Commit ed041919 authored by Bart Van Assche's avatar Bart Van Assche Committed by Doug Ledford

RDMA/srpt: Fix a use-after-free in the channel release code

This patch avoids that KASAN sporadically reports the following:

BUG: KASAN: use-after-free in rxe_run_task+0x1e/0x60 [rdma_rxe]
Read of size 1 at addr ffff88801c50d8f4 by task check/24830

CPU: 4 PID: 24830 Comm: check Not tainted 4.20.0-rc6-dbg+ #3
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
Call Trace:
 dump_stack+0x86/0xca
 print_address_description+0x71/0x239
 kasan_report.cold.5+0x242/0x301
 __asan_load1+0x47/0x50
 rxe_run_task+0x1e/0x60 [rdma_rxe]
 rxe_post_send+0x4bd/0x8d0 [rdma_rxe]
 srpt_zerolength_write+0xe1/0x160 [ib_srpt]
 srpt_close_ch+0x8b/0xe0 [ib_srpt]
 srpt_set_enabled+0xe7/0x150 [ib_srpt]
 srpt_tpg_enable_store+0xc0/0x100 [ib_srpt]
 configfs_write_file+0x157/0x1d0
 __vfs_write+0xd7/0x3d0
 vfs_write+0x102/0x290
 ksys_write+0xab/0x130
 __x64_sys_write+0x43/0x50
 do_syscall_64+0x71/0x210
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Allocated by task 13856:
 save_stack+0x43/0xd0
 kasan_kmalloc+0xc7/0xe0
 kasan_slab_alloc+0x11/0x20
 kmem_cache_alloc+0x105/0x320
 rxe_alloc+0xff/0x1f0 [rdma_rxe]
 rxe_create_qp+0x9f/0x160 [rdma_rxe]
 ib_create_qp+0xf5/0x690 [ib_core]
 rdma_create_qp+0x6a/0x140 [rdma_cm]
 srpt_cm_req_recv.cold.59+0x1588/0x237b [ib_srpt]
 srpt_rdma_cm_req_recv.isra.35+0x1d5/0x220 [ib_srpt]
 srpt_rdma_cm_handler+0x6f/0x100 [ib_srpt]
 cma_listen_handler+0x59/0x60 [rdma_cm]
 cma_ib_req_handler+0xd5b/0x2570 [rdma_cm]
 cm_process_work+0x2e/0x110 [ib_cm]
 cm_work_handler+0x2aae/0x502b [ib_cm]
 process_one_work+0x481/0x9e0
 worker_thread+0x67/0x5b0
 kthread+0x1cf/0x1f0
 ret_from_fork+0x24/0x30

Freed by task 3440:
 save_stack+0x43/0xd0
 __kasan_slab_free+0x139/0x190
 kasan_slab_free+0xe/0x10
 kmem_cache_free+0xbc/0x330
 rxe_elem_release+0x66/0xe0 [rdma_rxe]
 rxe_destroy_qp+0x3f/0x50 [rdma_rxe]
 ib_destroy_qp+0x140/0x360 [ib_core]
 srpt_release_channel_work+0xdc/0x310 [ib_srpt]
 process_one_work+0x481/0x9e0
 worker_thread+0x67/0x5b0
 kthread+0x1cf/0x1f0
 ret_from_fork+0x24/0x30

Cc: Sergey Gorenko <sergeygo@mellanox.com>
Cc: Max Gurtovoy <maxg@mellanox.com>
Cc: Laurence Oberman <loberman@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: default avatarBart Van Assche <bvanassche@acm.org>
Signed-off-by: default avatarDoug Ledford <dledford@redhat.com>
parent 882981f4
......@@ -2010,6 +2010,14 @@ static void srpt_free_ch(struct kref *kref)
kfree_rcu(ch, rcu);
}
/*
* Shut down the SCSI target session, tell the connection manager to
* disconnect the associated RDMA channel, transition the QP to the error
* state and remove the channel from the channel list. This function is
* typically called from inside srpt_zerolength_write_done(). Concurrent
* srpt_zerolength_write() calls from inside srpt_close_ch() are possible
* as long as the channel is on sport->nexus_list.
*/
static void srpt_release_channel_work(struct work_struct *w)
{
struct srpt_rdma_ch *ch;
......@@ -2037,6 +2045,11 @@ static void srpt_release_channel_work(struct work_struct *w)
else
ib_destroy_cm_id(ch->ib_cm.cm_id);
sport = ch->sport;
mutex_lock(&sport->mutex);
list_del_rcu(&ch->list);
mutex_unlock(&sport->mutex);
srpt_destroy_ch_ib(ch);
srpt_free_ioctx_ring((struct srpt_ioctx **)ch->ioctx_ring,
......@@ -2047,11 +2060,6 @@ static void srpt_release_channel_work(struct work_struct *w)
sdev, ch->rq_size,
srp_max_req_size, DMA_FROM_DEVICE);
sport = ch->sport;
mutex_lock(&sport->mutex);
list_del_rcu(&ch->list);
mutex_unlock(&sport->mutex);
wake_up(&sport->ch_releaseQ);
kref_put(&ch->kref, srpt_free_ch);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment