work_handler+0x194/0x1588 [ib_cm] [ 27.076525] process_one_work+0x160/0x410 [ 27.076565] worker_thread+0x137/0x4a0 [ 27.076614] kthread+0x112/0x150 [ 27.076684] ? max_active_store+0x60/0x60 [ 27.077642] ? kthread_park+0x90/0x90 [ 27.078530] ret_from_fork+0x2c/0x40 This patch moves it back to the common SA Path Record structure and removes the redundant setter and getter. Tested on Connect-IB and Connect-X4 in Infiniband and RoCE respectively. Fixes: 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB ands ROCE specific fields) Signed-off-by: Majd Dibbiny <majd@mellanox.com> Reviewed-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-05-22 00:09:54 +08:00
alt_path->service_id = req_msg->service_id;
}
cm_format_path_lid_from_req(req_msg, primary_path, alt_path);
}
static u16 cm_get_bth_pkey(struct cm_work *work)
{
struct ib_device *ib_dev = work->port->cm_dev->ib_device;
u8 port_num = work->port->port_num;
u16 pkey_index = work->mad_recv_wc->wc->pkey_index;
u16 pkey;
int ret;
ret = ib_get_cached_pkey(ib_dev, port_num, pkey_index, &pkey);
if (ret) {
dev_warn_ratelimited(&ib_dev->dev, "ib_cm: Couldn't retrieve pkey for incoming request (port %d, pkey index %d). %d\n",
port_num, pkey_index, ret);
return 0;
}
return pkey;
}
static void cm_format_req_event(struct cm_work *work,
struct cm_id_private *cm_id_priv,
struct ib_cm_id *listen_id)
{
struct cm_req_msg *req_msg;
struct ib_cm_req_event_param *param;
req_msg = (struct cm_req_msg *)work->mad_recv_wc->recv_buf.mad;
param = &work->cm_event.param.req_rcvd;
param->listen_id = listen_id;
param->bth_pkey = cm_get_bth_pkey(work);
param->port = cm_id_priv->av.port->port_num;
param->primary_path = &work->path[0];
if (cm_req_has_alt_path(req_msg))
param->alternate_path = &work->path[1];
else
param->alternate_path = NULL;
param->remote_ca_guid = req_msg->local_ca_guid;
param->remote_qkey = be32_to_cpu(req_msg->local_qkey);
param->remote_qpn = be32_to_cpu(cm_req_get_local_qpn(req_msg));
param->qp_type = cm_req_get_qp_type(req_msg);
param->starting_psn = be32_to_cpu(cm_req_get_starting_psn(req_msg));
param->responder_resources = cm_req_get_init_depth(req_msg);
param->initiator_depth = cm_req_get_resp_res(req_msg);
param->local_cm_response_timeout =
cm_req_get_remote_resp_timeout(req_msg);
param->flow_control = cm_req_get_flow_ctrl(req_msg);
param->remote_cm_response_timeout =
cm_req_get_local_resp_timeout(req_msg);
param->retry_count = cm_req_get_retry_count(req_msg);
param->rnr_retry_count = cm_req_get_rnr_retry_count(req_msg);
param->srq = cm_req_get_srq(req_msg);
work->cm_event.private_data = &req_msg->private_data;
}
static void cm_process_work(struct cm_id_private *cm_id_priv,
struct cm_work *work)
{
int ret;
/* We will typically only have the current event to report. */
ret = cm_id_priv->id.cm_handler(&cm_id_priv->id, &work->cm_event);
cm_free_work(work);
while (!ret && !atomic_add_negative(-1, &cm_id_priv->work_count)) {
spin_lock_irq(&cm_id_priv->lock);
work = cm_dequeue_work(cm_id_priv);
spin_unlock_irq(&cm_id_priv->lock);
BUG_ON(!work);
ret = cm_id_priv->id.cm_handler(&cm_id_priv->id,
&work->cm_event);
cm_free_work(work);
}
cm_deref_id(cm_id_priv);
if (ret)
cm_destroy_id(&cm_id_priv->id, ret);
}
static void cm_format_mra(struct cm_mra_msg *mra_msg,
struct cm_id_private *cm_id_priv,
enum cm_msg_response msg_mraed, u8 service_timeout,
const void *private_data, u8 private_data_len)
{
cm_format_mad_hdr(&mra_msg->hdr, CM_MRA_ATTR_ID, cm_id_priv->tid);
cm_mra_set_msg_mraed(mra_msg, msg_mraed);
mra_msg->local_comm_id = cm_id_priv->id.local_id;
mra_msg->remote_comm_id = cm_id_priv->id.remote_id;
cm_mra_set_service_timeout(mra_msg, service_timeout);
if (private_data && private_data_len)
memcpy(mra_msg->private_data, private_data, private_data_len);
}
static void cm_format_rej(struct cm_rej_msg *rej_msg,
struct cm_id_private *cm_id_priv,
enum ib_cm_rej_reason reason,
void *ari,
u8 ari_length,
const void *private_data,
u8 private_data_len)
{
cm_format_mad_hdr(&rej_msg->hdr, CM_REJ_ATTR_ID, cm_id_priv->tid);
rej_msg->remote_comm_id = cm_id_priv->id.remote_id;
switch(cm_id_priv->id.state) {
case IB_CM_REQ_RCVD:
rej_msg->local_comm_id = 0;
cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_REQ);
break;
case IB_CM_MRA_REQ_SENT:
rej_msg->local_comm_id = cm_id_priv->id.local_id;
cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_REQ);
break;
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
rej_msg->local_comm_id = cm_id_priv->id.local_id;
cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_REP);
break;
default:
rej_msg->local_comm_id = cm_id_priv->id.local_id;
cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_OTHER);
break;
}
rej_msg->reason = cpu_to_be16(reason);
if (ari && ari_length) {
cm_rej_set_reject_info_len(rej_msg, ari_length);
memcpy(rej_msg->ari, ari, ari_length);
}
if (private_data && private_data_len)
memcpy(rej_msg->private_data, private_data, private_data_len);
}
static void cm_dup_req_handler(struct cm_work *work,
struct cm_id_private *cm_id_priv)
{
struct ib_mad_send_buf *msg = NULL;
int ret;
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_REQ_COUNTER]);
/* Quick state check to discard duplicate REQs. */
if (cm_id_priv->id.state == IB_CM_REQ_RCVD)
return;
ret = cm_alloc_response_msg(work->port, work->mad_recv_wc, &msg);
if (ret)
return;
spin_lock_irq(&cm_id_priv->lock);
switch (cm_id_priv->id.state) {
case IB_CM_MRA_REQ_SENT:
cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
CM_MSG_RESPONSE_REQ, cm_id_priv->service_timeout,
cm_id_priv->private_data,
cm_id_priv->private_data_len);
break;
case IB_CM_TIMEWAIT:
cm_format_rej((struct cm_rej_msg *) msg->mad, cm_id_priv,
IB_CM_REJ_STALE_CONN, NULL, 0, NULL, 0);
break;
default:
goto unlock;
}
spin_unlock_irq(&cm_id_priv->lock);
ret = ib_post_send_mad(msg, NULL);
if (ret)
goto free;
return;
unlock: spin_unlock_irq(&cm_id_priv->lock);
free: cm_free_msg(msg);
}
static struct cm_id_private * cm_match_req(struct cm_work *work,
struct cm_id_private *cm_id_priv)
{
struct cm_id_private *listen_cm_id_priv, *cur_cm_id_priv;
struct cm_timewait_info *timewait_info;
struct cm_req_msg *req_msg;
struct ib_cm_id *cm_id;
req_msg = (struct cm_req_msg *)work->mad_recv_wc->recv_buf.mad;
/* Check for possible duplicate REQ. */
spin_lock_irq(&cm.lock);
timewait_info = cm_insert_remote_id(cm_id_priv->timewait_info);
if (timewait_info) {
cur_cm_id_priv = cm_get_id(timewait_info->work.local_id,
timewait_info->work.remote_id);
spin_unlock_irq(&cm.lock);
if (cur_cm_id_priv) {
cm_dup_req_handler(work, cur_cm_id_priv);
cm_deref_id(cur_cm_id_priv);
}
return NULL;
}
/* Check for stale connections. */
timewait_info = cm_insert_remote_qpn(cm_id_priv->timewait_info);
if (timewait_info) {
cm_cleanup_timewait(cm_id_priv->timewait_info);
cur_cm_id_priv = cm_get_id(timewait_info->work.local_id,
timewait_info->work.remote_id);
spin_unlock_irq(&cm.lock);
cm_issue_rej(work->port, work->mad_recv_wc,
IB_CM_REJ_STALE_CONN, CM_MSG_RESPONSE_REQ,
NULL, 0);
if (cur_cm_id_priv) {
cm_id = &cur_cm_id_priv->id;
ib_send_cm_dreq(cm_id, NULL, 0);
cm_deref_id(cur_cm_id_priv);
}
return NULL;
}
/* Find matching listen request. */
listen_cm_id_priv = cm_find_listen(cm_id_priv->id.device,
req_msg->service_id);
if (!listen_cm_id_priv) {
cm_cleanup_timewait(cm_id_priv->timewait_info);
spin_unlock_irq(&cm.lock);
cm_issue_rej(work->port, work->mad_recv_wc,
IB_CM_REJ_INVALID_SERVICE_ID, CM_MSG_RESPONSE_REQ,
NULL, 0);
goto out;
}
atomic_inc(&listen_cm_id_priv->refcount);
atomic_inc(&cm_id_priv->refcount);
cm_id_priv->id.state = IB_CM_REQ_RCVD;
atomic_inc(&cm_id_priv->work_count);
spin_unlock_irq(&cm.lock);
out:
return listen_cm_id_priv;
}
/*
* Work-around for inter-subnet connections. If the LIDs are permissive,
* we need to override the LID/SL data in the REQ with the LID information
* in the work completion.
*/
static void cm_process_routed_req(struct cm_req_msg *req_msg, struct ib_wc *wc)
{
if (!cm_req_get_primary_subnet_local(req_msg)) {
if (req_msg->primary_local_lid == IB_LID_PERMISSIVE) {
req_msg->primary_local_lid = ib_lid_be16(wc->slid);
cm_req_set_primary_sl(req_msg, wc->sl);
}
if (req_msg->primary_remote_lid == IB_LID_PERMISSIVE)
req_msg->primary_remote_lid = cpu_to_be16(wc->dlid_path_bits);
}
if (!cm_req_get_alt_subnet_local(req_msg)) {
if (req_msg->alt_local_lid == IB_LID_PERMISSIVE) {
req_msg->alt_local_lid = ib_lid_be16(wc->slid);
cm_req_set_alt_sl(req_msg, wc->sl);
}
if (req_msg->alt_remote_lid == IB_LID_PERMISSIVE)
req_msg->alt_remote_lid = cpu_to_be16(wc->dlid_path_bits);
}
}
static int cm_req_handler(struct cm_work *work)
{
struct ib_cm_id *cm_id;
struct cm_id_private *cm_id_priv, *listen_cm_id_priv;
struct cm_req_msg *req_msg;
union ib_gid gid;
struct ib_gid_attr gid_attr;
const struct ib_global_route *grh;
int ret;
req_msg = (struct cm_req_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id = ib_create_cm_id(work->port->cm_dev->ib_device, NULL, NULL);
if (IS_ERR(cm_id))
return PTR_ERR(cm_id);
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
cm_id_priv->id.remote_id = req_msg->local_comm_id;
cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
work->mad_recv_wc->recv_buf.grh,
&cm_id_priv->av);
cm_id_priv->timewait_info = cm_create_timewait_info(cm_id_priv->
id.local_id);
if (IS_ERR(cm_id_priv->timewait_info)) {
ret = PTR_ERR(cm_id_priv->timewait_info);
goto destroy;
}
cm_id_priv->timewait_info->work.remote_id = req_msg->local_comm_id;
cm_id_priv->timewait_info->remote_ca_guid = req_msg->local_ca_guid;
cm_id_priv->timewait_info->remote_qpn = cm_req_get_local_qpn(req_msg);
listen_cm_id_priv = cm_match_req(work, cm_id_priv);
if (!listen_cm_id_priv) {
ret = -EINVAL;
kfree(cm_id_priv->timewait_info);
goto destroy;
}
cm_id_priv->id.cm_handler = listen_cm_id_priv->id.cm_handler;
cm_id_priv->id.context = listen_cm_id_priv->id.context;
cm_id_priv->id.service_id = req_msg->service_id;
cm_id_priv->id.service_mask = ~cpu_to_be64(0);
cm_process_routed_req(req_msg, work->mad_recv_wc->wc);
IB/core: Ethernet L2 attributes in verbs/cm structures This patch add the support for Ethernet L2 attributes in the verbs/cm/cma structures. When dealing with L2 Ethernet, we should use smac, dmac, vlan ID and priority in a similar manner that the IB L2 (and the L4 PKEY) attributes are used. Thus, those attributes were added to the following structures: * ib_ah_attr - added dmac * ib_qp_attr - added smac and vlan_id, (sl remains vlan priority) * ib_wc - added smac, vlan_id * ib_sa_path_rec - added smac, dmac, vlan_id * cm_av - added smac and vlan_id For the path record structure, extra care was taken to avoid the new fields when packing it into wire format, so we don't break the IB CM and SA wire protocol. On the active side, the CM fills. its internal structures from the path provided by the ULP. We add there taking the ETH L2 attributes and placing them into the CM Address Handle (struct cm_av). On the passive side, the CM fills its internal structures from the WC associated with the REQ message. We add there taking the ETH L2 attributes from the WC. When the HW driver provides the required ETH L2 attributes in the WC, they set the IB_WC_WITH_SMAC and IB_WC_WITH_VLAN flags. The IB core code checks for the presence of these flags, and in their absence does address resolution from the ib_init_ah_from_wc() helper function. ib_modify_qp_is_ok is also updated to consider the link layer. Some parameters are mandatory for Ethernet link layer, while they are irrelevant for IB. Vendor drivers are modified to support the new function signature. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2013-12-13 00:03:11 +08:00
memset(&work->path[0], 0, sizeof(work->path[0]));
if (cm_req_has_alt_path(req_msg))
memset(&work->path[1], 0, sizeof(work->path[1]));
grh = rdma_ah_read_grh(&cm_id_priv->av.ah_attr);
ret = ib_get_cached_gid(work->port->cm_dev->ib_device,
work->port->port_num,
grh->sgid_index,
&gid, &gid_attr);
if (!ret) {
if (gid_attr.ndev) {
work->path[0].rec_type =
sa_conv_gid_to_pathrec_type(gid_attr.gid_type);
sa_path_set_ifindex(&work->path[0],
gid_attr.ndev->ifindex);
sa_path_set_ndev(&work->path[0],
dev_net(gid_attr.ndev));
dev_put(gid_attr.ndev);
} else {
cm_path_set_rec_type(work->port->cm_dev->ib_device,
work->port->port_num,
&work->path[0],
&req_msg->primary_local_gid);
}
if (cm_req_has_alt_path(req_msg))
work->path[1].rec_type = work->path[0].rec_type;
cm_format_paths_from_req(req_msg, &work->path[0],
&work->path[1]);
if (cm_id_priv->av.ah_attr.type == RDMA_AH_ATTR_TYPE_ROCE)
sa_path_set_dmac(&work->path[0],
cm_id_priv->av.ah_attr.roce.dmac);
work->path[0].hop_limit = grh->hop_limit;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
ret = cm_init_av_by_path(&work->path[0], &cm_id_priv->av,
cm_id_priv);
}
if (ret) {
int err = ib_get_cached_gid(work->port->cm_dev->ib_device,
work->port->port_num, 0,
&work->path[0].sgid,
&gid_attr);
if (!err && gid_attr.ndev) {
work->path[0].rec_type =
sa_conv_gid_to_pathrec_type(gid_attr.gid_type);
sa_path_set_ifindex(&work->path[0],
gid_attr.ndev->ifindex);
sa_path_set_ndev(&work->path[0],
dev_net(gid_attr.ndev));
dev_put(gid_attr.ndev);
} else {
cm_path_set_rec_type(work->port->cm_dev->ib_device,
work->port->port_num,
&work->path[0],
&req_msg->primary_local_gid);
}
if (cm_req_has_alt_path(req_msg))
work->path[1].rec_type = work->path[0].rec_type;
ib_send_cm_rej(cm_id, IB_CM_REJ_INVALID_GID,
&work->path[0].sgid, sizeof work->path[0].sgid,
NULL, 0);
goto rejected;
}
if (cm_req_has_alt_path(req_msg)) {
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
ret = cm_init_av_by_path(&work->path[1], &cm_id_priv->alt_av,
cm_id_priv);
if (ret) {
ib_send_cm_rej(cm_id, IB_CM_REJ_INVALID_ALT_GID,
&work->path[0].sgid,
sizeof work->path[0].sgid, NULL, 0);
goto rejected;
}
}
cm_id_priv->tid = req_msg->hdr.tid;
cm_id_priv->timeout_ms = cm_convert_to_ms(
cm_req_get_local_resp_timeout(req_msg));
cm_id_priv->max_cm_retries = cm_req_get_max_cm_retries(req_msg);
cm_id_priv->remote_qpn = cm_req_get_local_qpn(req_msg);
cm_id_priv->initiator_depth = cm_req_get_resp_res(req_msg);
cm_id_priv->responder_resources = cm_req_get_init_depth(req_msg);
cm_id_priv->path_mtu = cm_req_get_path_mtu(req_msg);
cm_id_priv->pkey = req_msg->pkey;
cm_id_priv->sq_psn = cm_req_get_starting_psn(req_msg);
cm_id_priv->retry_count = cm_req_get_retry_count(req_msg);
cm_id_priv->rnr_retry_count = cm_req_get_rnr_retry_count(req_msg);
cm_id_priv->qp_type = cm_req_get_qp_type(req_msg);
cm_format_req_event(work, cm_id_priv, &listen_cm_id_priv->id);
cm_process_work(cm_id_priv, work);
cm_deref_id(listen_cm_id_priv);
return 0;
rejected:
atomic_dec(&cm_id_priv->refcount);
cm_deref_id(listen_cm_id_priv);
destroy:
ib_destroy_cm_id(cm_id);
return ret;
}
static void cm_format_rep(struct cm_rep_msg *rep_msg,
struct cm_id_private *cm_id_priv,
struct ib_cm_rep_param *param)
{
cm_format_mad_hdr(&rep_msg->hdr, CM_REP_ATTR_ID, cm_id_priv->tid);
rep_msg->local_comm_id = cm_id_priv->id.local_id;
rep_msg->remote_comm_id = cm_id_priv->id.remote_id;
cm_rep_set_starting_psn(rep_msg, cpu_to_be32(param->starting_psn));
rep_msg->resp_resources = param->responder_resources;
cm_rep_set_target_ack_delay(rep_msg,
cm_id_priv->av.port->cm_dev->ack_delay);
cm_rep_set_failover(rep_msg, param->failover_accepted);
cm_rep_set_rnr_retry_count(rep_msg, param->rnr_retry_count);
rep_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
if (cm_id_priv->qp_type != IB_QPT_XRC_TGT) {
rep_msg->initiator_depth = param->initiator_depth;
cm_rep_set_flow_ctrl(rep_msg, param->flow_control);
cm_rep_set_srq(rep_msg, param->srq);
cm_rep_set_local_qpn(rep_msg, cpu_to_be32(param->qp_num));
} else {
cm_rep_set_srq(rep_msg, 1);
cm_rep_set_local_eecn(rep_msg, cpu_to_be32(param->qp_num));
}
if (param->private_data && param->private_data_len)
memcpy(rep_msg->private_data, param->private_data,
param->private_data_len);
}
int ib_send_cm_rep(struct ib_cm_id *cm_id,
struct ib_cm_rep_param *param)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
struct cm_rep_msg *rep_msg;
unsigned long flags;
int ret;
if (param->private_data &&
param->private_data_len > IB_CM_REP_PRIVATE_DATA_SIZE)
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_REQ_RCVD &&
cm_id->state != IB_CM_MRA_REQ_SENT) {
ret = -EINVAL;
goto out;
}
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto out;
rep_msg = (struct cm_rep_msg *) msg->mad;
cm_format_rep(rep_msg, cm_id_priv, param);
msg->timeout_ms = cm_id_priv->timeout_ms;
msg->context[1] = (void *) (unsigned long) IB_CM_REP_SENT;
ret = ib_post_send_mad(msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
return ret;
}
cm_id->state = IB_CM_REP_SENT;
cm_id_priv->msg = msg;
cm_id_priv->initiator_depth = param->initiator_depth;
cm_id_priv->responder_resources = param->responder_resources;
cm_id_priv->rq_psn = cm_rep_get_starting_psn(rep_msg);
cm_id_priv->local_qpn = cpu_to_be32(param->qp_num & 0xFFFFFF);
out: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_rep);
static void cm_format_rtu(struct cm_rtu_msg *rtu_msg,
struct cm_id_private *cm_id_priv,
const void *private_data,
u8 private_data_len)
{
cm_format_mad_hdr(&rtu_msg->hdr, CM_RTU_ATTR_ID, cm_id_priv->tid);
rtu_msg->local_comm_id = cm_id_priv->id.local_id;
rtu_msg->remote_comm_id = cm_id_priv->id.remote_id;
if (private_data && private_data_len)
memcpy(rtu_msg->private_data, private_data, private_data_len);
}
int ib_send_cm_rtu(struct ib_cm_id *cm_id,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
void *data;
int ret;
if (private_data && private_data_len > IB_CM_RTU_PRIVATE_DATA_SIZE)
return -EINVAL;
data = cm_copy_private_data(private_data, private_data_len);
if (IS_ERR(data))
return PTR_ERR(data);
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_REP_RCVD &&
cm_id->state != IB_CM_MRA_REP_SENT) {
ret = -EINVAL;
goto error;
}
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto error;
cm_format_rtu((struct cm_rtu_msg *) msg->mad, cm_id_priv,
private_data, private_data_len);
ret = ib_post_send_mad(msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
kfree(data);
return ret;
}
cm_id->state = IB_CM_ESTABLISHED;
cm_set_private_data(cm_id_priv, data, private_data_len);
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return 0;
error: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
kfree(data);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_rtu);
static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type)
{
struct cm_rep_msg *rep_msg;
struct ib_cm_rep_event_param *param;
rep_msg = (struct cm_rep_msg *)work->mad_recv_wc->recv_buf.mad;
param = &work->cm_event.param.rep_rcvd;
param->remote_ca_guid = rep_msg->local_ca_guid;
param->remote_qkey = be32_to_cpu(rep_msg->local_qkey);
param->remote_qpn = be32_to_cpu(cm_rep_get_qpn(rep_msg, qp_type));
param->starting_psn = be32_to_cpu(cm_rep_get_starting_psn(rep_msg));
param->responder_resources = rep_msg->initiator_depth;
param->initiator_depth = rep_msg->resp_resources;
param->target_ack_delay = cm_rep_get_target_ack_delay(rep_msg);
param->failover_accepted = cm_rep_get_failover(rep_msg);
param->flow_control = cm_rep_get_flow_ctrl(rep_msg);
param->rnr_retry_count = cm_rep_get_rnr_retry_count(rep_msg);
param->srq = cm_rep_get_srq(rep_msg);
work->cm_event.private_data = &rep_msg->private_data;
}
static void cm_dup_rep_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_rep_msg *rep_msg;
struct ib_mad_send_buf *msg = NULL;
int ret;
rep_msg = (struct cm_rep_msg *) work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(rep_msg->remote_comm_id,
rep_msg->local_comm_id);
if (!cm_id_priv)
return;
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_REP_COUNTER]);
ret = cm_alloc_response_msg(work->port, work->mad_recv_wc, &msg);
if (ret)
goto deref;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state == IB_CM_ESTABLISHED)
cm_format_rtu((struct cm_rtu_msg *) msg->mad, cm_id_priv,
cm_id_priv->private_data,
cm_id_priv->private_data_len);
else if (cm_id_priv->id.state == IB_CM_MRA_REP_SENT)
cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
CM_MSG_RESPONSE_REP, cm_id_priv->service_timeout,
cm_id_priv->private_data,
cm_id_priv->private_data_len);
else
goto unlock;
spin_unlock_irq(&cm_id_priv->lock);
ret = ib_post_send_mad(msg, NULL);
if (ret)
goto free;
goto deref;
unlock: spin_unlock_irq(&cm_id_priv->lock);
free: cm_free_msg(msg);
deref: cm_deref_id(cm_id_priv);
}
static int cm_rep_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_rep_msg *rep_msg;
int ret;
struct cm_id_private *cur_cm_id_priv;
struct ib_cm_id *cm_id;
struct cm_timewait_info *timewait_info;
rep_msg = (struct cm_rep_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(rep_msg->remote_comm_id, 0);
if (!cm_id_priv) {
cm_dup_rep_handler(work);
return -EINVAL;
}
cm_format_rep_event(work, cm_id_priv->qp_type);
spin_lock_irq(&cm_id_priv->lock);
switch (cm_id_priv->id.state) {
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
break;
default:
spin_unlock_irq(&cm_id_priv->lock);
ret = -EINVAL;
goto error;
}
cm_id_priv->timewait_info->work.remote_id = rep_msg->local_comm_id;
cm_id_priv->timewait_info->remote_ca_guid = rep_msg->local_ca_guid;
cm_id_priv->timewait_info->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type);
spin_lock(&cm.lock);
/* Check for duplicate REP. */
if (cm_insert_remote_id(cm_id_priv->timewait_info)) {
spin_unlock(&cm.lock);
spin_unlock_irq(&cm_id_priv->lock);
ret = -EINVAL;
goto error;
}
/* Check for a stale connection. */
timewait_info = cm_insert_remote_qpn(cm_id_priv->timewait_info);
if (timewait_info) {
rb_erase(&cm_id_priv->timewait_info->remote_id_node,
&cm.remote_id_table);
cm_id_priv->timewait_info->inserted_remote_id = 0;
cur_cm_id_priv = cm_get_id(timewait_info->work.local_id,
timewait_info->work.remote_id);
spin_unlock(&cm.lock);
spin_unlock_irq(&cm_id_priv->lock);
cm_issue_rej(work->port, work->mad_recv_wc,
IB_CM_REJ_STALE_CONN, CM_MSG_RESPONSE_REP,
NULL, 0);
ret = -EINVAL;
if (cur_cm_id_priv) {
cm_id = &cur_cm_id_priv->id;
ib_send_cm_dreq(cm_id, NULL, 0);
cm_deref_id(cur_cm_id_priv);
}
goto error;
}
spin_unlock(&cm.lock);
cm_id_priv->id.state = IB_CM_REP_RCVD;
cm_id_priv->id.remote_id = rep_msg->local_comm_id;
cm_id_priv->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type);
cm_id_priv->initiator_depth = rep_msg->resp_resources;
cm_id_priv->responder_resources = rep_msg->initiator_depth;
cm_id_priv->sq_psn = cm_rep_get_starting_psn(rep_msg);
cm_id_priv->rnr_retry_count = cm_rep_get_rnr_retry_count(rep_msg);
cm_id_priv->target_ack_delay = cm_rep_get_target_ack_delay(rep_msg);
cm_id_priv->av.timeout =
cm_ack_timeout(cm_id_priv->target_ack_delay,
cm_id_priv->av.timeout - 1);
cm_id_priv->alt_av.timeout =
cm_ack_timeout(cm_id_priv->target_ack_delay,
cm_id_priv->alt_av.timeout - 1);
/* todo: handle peer_to_peer */
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
error:
cm_deref_id(cm_id_priv);
return ret;
}
static int cm_establish_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
int ret;
/* See comment in cm_establish about lookup. */
cm_id_priv = cm_acquire_id(work->local_id, work->remote_id);
if (!cm_id_priv)
return -EINVAL;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_ESTABLISHED) {
spin_unlock_irq(&cm_id_priv->lock);
goto out;
}
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
static int cm_rtu_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_rtu_msg *rtu_msg;
int ret;
rtu_msg = (struct cm_rtu_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(rtu_msg->remote_comm_id,
rtu_msg->local_comm_id);
if (!cm_id_priv)
return -EINVAL;
work->cm_event.private_data = &rtu_msg->private_data;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_REP_SENT &&
cm_id_priv->id.state != IB_CM_MRA_REP_RCVD) {
spin_unlock_irq(&cm_id_priv->lock);
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_RTU_COUNTER]);
goto out;
}
cm_id_priv->id.state = IB_CM_ESTABLISHED;
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
static void cm_format_dreq(struct cm_dreq_msg *dreq_msg,
struct cm_id_private *cm_id_priv,
const void *private_data,
u8 private_data_len)
{
cm_format_mad_hdr(&dreq_msg->hdr, CM_DREQ_ATTR_ID,
cm_form_tid(cm_id_priv, CM_MSG_SEQUENCE_DREQ));
dreq_msg->local_comm_id = cm_id_priv->id.local_id;
dreq_msg->remote_comm_id = cm_id_priv->id.remote_id;
cm_dreq_set_remote_qpn(dreq_msg, cm_id_priv->remote_qpn);
if (private_data && private_data_len)
memcpy(dreq_msg->private_data, private_data, private_data_len);
}
int ib_send_cm_dreq(struct ib_cm_id *cm_id,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
int ret;
if (private_data && private_data_len > IB_CM_DREQ_PRIVATE_DATA_SIZE)
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_ESTABLISHED) {
ret = -EINVAL;
goto out;
}
if (cm_id->lap_state == IB_CM_LAP_SENT ||
cm_id->lap_state == IB_CM_MRA_LAP_RCVD)
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret) {
cm_enter_timewait(cm_id_priv);
goto out;
}
cm_format_dreq((struct cm_dreq_msg *) msg->mad, cm_id_priv,
private_data, private_data_len);
msg->timeout_ms = cm_id_priv->timeout_ms;
msg->context[1] = (void *) (unsigned long) IB_CM_DREQ_SENT;
ret = ib_post_send_mad(msg, NULL);
if (ret) {
cm_enter_timewait(cm_id_priv);
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
return ret;
}
cm_id->state = IB_CM_DREQ_SENT;
cm_id_priv->msg = msg;
out: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_dreq);
static void cm_format_drep(struct cm_drep_msg *drep_msg,
struct cm_id_private *cm_id_priv,
const void *private_data,
u8 private_data_len)
{
cm_format_mad_hdr(&drep_msg->hdr, CM_DREP_ATTR_ID, cm_id_priv->tid);
drep_msg->local_comm_id = cm_id_priv->id.local_id;
drep_msg->remote_comm_id = cm_id_priv->id.remote_id;
if (private_data && private_data_len)
memcpy(drep_msg->private_data, private_data, private_data_len);
}
int ib_send_cm_drep(struct ib_cm_id *cm_id,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
void *data;
int ret;
if (private_data && private_data_len > IB_CM_DREP_PRIVATE_DATA_SIZE)
return -EINVAL;
data = cm_copy_private_data(private_data, private_data_len);
if (IS_ERR(data))
return PTR_ERR(data);
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_DREQ_RCVD) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
kfree(data);
return -EINVAL;
}
cm_set_private_data(cm_id_priv, data, private_data_len);
cm_enter_timewait(cm_id_priv);
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto out;
cm_format_drep((struct cm_drep_msg *) msg->mad, cm_id_priv,
private_data, private_data_len);
ret = ib_post_send_mad(msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
return ret;
}
out: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_drep);
static int cm_issue_drep(struct cm_port *port,
struct ib_mad_recv_wc *mad_recv_wc)
{
struct ib_mad_send_buf *msg = NULL;
struct cm_dreq_msg *dreq_msg;
struct cm_drep_msg *drep_msg;
int ret;
ret = cm_alloc_response_msg(port, mad_recv_wc, &msg);
if (ret)
return ret;
dreq_msg = (struct cm_dreq_msg *) mad_recv_wc->recv_buf.mad;
drep_msg = (struct cm_drep_msg *) msg->mad;
cm_format_mad_hdr(&drep_msg->hdr, CM_DREP_ATTR_ID, dreq_msg->hdr.tid);
drep_msg->remote_comm_id = dreq_msg->local_comm_id;
drep_msg->local_comm_id = dreq_msg->remote_comm_id;
ret = ib_post_send_mad(msg, NULL);
if (ret)
cm_free_msg(msg);
return ret;
}
static int cm_dreq_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_dreq_msg *dreq_msg;
struct ib_mad_send_buf *msg = NULL;
int ret;
dreq_msg = (struct cm_dreq_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(dreq_msg->remote_comm_id,
dreq_msg->local_comm_id);
if (!cm_id_priv) {
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_DREQ_COUNTER]);
cm_issue_drep(work->port, work->mad_recv_wc);
return -EINVAL;
}
work->cm_event.private_data = &dreq_msg->private_data;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->local_qpn != cm_dreq_get_remote_qpn(dreq_msg))
goto unlock;
switch (cm_id_priv->id.state) {
case IB_CM_REP_SENT:
case IB_CM_DREQ_SENT:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
break;
case IB_CM_ESTABLISHED:
if (cm_id_priv->id.lap_state == IB_CM_LAP_SENT ||
cm_id_priv->id.lap_state == IB_CM_MRA_LAP_RCVD)
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
break;
case IB_CM_MRA_REP_RCVD:
break;
case IB_CM_TIMEWAIT:
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_DREQ_COUNTER]);
msg = cm_alloc_response_msg_no_ah(work->port, work->mad_recv_wc);
if (IS_ERR(msg))
goto unlock;
cm_format_drep((struct cm_drep_msg *) msg->mad, cm_id_priv,
cm_id_priv->private_data,
cm_id_priv->private_data_len);
spin_unlock_irq(&cm_id_priv->lock);
if (cm_create_response_msg_ah(work->port, work->mad_recv_wc, msg) ||
ib_post_send_mad(msg, NULL))
cm_free_msg(msg);
goto deref;
case IB_CM_DREQ_RCVD:
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_DREQ_COUNTER]);
goto unlock;
default:
goto unlock;
}
cm_id_priv->id.state = IB_CM_DREQ_RCVD;
cm_id_priv->tid = dreq_msg->hdr.tid;
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
unlock: spin_unlock_irq(&cm_id_priv->lock);
deref: cm_deref_id(cm_id_priv);
return -EINVAL;
}
static int cm_drep_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_drep_msg *drep_msg;
int ret;
drep_msg = (struct cm_drep_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(drep_msg->remote_comm_id,
drep_msg->local_comm_id);
if (!cm_id_priv)
return -EINVAL;
work->cm_event.private_data = &drep_msg->private_data;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_DREQ_SENT &&
cm_id_priv->id.state != IB_CM_DREQ_RCVD) {
spin_unlock_irq(&cm_id_priv->lock);
goto out;
}
cm_enter_timewait(cm_id_priv);
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
int ib_send_cm_rej(struct ib_cm_id *cm_id,
enum ib_cm_rej_reason reason,
void *ari,
u8 ari_length,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
int ret;
if ((private_data && private_data_len > IB_CM_REJ_PRIVATE_DATA_SIZE) ||
(ari && ari_length > IB_CM_REJ_ARI_LENGTH))
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
switch (cm_id->state) {
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
case IB_CM_REQ_RCVD:
case IB_CM_MRA_REQ_SENT:
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
ret = cm_alloc_msg(cm_id_priv, &msg);
if (!ret)
cm_format_rej((struct cm_rej_msg *) msg->mad,
cm_id_priv, reason, ari, ari_length,
private_data, private_data_len);
cm_reset_to_idle(cm_id_priv);
break;
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
ret = cm_alloc_msg(cm_id_priv, &msg);
if (!ret)
cm_format_rej((struct cm_rej_msg *) msg->mad,
cm_id_priv, reason, ari, ari_length,
private_data, private_data_len);
cm_enter_timewait(cm_id_priv);
break;
default:
ret = -EINVAL;
goto out;
}
if (ret)
goto out;
ret = ib_post_send_mad(msg, NULL);
if (ret)
cm_free_msg(msg);
out: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_rej);
static void cm_format_rej_event(struct cm_work *work)
{
struct cm_rej_msg *rej_msg;
struct ib_cm_rej_event_param *param;
rej_msg = (struct cm_rej_msg *)work->mad_recv_wc->recv_buf.mad;
param = &work->cm_event.param.rej_rcvd;
param->ari = rej_msg->ari;
param->ari_length = cm_rej_get_reject_info_len(rej_msg);
param->reason = __be16_to_cpu(rej_msg->reason);
work->cm_event.private_data = &rej_msg->private_data;
}
static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
{
struct cm_timewait_info *timewait_info;
struct cm_id_private *cm_id_priv;
__be32 remote_id;
remote_id = rej_msg->local_comm_id;
if (__be16_to_cpu(rej_msg->reason) == IB_CM_REJ_TIMEOUT) {
spin_lock_irq(&cm.lock);
timewait_info = cm_find_remote_id( *((__be64 *) rej_msg->ari),
remote_id);
if (!timewait_info) {
spin_unlock_irq(&cm.lock);
return NULL;
}
cm_id_priv = idr_find(&cm.local_id_table, (__force int)
(timewait_info->work.local_id ^
cm.random_id_operand));
if (cm_id_priv) {
if (cm_id_priv->id.remote_id == remote_id)
atomic_inc(&cm_id_priv->refcount);
else
cm_id_priv = NULL;
}
spin_unlock_irq(&cm.lock);
} else if (cm_rej_get_msg_rejected(rej_msg) == CM_MSG_RESPONSE_REQ)
cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, 0);
else
cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, remote_id);
return cm_id_priv;
}
static int cm_rej_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_rej_msg *rej_msg;
int ret;
rej_msg = (struct cm_rej_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_rejected_id(rej_msg);
if (!cm_id_priv)
return -EINVAL;
cm_format_rej_event(work);
spin_lock_irq(&cm_id_priv->lock);
switch (cm_id_priv->id.state) {
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
/* fall through */
case IB_CM_REQ_RCVD:
case IB_CM_MRA_REQ_SENT:
if (__be16_to_cpu(rej_msg->reason) == IB_CM_REJ_STALE_CONN)
cm_enter_timewait(cm_id_priv);
else
cm_reset_to_idle(cm_id_priv);
break;
case IB_CM_DREQ_SENT:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
/* fall through */
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
cm_enter_timewait(cm_id_priv);
break;
case IB_CM_ESTABLISHED:
if (cm_id_priv->id.lap_state == IB_CM_LAP_UNINIT ||
cm_id_priv->id.lap_state == IB_CM_LAP_SENT) {
if (cm_id_priv->id.lap_state == IB_CM_LAP_SENT)
ib_cancel_mad(cm_id_priv->av.port->mad_agent,
cm_id_priv->msg);
cm_enter_timewait(cm_id_priv);
break;
}
/* fall through */
default:
spin_unlock_irq(&cm_id_priv->lock);
ret = -EINVAL;
goto out;
}
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
int ib_send_cm_mra(struct ib_cm_id *cm_id,
u8 service_timeout,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
enum ib_cm_state cm_state;
enum ib_cm_lap_state lap_state;
enum cm_msg_response msg_response;
void *data;
unsigned long flags;
int ret;
if (private_data && private_data_len > IB_CM_MRA_PRIVATE_DATA_SIZE)
return -EINVAL;
data = cm_copy_private_data(private_data, private_data_len);
if (IS_ERR(data))
return PTR_ERR(data);
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
switch(cm_id_priv->id.state) {
case IB_CM_REQ_RCVD:
cm_state = IB_CM_MRA_REQ_SENT;
lap_state = cm_id->lap_state;
msg_response = CM_MSG_RESPONSE_REQ;
break;
case IB_CM_REP_RCVD:
cm_state = IB_CM_MRA_REP_SENT;
lap_state = cm_id->lap_state;
msg_response = CM_MSG_RESPONSE_REP;
break;
case IB_CM_ESTABLISHED:
if (cm_id->lap_state == IB_CM_LAP_RCVD) {
cm_state = cm_id->state;
lap_state = IB_CM_MRA_LAP_SENT;
msg_response = CM_MSG_RESPONSE_OTHER;
break;
}
/* fall through */
default:
ret = -EINVAL;
goto error1;
}
if (!(service_timeout & IB_CM_MRA_FLAG_DELAY)) {
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto error1;
cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
msg_response, service_timeout,
private_data, private_data_len);
ret = ib_post_send_mad(msg, NULL);
if (ret)
goto error2;
}
cm_id->state = cm_state;
cm_id->lap_state = lap_state;
cm_id_priv->service_timeout = service_timeout;
cm_set_private_data(cm_id_priv, data, private_data_len);
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return 0;
error1: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
kfree(data);
return ret;
error2: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
kfree(data);
cm_free_msg(msg);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_mra);
static struct cm_id_private * cm_acquire_mraed_id(struct cm_mra_msg *mra_msg)
{
switch (cm_mra_get_msg_mraed(mra_msg)) {
case CM_MSG_RESPONSE_REQ:
return cm_acquire_id(mra_msg->remote_comm_id, 0);
case CM_MSG_RESPONSE_REP:
case CM_MSG_RESPONSE_OTHER:
return cm_acquire_id(mra_msg->remote_comm_id,
mra_msg->local_comm_id);
default:
return NULL;
}
}
static int cm_mra_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_mra_msg *mra_msg;
int timeout, ret;
mra_msg = (struct cm_mra_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_mraed_id(mra_msg);
if (!cm_id_priv)
return -EINVAL;
work->cm_event.private_data = &mra_msg->private_data;
work->cm_event.param.mra_rcvd.service_timeout =
cm_mra_get_service_timeout(mra_msg);
timeout = cm_convert_to_ms(cm_mra_get_service_timeout(mra_msg)) +
cm_convert_to_ms(cm_id_priv->av.timeout);
spin_lock_irq(&cm_id_priv->lock);
switch (cm_id_priv->id.state) {
case IB_CM_REQ_SENT:
if (cm_mra_get_msg_mraed(mra_msg) != CM_MSG_RESPONSE_REQ ||
ib_modify_mad(cm_id_priv->av.port->mad_agent,
cm_id_priv->msg, timeout))
goto out;
cm_id_priv->id.state = IB_CM_MRA_REQ_RCVD;
break;
case IB_CM_REP_SENT:
if (cm_mra_get_msg_mraed(mra_msg) != CM_MSG_RESPONSE_REP ||
ib_modify_mad(cm_id_priv->av.port->mad_agent,
cm_id_priv->msg, timeout))
goto out;
cm_id_priv->id.state = IB_CM_MRA_REP_RCVD;
break;
case IB_CM_ESTABLISHED:
if (cm_mra_get_msg_mraed(mra_msg) != CM_MSG_RESPONSE_OTHER ||
cm_id_priv->id.lap_state != IB_CM_LAP_SENT ||
ib_modify_mad(cm_id_priv->av.port->mad_agent,
cm_id_priv->msg, timeout)) {
if (cm_id_priv->id.lap_state == IB_CM_MRA_LAP_RCVD)
atomic_long_inc(&work->port->
counter_group[CM_RECV_DUPLICATES].
counter[CM_MRA_COUNTER]);
goto out;
}
cm_id_priv->id.lap_state = IB_CM_MRA_LAP_RCVD;
break;
case IB_CM_MRA_REQ_RCVD:
case IB_CM_MRA_REP_RCVD:
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_MRA_COUNTER]);
/* fall through */
default:
goto out;
}
cm_id_priv->msg->context[1] = (void *) (unsigned long)
cm_id_priv->id.state;
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
spin_unlock_irq(&cm_id_priv->lock);
cm_deref_id(cm_id_priv);
return -EINVAL;
}
static void cm_format_lap(struct cm_lap_msg *lap_msg,
struct cm_id_private *cm_id_priv,
struct sa_path_rec *alternate_path,
const void *private_data,
u8 private_data_len)
{
bool alt_ext = false;
if (alternate_path->rec_type == SA_PATH_REC_TYPE_OPA)
alt_ext = opa_is_extended_lid(alternate_path->opa.dlid,
alternate_path->opa.slid);
cm_format_mad_hdr(&lap_msg->hdr, CM_LAP_ATTR_ID,
cm_form_tid(cm_id_priv, CM_MSG_SEQUENCE_LAP));
lap_msg->local_comm_id = cm_id_priv->id.local_id;
lap_msg->remote_comm_id = cm_id_priv->id.remote_id;
cm_lap_set_remote_qpn(lap_msg, cm_id_priv->remote_qpn);
/* todo: need remote CM response timeout */
cm_lap_set_remote_resp_timeout(lap_msg, 0x1F);
lap_msg->alt_local_lid =
htons(ntohl(sa_path_get_slid(alternate_path)));
lap_msg->alt_remote_lid =
htons(ntohl(sa_path_get_dlid(alternate_path)));
lap_msg->alt_local_gid = alternate_path->sgid;
lap_msg->alt_remote_gid = alternate_path->dgid;
if (alt_ext) {
lap_msg->alt_local_gid.global.interface_id
= OPA_MAKE_ID(be32_to_cpu(alternate_path->opa.slid));
lap_msg->alt_remote_gid.global.interface_id
= OPA_MAKE_ID(be32_to_cpu(alternate_path->opa.dlid));
}
cm_lap_set_flow_label(lap_msg, alternate_path->flow_label);
cm_lap_set_traffic_class(lap_msg, alternate_path->traffic_class);
lap_msg->alt_hop_limit = alternate_path->hop_limit;
cm_lap_set_packet_rate(lap_msg, alternate_path->rate);
cm_lap_set_sl(lap_msg, alternate_path->sl);
cm_lap_set_subnet_local(lap_msg, 1); /* local only... */
cm_lap_set_local_ack_timeout(lap_msg,
cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay,
alternate_path->packet_life_time));
if (private_data && private_data_len)
memcpy(lap_msg->private_data, private_data, private_data_len);
}
int ib_send_cm_lap(struct ib_cm_id *cm_id,
struct sa_path_rec *alternate_path,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
int ret;
if (private_data && private_data_len > IB_CM_LAP_PRIVATE_DATA_SIZE)
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_ESTABLISHED ||
(cm_id->lap_state != IB_CM_LAP_UNINIT &&
cm_id->lap_state != IB_CM_LAP_IDLE)) {
ret = -EINVAL;
goto out;
}
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
ret = cm_init_av_by_path(alternate_path, &cm_id_priv->alt_av,
cm_id_priv);
if (ret)
goto out;
cm_id_priv->alt_av.timeout =
cm_ack_timeout(cm_id_priv->target_ack_delay,
cm_id_priv->alt_av.timeout - 1);
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto out;
cm_format_lap((struct cm_lap_msg *) msg->mad, cm_id_priv,
alternate_path, private_data, private_data_len);
msg->timeout_ms = cm_id_priv->timeout_ms;
msg->context[1] = (void *) (unsigned long) IB_CM_ESTABLISHED;
ret = ib_post_send_mad(msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
return ret;
}
cm_id->lap_state = IB_CM_LAP_SENT;
cm_id_priv->msg = msg;
out: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_lap);
static void cm_format_path_lid_from_lap(struct cm_lap_msg *lap_msg,
struct sa_path_rec *path)
{
u32 lid;
if (path->rec_type != SA_PATH_REC_TYPE_OPA) {
sa_path_set_dlid(path, ntohs(lap_msg->alt_local_lid));
sa_path_set_slid(path, ntohs(lap_msg->alt_remote_lid));
} else {
lid = opa_get_lid_from_gid(&lap_msg->alt_local_gid);
sa_path_set_dlid(path, lid);
lid = opa_get_lid_from_gid(&lap_msg->alt_remote_gid);
sa_path_set_slid(path, lid);
}
}
static void cm_format_path_from_lap(struct cm_id_private *cm_id_priv,
struct sa_path_rec *path,
struct cm_lap_msg *lap_msg)
{
path->dgid = lap_msg->alt_local_gid;
path->sgid = lap_msg->alt_remote_gid;
path->flow_label = cm_lap_get_flow_label(lap_msg);
path->hop_limit = lap_msg->alt_hop_limit;
path->traffic_class = cm_lap_get_traffic_class(lap_msg);
path->reversible = 1;
path->pkey = cm_id_priv->pkey;
path->sl = cm_lap_get_sl(lap_msg);
path->mtu_selector = IB_SA_EQ;
path->mtu = cm_id_priv->path_mtu;
path->rate_selector = IB_SA_EQ;
path->rate = cm_lap_get_packet_rate(lap_msg);
path->packet_life_time_selector = IB_SA_EQ;
path->packet_life_time = cm_lap_get_local_ack_timeout(lap_msg);
path->packet_life_time -= (path->packet_life_time > 0);
cm_format_path_lid_from_lap(lap_msg, path);
}
static int cm_lap_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_lap_msg *lap_msg;
struct ib_cm_lap_event_param *param;
struct ib_mad_send_buf *msg = NULL;
int ret;
/* todo: verify LAP request and send reject APR if invalid. */
lap_msg = (struct cm_lap_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(lap_msg->remote_comm_id,
lap_msg->local_comm_id);
if (!cm_id_priv)
return -EINVAL;
param = &work->cm_event.param.lap_rcvd;
memset(&work->path[0], 0, sizeof(work->path[1]));
cm_path_set_rec_type(work->port->cm_dev->ib_device,
work->port->port_num,
&work->path[0],
&lap_msg->alt_local_gid);
param->alternate_path = &work->path[0];
cm_format_path_from_lap(cm_id_priv, param->alternate_path, lap_msg);
work->cm_event.private_data = &lap_msg->private_data;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_ESTABLISHED)
goto unlock;
switch (cm_id_priv->id.lap_state) {
case IB_CM_LAP_UNINIT:
case IB_CM_LAP_IDLE:
break;
case IB_CM_MRA_LAP_SENT:
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_LAP_COUNTER]);
msg = cm_alloc_response_msg_no_ah(work->port, work->mad_recv_wc);
if (IS_ERR(msg))
goto unlock;
cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
CM_MSG_RESPONSE_OTHER,
cm_id_priv->service_timeout,
cm_id_priv->private_data,
cm_id_priv->private_data_len);
spin_unlock_irq(&cm_id_priv->lock);
if (cm_create_response_msg_ah(work->port, work->mad_recv_wc, msg) ||
ib_post_send_mad(msg, NULL))
cm_free_msg(msg);
goto deref;
case IB_CM_LAP_RCVD:
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_LAP_COUNTER]);
goto unlock;
default:
goto unlock;
}
cm_id_priv->id.lap_state = IB_CM_LAP_RCVD;
cm_id_priv->tid = lap_msg->hdr.tid;
cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
work->mad_recv_wc->recv_buf.grh,
&cm_id_priv->av);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
cm_init_av_by_path(param->alternate_path, &cm_id_priv->alt_av,
cm_id_priv);
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
unlock: spin_unlock_irq(&cm_id_priv->lock);
deref: cm_deref_id(cm_id_priv);
return -EINVAL;
}
static void cm_format_apr(struct cm_apr_msg *apr_msg,
struct cm_id_private *cm_id_priv,
enum ib_cm_apr_status status,
void *info,
u8 info_length,
const void *private_data,
u8 private_data_len)
{
cm_format_mad_hdr(&apr_msg->hdr, CM_APR_ATTR_ID, cm_id_priv->tid);
apr_msg->local_comm_id = cm_id_priv->id.local_id;
apr_msg->remote_comm_id = cm_id_priv->id.remote_id;
apr_msg->ap_status = (u8) status;
if (info && info_length) {
apr_msg->info_length = info_length;
memcpy(apr_msg->info, info, info_length);
}
if (private_data && private_data_len)
memcpy(apr_msg->private_data, private_data, private_data_len);
}
int ib_send_cm_apr(struct ib_cm_id *cm_id,
enum ib_cm_apr_status status,
void *info,
u8 info_length,
const void *private_data,
u8 private_data_len)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
int ret;
if ((private_data && private_data_len > IB_CM_APR_PRIVATE_DATA_SIZE) ||
(info && info_length > IB_CM_APR_INFO_LENGTH))
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_ESTABLISHED ||
(cm_id->lap_state != IB_CM_LAP_RCVD &&
cm_id->lap_state != IB_CM_MRA_LAP_SENT)) {
ret = -EINVAL;
goto out;
}
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto out;
cm_format_apr((struct cm_apr_msg *) msg->mad, cm_id_priv, status,
info, info_length, private_data, private_data_len);
ret = ib_post_send_mad(msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
return ret;
}
cm_id->lap_state = IB_CM_LAP_IDLE;
out: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_apr);
static int cm_apr_handler(struct cm_work *work)
{
struct cm_id_private *cm_id_priv;
struct cm_apr_msg *apr_msg;
int ret;
apr_msg = (struct cm_apr_msg *)work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(apr_msg->remote_comm_id,
apr_msg->local_comm_id);
if (!cm_id_priv)
return -EINVAL; /* Unmatched reply. */
work->cm_event.param.apr_rcvd.ap_status = apr_msg->ap_status;
work->cm_event.param.apr_rcvd.apr_info = &apr_msg->info;
work->cm_event.param.apr_rcvd.info_len = apr_msg->info_length;
work->cm_event.private_data = &apr_msg->private_data;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_ESTABLISHED ||
(cm_id_priv->id.lap_state != IB_CM_LAP_SENT &&
cm_id_priv->id.lap_state != IB_CM_MRA_LAP_RCVD)) {
spin_unlock_irq(&cm_id_priv->lock);
goto out;
}
cm_id_priv->id.lap_state = IB_CM_LAP_IDLE;
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
cm_id_priv->msg = NULL;
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
static int cm_timewait_handler(struct cm_work *work)
{
struct cm_timewait_info *timewait_info;
struct cm_id_private *cm_id_priv;
int ret;
timewait_info = (struct cm_timewait_info *)work;
spin_lock_irq(&cm.lock);
list_del(&timewait_info->list);
spin_unlock_irq(&cm.lock);
cm_id_priv = cm_acquire_id(timewait_info->work.local_id,
timewait_info->work.remote_id);
if (!cm_id_priv)
return -EINVAL;
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_TIMEWAIT ||
cm_id_priv->remote_qpn != timewait_info->remote_qpn) {
spin_unlock_irq(&cm_id_priv->lock);
goto out;
}
cm_id_priv->id.state = IB_CM_IDLE;
ret = atomic_inc_and_test(&cm_id_priv->work_count);
if (!ret)
list_add_tail(&work->list, &cm_id_priv->work_list);
spin_unlock_irq(&cm_id_priv->lock);
if (ret)
cm_process_work(cm_id_priv, work);
else
cm_deref_id(cm_id_priv);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
static void cm_format_sidr_req(struct cm_sidr_req_msg *sidr_req_msg,
struct cm_id_private *cm_id_priv,
struct ib_cm_sidr_req_param *param)
{
cm_format_mad_hdr(&sidr_req_msg->hdr, CM_SIDR_REQ_ATTR_ID,
cm_form_tid(cm_id_priv, CM_MSG_SEQUENCE_SIDR));
sidr_req_msg->request_id = cm_id_priv->id.local_id;
sidr_req_msg->pkey = param->path->pkey;
sidr_req_msg->service_id = param->service_id;
if (param->private_data && param->private_data_len)
memcpy(sidr_req_msg->private_data, param->private_data,
param->private_data_len);
}
int ib_send_cm_sidr_req(struct ib_cm_id *cm_id,
struct ib_cm_sidr_req_param *param)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
int ret;
if (!param->path || (param->private_data &&
param->private_data_len > IB_CM_SIDR_REQ_PRIVATE_DATA_SIZE))
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
ret = cm_init_av_by_path(param->path, &cm_id_priv->av, cm_id_priv);
if (ret)
goto out;
cm_id->service_id = param->service_id;
cm_id->service_mask = ~cpu_to_be64(0);
cm_id_priv->timeout_ms = param->timeout_ms;
cm_id_priv->max_cm_retries = param->max_cm_retries;
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto out;
cm_format_sidr_req((struct cm_sidr_req_msg *) msg->mad, cm_id_priv,
param);
msg->timeout_ms = cm_id_priv->timeout_ms;
msg->context[1] = (void *) (unsigned long) IB_CM_SIDR_REQ_SENT;
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state == IB_CM_IDLE)
ret = ib_post_send_mad(msg, NULL);
else
ret = -EINVAL;
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
goto out;
}
cm_id->state = IB_CM_SIDR_REQ_SENT;
cm_id_priv->msg = msg;
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
out:
return ret;
}
EXPORT_SYMBOL(ib_send_cm_sidr_req);
static void cm_format_sidr_req_event(struct cm_work *work,
struct ib_cm_id *listen_id)
{
struct cm_sidr_req_msg *sidr_req_msg;
struct ib_cm_sidr_req_event_param *param;
sidr_req_msg = (struct cm_sidr_req_msg *)
work->mad_recv_wc->recv_buf.mad;
param = &work->cm_event.param.sidr_req_rcvd;
param->pkey = __be16_to_cpu(sidr_req_msg->pkey);
param->listen_id = listen_id;
param->service_id = sidr_req_msg->service_id;
param->bth_pkey = cm_get_bth_pkey(work);
param->port = work->port->port_num;
work->cm_event.private_data = &sidr_req_msg->private_data;
}
static int cm_sidr_req_handler(struct cm_work *work)
{
struct ib_cm_id *cm_id;
struct cm_id_private *cm_id_priv, *cur_cm_id_priv;
struct cm_sidr_req_msg *sidr_req_msg;
struct ib_wc *wc;
cm_id = ib_create_cm_id(work->port->cm_dev->ib_device, NULL, NULL);
if (IS_ERR(cm_id))
return PTR_ERR(cm_id);
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
/* Record SGID/SLID and request ID for lookup. */
sidr_req_msg = (struct cm_sidr_req_msg *)
work->mad_recv_wc->recv_buf.mad;
wc = work->mad_recv_wc->wc;
cm_id_priv->av.dgid.global.subnet_prefix = cpu_to_be64(wc->slid);
cm_id_priv->av.dgid.global.interface_id = 0;
cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
work->mad_recv_wc->recv_buf.grh,
&cm_id_priv->av);
cm_id_priv->id.remote_id = sidr_req_msg->request_id;
cm_id_priv->tid = sidr_req_msg->hdr.tid;
atomic_inc(&cm_id_priv->work_count);
spin_lock_irq(&cm.lock);
cur_cm_id_priv = cm_insert_remote_sidr(cm_id_priv);
if (cur_cm_id_priv) {
spin_unlock_irq(&cm.lock);
atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
counter[CM_SIDR_REQ_COUNTER]);
goto out; /* Duplicate message. */
}
cm_id_priv->id.state = IB_CM_SIDR_REQ_RCVD;
cur_cm_id_priv = cm_find_listen(cm_id->device,
sidr_req_msg->service_id);
if (!cur_cm_id_priv) {
spin_unlock_irq(&cm.lock);
cm_reject_sidr_req(cm_id_priv, IB_SIDR_UNSUPPORTED);
goto out; /* No match. */
}
atomic_inc(&cur_cm_id_priv->refcount);
atomic_inc(&cm_id_priv->refcount);
spin_unlock_irq(&cm.lock);
cm_id_priv->id.cm_handler = cur_cm_id_priv->id.cm_handler;
cm_id_priv->id.context = cur_cm_id_priv->id.context;
cm_id_priv->id.service_id = sidr_req_msg->service_id;
cm_id_priv->id.service_mask = ~cpu_to_be64(0);
cm_format_sidr_req_event(work, &cur_cm_id_priv->id);
cm_process_work(cm_id_priv, work);
cm_deref_id(cur_cm_id_priv);
return 0;
out:
ib_destroy_cm_id(&cm_id_priv->id);
return -EINVAL;
}
static void cm_format_sidr_rep(struct cm_sidr_rep_msg *sidr_rep_msg,
struct cm_id_private *cm_id_priv,
struct ib_cm_sidr_rep_param *param)
{
cm_format_mad_hdr(&sidr_rep_msg->hdr, CM_SIDR_REP_ATTR_ID,
cm_id_priv->tid);
sidr_rep_msg->request_id = cm_id_priv->id.remote_id;
sidr_rep_msg->status = param->status;
cm_sidr_rep_set_qpn(sidr_rep_msg, cpu_to_be32(param->qp_num));
sidr_rep_msg->service_id = cm_id_priv->id.service_id;
sidr_rep_msg->qkey = cpu_to_be32(param->qkey);
if (param->info && param->info_length)
memcpy(sidr_rep_msg->info, param->info, param->info_length);
if (param->private_data && param->private_data_len)
memcpy(sidr_rep_msg->private_data, param->private_data,
param->private_data_len);
}
int ib_send_cm_sidr_rep(struct ib_cm_id *cm_id,
struct ib_cm_sidr_rep_param *param)
{
struct cm_id_private *cm_id_priv;
struct ib_mad_send_buf *msg;
unsigned long flags;
int ret;
if ((param->info && param->info_length > IB_CM_SIDR_REP_INFO_LENGTH) ||
(param->private_data &&
param->private_data_len > IB_CM_SIDR_REP_PRIVATE_DATA_SIZE))
return -EINVAL;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state != IB_CM_SIDR_REQ_RCVD) {
ret = -EINVAL;
goto error;
}
ret = cm_alloc_msg(cm_id_priv, &msg);
if (ret)
goto error;
cm_format_sidr_rep((struct cm_sidr_rep_msg *) msg->mad, cm_id_priv,
param);
ret = ib_post_send_mad(msg, NULL);
if (ret) {
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
cm_free_msg(msg);
return ret;
}
cm_id->state = IB_CM_IDLE;
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
spin_lock_irqsave(&cm.lock, flags);
if (!RB_EMPTY_NODE(&cm_id_priv->sidr_id_node)) {
rb_erase(&cm_id_priv->sidr_id_node, &cm.remote_sidr_table);
RB_CLEAR_NODE(&cm_id_priv->sidr_id_node);
}
spin_unlock_irqrestore(&cm.lock, flags);
return 0;
error: spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
EXPORT_SYMBOL(ib_send_cm_sidr_rep);
static void cm_format_sidr_rep_event(struct cm_work *work)
{
struct cm_sidr_rep_msg *sidr_rep_msg;
struct ib_cm_sidr_rep_event_param *param;
sidr_rep_msg = (struct cm_sidr_rep_msg *)
work->mad_recv_wc->recv_buf.mad;
param = &work->cm_event.param.sidr_rep_rcvd;
param->status = sidr_rep_msg->status;
param->qkey = be32_to_cpu(sidr_rep_msg->qkey);
param->qpn = be32_to_cpu(cm_sidr_rep_get_qpn(sidr_rep_msg));
param->info = &sidr_rep_msg->info;
param->info_len = sidr_rep_msg->info_length;
work->cm_event.private_data = &sidr_rep_msg->private_data;
}
static int cm_sidr_rep_handler(struct cm_work *work)
{
struct cm_sidr_rep_msg *sidr_rep_msg;
struct cm_id_private *cm_id_priv;
sidr_rep_msg = (struct cm_sidr_rep_msg *)
work->mad_recv_wc->recv_buf.mad;
cm_id_priv = cm_acquire_id(sidr_rep_msg->request_id, 0);
if (!cm_id_priv)
return -EINVAL; /* Unmatched reply. */
spin_lock_irq(&cm_id_priv->lock);
if (cm_id_priv->id.state != IB_CM_SIDR_REQ_SENT) {
spin_unlock_irq(&cm_id_priv->lock);
goto out;
}
cm_id_priv->id.state = IB_CM_IDLE;
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
spin_unlock_irq(&cm_id_priv->lock);
cm_format_sidr_rep_event(work);
cm_process_work(cm_id_priv, work);
return 0;
out:
cm_deref_id(cm_id_priv);
return -EINVAL;
}
static void cm_process_send_error(struct ib_mad_send_buf *msg,
enum ib_wc_status wc_status)
{
struct cm_id_private *cm_id_priv;
struct ib_cm_event cm_event;
enum ib_cm_state state;
int ret;
memset(&cm_event, 0, sizeof cm_event);
cm_id_priv = msg->context[0];
/* Discard old sends or ones without a response. */
spin_lock_irq(&cm_id_priv->lock);
state = (enum ib_cm_state) (unsigned long) msg->context[1];
if (msg != cm_id_priv->msg || state != cm_id_priv->id.state)
goto discard;
pr_debug_ratelimited("CM: failed sending MAD in state %d. (%s)\n",
state, ib_wc_status_msg(wc_status));
switch (state) {
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
cm_reset_to_idle(cm_id_priv);
cm_event.event = IB_CM_REQ_ERROR;
break;
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
cm_reset_to_idle(cm_id_priv);
cm_event.event = IB_CM_REP_ERROR;
break;
case IB_CM_DREQ_SENT:
cm_enter_timewait(cm_id_priv);
cm_event.event = IB_CM_DREQ_ERROR;
break;
case IB_CM_SIDR_REQ_SENT:
cm_id_priv->id.state = IB_CM_IDLE;
cm_event.event = IB_CM_SIDR_REQ_ERROR;
break;
default:
goto discard;
}
spin_unlock_irq(&cm_id_priv->lock);
cm_event.param.send_status = wc_status;
/* No other events can occur on the cm_id at this point. */
ret = cm_id_priv->id.cm_handler(&cm_id_priv->id, &cm_event);
cm_free_msg(msg);
if (ret)
ib_destroy_cm_id(&cm_id_priv->id);
return;
discard:
spin_unlock_irq(&cm_id_priv->lock);
cm_free_msg(msg);
}
static void cm_send_handler(struct ib_mad_agent *mad_agent,
struct ib_mad_send_wc *mad_send_wc)
{
struct ib_mad_send_buf *msg = mad_send_wc->send_buf;
struct cm_port *port;
u16 attr_index;
port = mad_agent->context;
attr_index = be16_to_cpu(((struct ib_mad_hdr *)
msg->mad)->attr_id) - CM_ATTR_ID_OFFSET;
/*
* If the send was in response to a received message (context[0] is not
* set to a cm_id), and is not a REJ, then it is a send that was
* manually retried.
*/
if (!msg->context[0] && (attr_index != CM_REJ_COUNTER))
msg->retries = 1;
atomic_long_add(1 + msg->retries,
&port->counter_group[CM_XMIT].counter[attr_index]);
if (msg->retries)
atomic_long_add(msg->retries,
&port->counter_group[CM_XMIT_RETRIES].
counter[attr_index]);
switch (mad_send_wc->status) {
case IB_WC_SUCCESS:
case IB_WC_WR_FLUSH_ERR:
cm_free_msg(msg);
break;
default:
if (msg->context[0] && msg->context[1])
cm_process_send_error(msg, mad_send_wc->status);
else
cm_free_msg(msg);
break;
}
}
static void cm_work_handler(struct work_struct *_work)
{
struct cm_work *work = container_of(_work, struct cm_work, work.work);
int ret;
switch (work->cm_event.event) {
case IB_CM_REQ_RECEIVED:
ret = cm_req_handler(work);
break;
case IB_CM_MRA_RECEIVED:
ret = cm_mra_handler(work);
break;
case IB_CM_REJ_RECEIVED:
ret = cm_rej_handler(work);
break;
case IB_CM_REP_RECEIVED:
ret = cm_rep_handler(work);
break;
case IB_CM_RTU_RECEIVED:
ret = cm_rtu_handler(work);
break;
case IB_CM_USER_ESTABLISHED:
ret = cm_establish_handler(work);
break;
case IB_CM_DREQ_RECEIVED:
ret = cm_dreq_handler(work);
break;
case IB_CM_DREP_RECEIVED:
ret = cm_drep_handler(work);
break;
case IB_CM_SIDR_REQ_RECEIVED:
ret = cm_sidr_req_handler(work);
break;
case IB_CM_SIDR_REP_RECEIVED:
ret = cm_sidr_rep_handler(work);
break;
case IB_CM_LAP_RECEIVED:
ret = cm_lap_handler(work);
break;
case IB_CM_APR_RECEIVED:
ret = cm_apr_handler(work);
break;
case IB_CM_TIMEWAIT_EXIT:
ret = cm_timewait_handler(work);
break;
default:
ret = -EINVAL;
break;
}
if (ret)
cm_free_work(work);
}
static int cm_establish(struct ib_cm_id *cm_id)
{
struct cm_id_private *cm_id_priv;
struct cm_work *work;
unsigned long flags;
int ret = 0;
struct cm_device *cm_dev;
cm_dev = ib_get_client_data(cm_id->device, &cm_client);
if (!cm_dev)
return -ENODEV;
work = kmalloc(sizeof *work, GFP_ATOMIC);
if (!work)
return -ENOMEM;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
switch (cm_id->state)
{
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
cm_id->state = IB_CM_ESTABLISHED;
break;
case IB_CM_ESTABLISHED:
ret = -EISCONN;
break;
default:
ret = -EINVAL;
break;
}
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
if (ret) {
kfree(work);
goto out;
}
/*
* The CM worker thread may try to destroy the cm_id before it
* can execute this work item. To prevent potential deadlock,
* we need to find the cm_id once we're in the context of the
* worker thread, rather than holding a reference on it.
*/
INIT_DELAYED_WORK(&work->work, cm_work_handler);
work->local_id = cm_id->local_id;
work->remote_id = cm_id->remote_id;
work->mad_recv_wc = NULL;
work->cm_event.event = IB_CM_USER_ESTABLISHED;
/* Check if the device started its remove_one */
IB/cm: Fix a recently introduced locking bug ib_cm_notify() can be called from interrupt context. Hence do not reenable interrupts unconditionally in cm_establish(). This patch avoids that lockdep reports the following warning: WARNING: CPU: 0 PID: 23317 at kernel/locking/lockdep.c:2624 trace _hardirqs_on_caller+0x112/0x1b0 DEBUG_LOCKS_WARN_ON(current->hardirq_context) Call Trace: <IRQ> [<ffffffff812bd0e5>] dump_stack+0x67/0x92 [<ffffffff81056f21>] __warn+0xc1/0xe0 [<ffffffff81056f8a>] warn_slowpath_fmt+0x4a/0x50 [<ffffffff810a5932>] trace_hardirqs_on_caller+0x112/0x1b0 [<ffffffff810a59dd>] trace_hardirqs_on+0xd/0x10 [<ffffffff815992c7>] _raw_spin_unlock_irq+0x27/0x40 [<ffffffffa0382e9c>] ib_cm_notify+0x25c/0x290 [ib_cm] [<ffffffffa068fbc1>] srpt_qp_event+0xa1/0xf0 [ib_srpt] [<ffffffffa04efb97>] mlx4_ib_qp_event+0x67/0xd0 [mlx4_ib] [<ffffffffa034ec0a>] mlx4_qp_event+0x5a/0xc0 [mlx4_core] [<ffffffffa03365f8>] mlx4_eq_int+0x3d8/0xcf0 [mlx4_core] [<ffffffffa0336f9c>] mlx4_msi_x_interrupt+0xc/0x20 [mlx4_core] [<ffffffff810b0914>] handle_irq_event_percpu+0x64/0x100 [<ffffffff810b09e4>] handle_irq_event+0x34/0x60 [<ffffffff810b3a6a>] handle_edge_irq+0x6a/0x150 [<ffffffff8101ad05>] handle_irq+0x15/0x20 [<ffffffff8101a66c>] do_IRQ+0x5c/0x110 [<ffffffff8159a2c9>] common_interrupt+0x89/0x89 [<ffffffff81297a17>] blk_run_queue_async+0x37/0x40 [<ffffffffa0163e53>] rq_completed+0x43/0x70 [dm_mod] [<ffffffffa0164896>] dm_softirq_done+0x176/0x280 [dm_mod] [<ffffffff812a26c2>] blk_done_softirq+0x52/0x90 [<ffffffff8105bc1f>] __do_softirq+0x10f/0x230 [<ffffffff8105bec8>] irq_exit+0xa8/0xb0 [<ffffffff8103653e>] smp_trace_call_function_single_interrupt+0x2e/0x30 [<ffffffff81036549>] smp_call_function_single_interrupt+0x9/0x10 [<ffffffff8159a959>] call_function_single_interrupt+0x89/0x90 <EOI> Fixes: commit be4b499323bf (IB/cm: Do not queue work to a device that's going away) Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Erez Shitrit <erezsh@mellanox.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Nikolay Borisov <kernel@kyup.com> Cc: stable <stable@vger.kernel.org> # v4.2+ Acked-by: Erez Shitrit <erezsh@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-25 23:33:16 +08:00
spin_lock_irqsave(&cm.lock, flags);
if (!cm_dev->going_down) {
queue_delayed_work(cm.wq, &work->work, 0);
} else {
kfree(work);
ret = -ENODEV;
}
IB/cm: Fix a recently introduced locking bug ib_cm_notify() can be called from interrupt context. Hence do not reenable interrupts unconditionally in cm_establish(). This patch avoids that lockdep reports the following warning: WARNING: CPU: 0 PID: 23317 at kernel/locking/lockdep.c:2624 trace _hardirqs_on_caller+0x112/0x1b0 DEBUG_LOCKS_WARN_ON(current->hardirq_context) Call Trace: <IRQ> [<ffffffff812bd0e5>] dump_stack+0x67/0x92 [<ffffffff81056f21>] __warn+0xc1/0xe0 [<ffffffff81056f8a>] warn_slowpath_fmt+0x4a/0x50 [<ffffffff810a5932>] trace_hardirqs_on_caller+0x112/0x1b0 [<ffffffff810a59dd>] trace_hardirqs_on+0xd/0x10 [<ffffffff815992c7>] _raw_spin_unlock_irq+0x27/0x40 [<ffffffffa0382e9c>] ib_cm_notify+0x25c/0x290 [ib_cm] [<ffffffffa068fbc1>] srpt_qp_event+0xa1/0xf0 [ib_srpt] [<ffffffffa04efb97>] mlx4_ib_qp_event+0x67/0xd0 [mlx4_ib] [<ffffffffa034ec0a>] mlx4_qp_event+0x5a/0xc0 [mlx4_core] [<ffffffffa03365f8>] mlx4_eq_int+0x3d8/0xcf0 [mlx4_core] [<ffffffffa0336f9c>] mlx4_msi_x_interrupt+0xc/0x20 [mlx4_core] [<ffffffff810b0914>] handle_irq_event_percpu+0x64/0x100 [<ffffffff810b09e4>] handle_irq_event+0x34/0x60 [<ffffffff810b3a6a>] handle_edge_irq+0x6a/0x150 [<ffffffff8101ad05>] handle_irq+0x15/0x20 [<ffffffff8101a66c>] do_IRQ+0x5c/0x110 [<ffffffff8159a2c9>] common_interrupt+0x89/0x89 [<ffffffff81297a17>] blk_run_queue_async+0x37/0x40 [<ffffffffa0163e53>] rq_completed+0x43/0x70 [dm_mod] [<ffffffffa0164896>] dm_softirq_done+0x176/0x280 [dm_mod] [<ffffffff812a26c2>] blk_done_softirq+0x52/0x90 [<ffffffff8105bc1f>] __do_softirq+0x10f/0x230 [<ffffffff8105bec8>] irq_exit+0xa8/0xb0 [<ffffffff8103653e>] smp_trace_call_function_single_interrupt+0x2e/0x30 [<ffffffff81036549>] smp_call_function_single_interrupt+0x9/0x10 [<ffffffff8159a959>] call_function_single_interrupt+0x89/0x90 <EOI> Fixes: commit be4b499323bf (IB/cm: Do not queue work to a device that's going away) Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Erez Shitrit <erezsh@mellanox.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Nikolay Borisov <kernel@kyup.com> Cc: stable <stable@vger.kernel.org> # v4.2+ Acked-by: Erez Shitrit <erezsh@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-25 23:33:16 +08:00
spin_unlock_irqrestore(&cm.lock, flags);
out:
return ret;
}
static int cm_migrate(struct ib_cm_id *cm_id)
{
struct cm_id_private *cm_id_priv;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
struct cm_av tmp_av;
unsigned long flags;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
int tmp_send_port_not_ready;
int ret = 0;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
spin_lock_irqsave(&cm_id_priv->lock, flags);
if (cm_id->state == IB_CM_ESTABLISHED &&
(cm_id->lap_state == IB_CM_LAP_UNINIT ||
cm_id->lap_state == IB_CM_LAP_IDLE)) {
cm_id->lap_state = IB_CM_LAP_IDLE;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
/* Swap address vector */
tmp_av = cm_id_priv->av;
cm_id_priv->av = cm_id_priv->alt_av;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
cm_id_priv->alt_av = tmp_av;
/* Swap port send ready state */
tmp_send_port_not_ready = cm_id_priv->prim_send_port_not_ready;
cm_id_priv->prim_send_port_not_ready = cm_id_priv->altr_send_port_not_ready;
cm_id_priv->altr_send_port_not_ready = tmp_send_port_not_ready;
} else
ret = -EINVAL;
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
int ib_cm_notify(struct ib_cm_id *cm_id, enum ib_event_type event)
{
int ret;
switch (event) {
case IB_EVENT_COMM_EST:
ret = cm_establish(cm_id);
break;
case IB_EVENT_PATH_MIG:
ret = cm_migrate(cm_id);
break;
default:
ret = -EINVAL;
}
return ret;
}
EXPORT_SYMBOL(ib_cm_notify);
static void cm_recv_handler(struct ib_mad_agent *mad_agent,
struct ib_mad_send_buf *send_buf,
struct ib_mad_recv_wc *mad_recv_wc)
{
struct cm_port *port = mad_agent->context;
struct cm_work *work;
enum ib_cm_event_type event;
bool alt_path = false;
u16 attr_id;
int paths = 0;
int going_down = 0;
switch (mad_recv_wc->recv_buf.mad->mad_hdr.attr_id) {
case CM_REQ_ATTR_ID:
alt_path = cm_req_has_alt_path((struct cm_req_msg *)
mad_recv_wc->recv_buf.mad);
paths = 1 + (alt_path != 0);
event = IB_CM_REQ_RECEIVED;
break;
case CM_MRA_ATTR_ID:
event = IB_CM_MRA_RECEIVED;
break;
case CM_REJ_ATTR_ID:
event = IB_CM_REJ_RECEIVED;
break;
case CM_REP_ATTR_ID:
event = IB_CM_REP_RECEIVED;
break;
case CM_RTU_ATTR_ID:
event = IB_CM_RTU_RECEIVED;
break;
case CM_DREQ_ATTR_ID:
event = IB_CM_DREQ_RECEIVED;
break;
case CM_DREP_ATTR_ID:
event = IB_CM_DREP_RECEIVED;
break;
case CM_SIDR_REQ_ATTR_ID:
event = IB_CM_SIDR_REQ_RECEIVED;
break;
case CM_SIDR_REP_ATTR_ID:
event = IB_CM_SIDR_REP_RECEIVED;
break;
case CM_LAP_ATTR_ID:
paths = 1;
event = IB_CM_LAP_RECEIVED;
break;
case CM_APR_ATTR_ID:
event = IB_CM_APR_RECEIVED;
break;
default:
ib_free_recv_mad(mad_recv_wc);
return;
}
attr_id = be16_to_cpu(mad_recv_wc->recv_buf.mad->mad_hdr.attr_id);
atomic_long_inc(&port->counter_group[CM_RECV].
counter[attr_id - CM_ATTR_ID_OFFSET]);
work = kmalloc(sizeof(*work) + sizeof(struct sa_path_rec) * paths,
GFP_KERNEL);
if (!work) {
ib_free_recv_mad(mad_recv_wc);
return;
}
INIT_DELAYED_WORK(&work->work, cm_work_handler);
work->cm_event.event = event;
work->mad_recv_wc = mad_recv_wc;
work->port = port;
/* Check if the device started its remove_one */
spin_lock_irq(&cm.lock);
if (!port->cm_dev->going_down)
queue_delayed_work(cm.wq, &work->work, 0);
else
going_down = 1;
spin_unlock_irq(&cm.lock);
if (going_down) {
kfree(work);
ib_free_recv_mad(mad_recv_wc);
}
}
static int cm_init_qp_init_attr(struct cm_id_private *cm_id_priv,
struct ib_qp_attr *qp_attr,
int *qp_attr_mask)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&cm_id_priv->lock, flags);
switch (cm_id_priv->id.state) {
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
case IB_CM_REQ_RCVD:
case IB_CM_MRA_REQ_SENT:
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
case IB_CM_ESTABLISHED:
*qp_attr_mask = IB_QP_STATE | IB_QP_ACCESS_FLAGS |
IB_QP_PKEY_INDEX | IB_QP_PORT;
qp_attr->qp_access_flags = IB_ACCESS_REMOTE_WRITE;
if (cm_id_priv->responder_resources)
qp_attr->qp_access_flags |= IB_ACCESS_REMOTE_READ |
IB_ACCESS_REMOTE_ATOMIC;
qp_attr->pkey_index = cm_id_priv->av.pkey_index;
qp_attr->port_num = cm_id_priv->av.port->port_num;
ret = 0;
break;
default:
ret = -EINVAL;
break;
}
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
static int cm_init_qp_rtr_attr(struct cm_id_private *cm_id_priv,
struct ib_qp_attr *qp_attr,
int *qp_attr_mask)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&cm_id_priv->lock, flags);
switch (cm_id_priv->id.state) {
case IB_CM_REQ_RCVD:
case IB_CM_MRA_REQ_SENT:
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
case IB_CM_ESTABLISHED:
*qp_attr_mask = IB_QP_STATE | IB_QP_AV | IB_QP_PATH_MTU |
IB_QP_DEST_QPN | IB_QP_RQ_PSN;
qp_attr->ah_attr = cm_id_priv->av.ah_attr;
qp_attr->path_mtu = cm_id_priv->path_mtu;
qp_attr->dest_qp_num = be32_to_cpu(cm_id_priv->remote_qpn);
qp_attr->rq_psn = be32_to_cpu(cm_id_priv->rq_psn);
if (cm_id_priv->qp_type == IB_QPT_RC ||
cm_id_priv->qp_type == IB_QPT_XRC_TGT) {
*qp_attr_mask |= IB_QP_MAX_DEST_RD_ATOMIC |
IB_QP_MIN_RNR_TIMER;
qp_attr->max_dest_rd_atomic =
cm_id_priv->responder_resources;
qp_attr->min_rnr_timer = 0;
}
if (rdma_ah_get_dlid(&cm_id_priv->alt_av.ah_attr)) {
*qp_attr_mask |= IB_QP_ALT_PATH;
qp_attr->alt_port_num = cm_id_priv->alt_av.port->port_num;
qp_attr->alt_pkey_index = cm_id_priv->alt_av.pkey_index;
qp_attr->alt_timeout = cm_id_priv->alt_av.timeout;
qp_attr->alt_ah_attr = cm_id_priv->alt_av.ah_attr;
}
ret = 0;
break;
default:
ret = -EINVAL;
break;
}
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
static int cm_init_qp_rts_attr(struct cm_id_private *cm_id_priv,
struct ib_qp_attr *qp_attr,
int *qp_attr_mask)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&cm_id_priv->lock, flags);
switch (cm_id_priv->id.state) {
/* Allow transition to RTS before sending REP */
case IB_CM_REQ_RCVD:
case IB_CM_MRA_REQ_SENT:
case IB_CM_REP_RCVD:
case IB_CM_MRA_REP_SENT:
case IB_CM_REP_SENT:
case IB_CM_MRA_REP_RCVD:
case IB_CM_ESTABLISHED:
if (cm_id_priv->id.lap_state == IB_CM_LAP_UNINIT) {
*qp_attr_mask = IB_QP_STATE | IB_QP_SQ_PSN;
qp_attr->sq_psn = be32_to_cpu(cm_id_priv->sq_psn);
switch (cm_id_priv->qp_type) {
case IB_QPT_RC:
case IB_QPT_XRC_INI:
*qp_attr_mask |= IB_QP_RETRY_CNT | IB_QP_RNR_RETRY |
IB_QP_MAX_QP_RD_ATOMIC;
qp_attr->retry_cnt = cm_id_priv->retry_count;
qp_attr->rnr_retry = cm_id_priv->rnr_retry_count;
qp_attr->max_rd_atomic = cm_id_priv->initiator_depth;
/* fall through */
case IB_QPT_XRC_TGT:
*qp_attr_mask |= IB_QP_TIMEOUT;
qp_attr->timeout = cm_id_priv->av.timeout;
break;
default:
break;
}
if (rdma_ah_get_dlid(&cm_id_priv->alt_av.ah_attr)) {
*qp_attr_mask |= IB_QP_PATH_MIG_STATE;
qp_attr->path_mig_state = IB_MIG_REARM;
}
} else {
*qp_attr_mask = IB_QP_ALT_PATH | IB_QP_PATH_MIG_STATE;
qp_attr->alt_port_num = cm_id_priv->alt_av.port->port_num;
qp_attr->alt_pkey_index = cm_id_priv->alt_av.pkey_index;
qp_attr->alt_timeout = cm_id_priv->alt_av.timeout;
qp_attr->alt_ah_attr = cm_id_priv->alt_av.ah_attr;
qp_attr->path_mig_state = IB_MIG_REARM;
}
ret = 0;
break;
default:
ret = -EINVAL;
break;
}
spin_unlock_irqrestore(&cm_id_priv->lock, flags);
return ret;
}
int ib_cm_init_qp_attr(struct ib_cm_id *cm_id,
struct ib_qp_attr *qp_attr,
int *qp_attr_mask)
{
struct cm_id_private *cm_id_priv;
int ret;
cm_id_priv = container_of(cm_id, struct cm_id_private, id);
switch (qp_attr->qp_state) {
case IB_QPS_INIT:
ret = cm_init_qp_init_attr(cm_id_priv, qp_attr, qp_attr_mask);
break;
case IB_QPS_RTR:
ret = cm_init_qp_rtr_attr(cm_id_priv, qp_attr, qp_attr_mask);
break;
case IB_QPS_RTS:
ret = cm_init_qp_rts_attr(cm_id_priv, qp_attr, qp_attr_mask);
break;
default:
ret = -EINVAL;
break;
}
return ret;
}
EXPORT_SYMBOL(ib_cm_init_qp_attr);
static ssize_t cm_show_counter(struct kobject *obj, struct attribute *attr,
char *buf)
{
struct cm_counter_group *group;
struct cm_counter_attribute *cm_attr;
group = container_of(obj, struct cm_counter_group, obj);
cm_attr = container_of(attr, struct cm_counter_attribute, attr);
return sprintf(buf, "%ld\n",
atomic_long_read(&group->counter[cm_attr->index]));
}
static const struct sysfs_ops cm_counter_ops = {
.show = cm_show_counter
};
static struct kobj_type cm_counter_obj_type = {
.sysfs_ops = &cm_counter_ops,
.default_attrs = cm_counter_default_attrs
};
static void cm_release_port_obj(struct kobject *obj)
{
struct cm_port *cm_port;
cm_port = container_of(obj, struct cm_port, port_obj);
kfree(cm_port);
}
static struct kobj_type cm_port_obj_type = {
.release = cm_release_port_obj
};
static char *cm_devnode(struct device *dev, umode_t *mode)
{
if (mode)
*mode = 0666;
return kasprintf(GFP_KERNEL, "infiniband/%s", dev_name(dev));
}
struct class cm_class = {
.owner = THIS_MODULE,
.name = "infiniband_cm",
.devnode = cm_devnode,
};
EXPORT_SYMBOL(cm_class);
static int cm_create_port_fs(struct cm_port *port)
{
int i, ret;
ret = kobject_init_and_add(&port->port_obj, &cm_port_obj_type,
&port->cm_dev->device->kobj,
"%d", port->port_num);
if (ret) {
kfree(port);
return ret;
}
for (i = 0; i < CM_COUNTER_GROUPS; i++) {
ret = kobject_init_and_add(&port->counter_group[i].obj,
&cm_counter_obj_type,
&port->port_obj,
"%s", counter_group_names[i]);
if (ret)
goto error;
}
return 0;
error:
while (i--)
kobject_put(&port->counter_group[i].obj);
kobject_put(&port->port_obj);
return ret;
}
static void cm_remove_port_fs(struct cm_port *port)
{
int i;
for (i = 0; i < CM_COUNTER_GROUPS; i++)
kobject_put(&port->counter_group[i].obj);
kobject_put(&port->port_obj);
}
static void cm_add_one(struct ib_device *ib_device)
{
struct cm_device *cm_dev;
struct cm_port *port;
struct ib_mad_reg_req reg_req = {
.mgmt_class = IB_MGMT_CLASS_CM,
.mgmt_class_version = IB_CM_CLASS_VERSION,
};
struct ib_port_modify port_modify = {
.set_port_cap_mask = IB_PORT_CM_SUP
};
unsigned long flags;
int ret;
int count = 0;
u8 i;
cm_dev = kzalloc(sizeof(*cm_dev) + sizeof(*port) *
ib_device->phys_port_cnt, GFP_KERNEL);
if (!cm_dev)
return;
cm_dev->ib_device = ib_device;
cm_dev->ack_delay = ib_device->attrs.local_ca_ack_delay;
cm_dev->going_down = 0;
cm_dev->device = device_create(&cm_class, &ib_device->dev,
MKDEV(0, 0), NULL,
"%s", ib_device->name);
if (IS_ERR(cm_dev->device)) {
kfree(cm_dev);
return;
}
set_bit(IB_MGMT_METHOD_SEND, reg_req.method_mask);
for (i = 1; i <= ib_device->phys_port_cnt; i++) {
if (!rdma_cap_ib_cm(ib_device, i))
continue;
port = kzalloc(sizeof *port, GFP_KERNEL);
if (!port)
goto error1;
cm_dev->port[i-1] = port;
port->cm_dev = cm_dev;
port->port_num = i;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
INIT_LIST_HEAD(&port->cm_priv_prim_list);
INIT_LIST_HEAD(&port->cm_priv_altr_list);
ret = cm_create_port_fs(port);
if (ret)
goto error1;
port->mad_agent = ib_register_mad_agent(ib_device, i,
IB_QPT_GSI,
&reg_req,
0,
cm_send_handler,
cm_recv_handler,
port,
0);
if (IS_ERR(port->mad_agent))
goto error2;
ret = ib_modify_port(ib_device, i, 0, &port_modify);
if (ret)
goto error3;
count++;
}
if (!count)
goto free;
ib_set_client_data(ib_device, &cm_client, cm_dev);
write_lock_irqsave(&cm.device_lock, flags);
list_add_tail(&cm_dev->list, &cm.device_list);
write_unlock_irqrestore(&cm.device_lock, flags);
return;
error3:
ib_unregister_mad_agent(port->mad_agent);
error2:
cm_remove_port_fs(port);
error1:
port_modify.set_port_cap_mask = 0;
port_modify.clr_port_cap_mask = IB_PORT_CM_SUP;
while (--i) {
if (!rdma_cap_ib_cm(ib_device, i))
continue;
port = cm_dev->port[i-1];
ib_modify_port(ib_device, port->port_num, 0, &port_modify);
ib_unregister_mad_agent(port->mad_agent);
cm_remove_port_fs(port);
}
free:
device_unregister(cm_dev->device);
kfree(cm_dev);
}
static void cm_remove_one(struct ib_device *ib_device, void *client_data)
{
struct cm_device *cm_dev = client_data;
struct cm_port *port;
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
struct cm_id_private *cm_id_priv;
struct ib_mad_agent *cur_mad_agent;
struct ib_port_modify port_modify = {
.clr_port_cap_mask = IB_PORT_CM_SUP
};
unsigned long flags;
int i;
if (!cm_dev)
return;
write_lock_irqsave(&cm.device_lock, flags);
list_del(&cm_dev->list);
write_unlock_irqrestore(&cm.device_lock, flags);
spin_lock_irq(&cm.lock);
cm_dev->going_down = 1;
spin_unlock_irq(&cm.lock);
for (i = 1; i <= ib_device->phys_port_cnt; i++) {
if (!rdma_cap_ib_cm(ib_device, i))
continue;
port = cm_dev->port[i-1];
ib_modify_port(ib_device, port->port_num, 0, &port_modify);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
/* Mark all the cm_id's as not valid */
spin_lock_irq(&cm.lock);
list_for_each_entry(cm_id_priv, &port->cm_priv_altr_list, altr_list)
cm_id_priv->altr_send_port_not_ready = 1;
list_for_each_entry(cm_id_priv, &port->cm_priv_prim_list, prim_list)
cm_id_priv->prim_send_port_not_ready = 1;
spin_unlock_irq(&cm.lock);
/*
* We flush the queue here after the going_down set, this
* verify that no new works will be queued in the recv handler,
* after that we can call the unregister_mad_agent
*/
flush_workqueue(cm.wq);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
spin_lock_irq(&cm.state_lock);
cur_mad_agent = port->mad_agent;
port->mad_agent = NULL;
spin_unlock_irq(&cm.state_lock);
ib_unregister_mad_agent(cur_mad_agent);
cm_remove_port_fs(port);
}
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
device_unregister(cm_dev->device);
kfree(cm_dev);
}
static int __init ib_cm_init(void)
{
int ret;
memset(&cm, 0, sizeof cm);
INIT_LIST_HEAD(&cm.device_list);
rwlock_init(&cm.device_lock);
spin_lock_init(&cm.lock);
IB/cm: Mark stale CM id's whenever the mad agent was unregistered When there is a CM id object that has port assigned to it, it means that the cm-id asked for the specific port that it should go by it, but if that port was removed (hot-unplug event) the cm-id was not updated. In order to fix that the port keeps a list of all the cm-id's that are planning to go by it, whenever the port is removed it marks all of them as invalid. This commit fixes a kernel panic which happens when running traffic between guests and we force reboot a guest mid traffic, it triggers a kernel panic: Call Trace: [<ffffffff815271fa>] ? panic+0xa7/0x16f [<ffffffff8152b534>] ? oops_end+0xe4/0x100 [<ffffffff8104a00b>] ? no_context+0xfb/0x260 [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30 [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0 [<ffffffff81084240>] ? process_timeout+0x0/0x10 [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20 [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480 [<ffffffff81065df0>] ? default_wake_function+0x0/0x20 [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core] [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core] [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0 [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0 [<ffffffff8152a815>] ? page_fault+0x25/0x30 [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm] [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm] [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm] [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm] [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib] [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib] [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib] [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0 [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0 [<ffffffff8109aef6>] ? kthread+0x96/0xa0 [<ffffffff8100c20a>] ? child_rip+0xa/0x20 [<ffffffff8109ae60>] ? kthread+0x0/0xa0 [<ffffffff8100c200>] ? child_rip+0x0/0x20 Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation") Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-10-27 21:36:27 +08:00
spin_lock_init(&cm.state_lock);
cm.listen_service_table = RB_ROOT;
cm.listen_service_id = be64_to_cpu(IB_CM_ASSIGN_SERVICE_ID);
cm.remote_id_table = RB_ROOT;
cm.remote_qp_table = RB_ROOT;
cm.remote_sidr_table = RB_ROOT;
idr_init(&cm.local_id_table);
get_random_bytes(&cm.random_id_operand, sizeof cm.random_id_operand);
INIT_LIST_HEAD(&cm.timewait_list);
ret = class_register(&cm_class);
if (ret) {
ret = -ENOMEM;
goto error1;
}
cm.wq = alloc_workqueue("ib_cm", 0, 1);
if (!cm.wq) {
ret = -ENOMEM;
goto error2;
}
ret = ib_register_client(&cm_client);
if (ret)
goto error3;
return 0;
error3:
destroy_workqueue(cm.wq);
error2:
class_unregister(&cm_class);
error1:
idr_destroy(&cm.local_id_table);
return ret;
}
static void __exit ib_cm_cleanup(void)
{
struct cm_timewait_info *timewait_info, *tmp;
spin_lock_irq(&cm.lock);
list_for_each_entry(timewait_info, &cm.timewait_list, list)
cancel_delayed_work(&timewait_info->work.work);
spin_unlock_irq(&cm.lock);
ib_unregister_client(&cm_client);
destroy_workqueue(cm.wq);
list_for_each_entry_safe(timewait_info, tmp, &cm.timewait_list, list) {
list_del(&timewait_info->list);
kfree(timewait_info);
}
class_unregister(&cm_class);
idr_destroy(&cm.local_id_table);
}
module_init(ib_cm_init);
module_exit(ib_cm_cleanup);