RDMA/mlx5: Fix NULL pointer dereference in destroy_prefetch_work
Subsystems (2)
PATCH 1/1 RDMA/mlx5: Fix NULL pointer dereference in destroy_prefetch_work
From: Maor Gottlieb <[email protected]>
q_deferred_work isn't initialized when user create an explicit ODP memory region. It could lead to NULL pointer dereference when user performs asynchronous prefetch MR.
Fix it by initialize q_deferred_work for explicit ODP.
[ 2360.844536] BUG: kernel NULL pointer dereference, address: 0000000000000000 [ 2360.855944] #PF: supervisor read access in kernel mode [ 2360.858276] #PF: error_code(0x0000) - not-present page [ 2360.860600] PGD 0 P4D 0 [ 2360.861636] Oops: 0000 [#1] SMP PTI [ 2360.862781] CPU: 4 PID: 6074 Comm: kworker/u16:6 Not tainted 5.7.0-rc1-for-upstream-perf-2020-04-17_07-03-39-64 #1 [ 2360.865369] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.12.1-0-ga5cab58e9a3f-prebuilt.qemu.org 04/01/2014 [ 2360.868090] Workqueue: events_unbound mlx5_ib_prefetch_mr_work [mlx5_ib] [ 2360.869641] RIP: 0010:__wake_up_common+0x49/0x120 [ 2360.870906] Code: 04 89 54 24 0c 89 4c 24 08 74 0a 41 f6 01 04 0f 85 8e 00 00 00 48 8b 47 08 48 83 e8 18 4c 8d 67 08 48 8d 50 18 49 39 d4 74 66 <48> 8b 70 18 31 db 4c 8d 7e e8 eb 17 49 8b 47 18 48 8d 50 e8 49 8d [ 2360.875211] RSP: 0000:ffffc9000097bd88 EFLAGS: 00010082 [ 2360.876594] RAX: ffffffffffffffe8 RBX: ffff888454cd9f90 RCX: 0000000000000000 [ 2360.878261] RDX: 0000000000000000 RSI: 0000000000000003 RDI: ffff888454cd9f90 [ 2360.879930] RBP: ffffc9000097bdd0 R08: 0000000000000000 R09: ffffc9000097bdd0 [ 2360.881593] R10: 0000000000000000 R11: 0000000000000001 R12: ffff888454cd9f98 [ 2360.883291] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000003 [ 2360.884970] FS: 0000000000000000(0000) GS:ffff88846fd00000(0000) knlGS:0000000000000000 [ 2360.887231] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2360.888656] CR2: 0000000000000000 CR3: 000000044c19e002 CR4: 0000000000760ee0 [ 2360.890269] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 2360.891881] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 2360.893489] PKRU: 55555554 [ 2360.894504] Call Trace: [ 2360.895484] __wake_up_common_lock+0x7a/0xc0 [ 2360.896728] destroy_prefetch_work+0x5a/0x60 [mlx5_ib] [ 2360.898106] mlx5_ib_prefetch_mr_work+0x64/0x80 [mlx5_ib] [ 2360.899525] process_one_work+0x15b/0x360 [ 2360.900726] worker_thread+0x49/0x3d0 [ 2360.901879] kthread+0xf5/0x130 [ 2360.902961] ? rescuer_thread+0x310/0x310 [ 2360.904172] ? kthread_bind+0x10/0x10 [ 2360.905334] ret_from_fork+0x1f/0x30
Fixes de5ed007a03d ("IB/mlx5: Fix implicit ODP race")
Signed-off-by Maor Gottlieb <[email protected]>
Signed-off-by Leon Romanovsky <[email protected]>
--- drivers/infiniband/hw/mlx5/mr.c | 1 + 1 file changed, 1 insertion(+)
drivers/infiniband/hw/mlx5/mr.c +1 -0
@@ -1439,6 +1439,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,if (is_odp_mr (mr)) {to_ib_umem_odp (mr->umem)->private = mr; +init_waitqueue_head (&mr->q_deferred_work);atomic_set (&mr->num_deferred_work,0 ); err = xa_err(xa_store(&dev->odp_mkeys, mlx5_base_mkey(mr->mmkey.key), &mr->mmkey,
Signature · 1 line
2.26.2
Jason Gunthorpe
AppliedQuoted 42 lines · Leon Romanovsky
From: Maor Gottlieb <[email protected]>
q_deferred_workisn't initialized when user create an explicit ODP memory region. It could lead to NULL pointer dereference when user performs asynchronous prefetch MR.
Fix it by initializeq_deferred_workfor explicit ODP.[ 2360.844536] BUG: kernel NULL pointer dereference, address: 0000000000000000 [ 2360.855944] #PF: supervisor read access in kernel mode [ 2360.858276] #PF: error_code(0x0000) - not-present page [ 2360.860600] PGD 0 P4D 0 [ 2360.861636] Oops: 0000 [#1] SMP PTI [ 2360.862781] CPU: 4 PID: 6074 Comm: kworker/u16:6 Not tainted 5.7.0-rc1-for-upstream-perf-2020-04-17_07-03-39-64 #1 [ 2360.865369] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.12.1-0-ga5cab58e9a3f-prebuilt.qemu.org 04/01/2014 [ 2360.868090] Workqueue: events_unbound mlx5_ib_prefetch_mr_work [mlx5_ib] [ 2360.869641] RIP: 0010:__wake_up_common+0x49/0x120 [ 2360.870906] Code: 04 89 54 24 0c 89 4c 24 08 74 0a 41 f6 01 04 0f 85 8e 00 00 00 48 8b 47 08 48 83 e8 18 4c 8d 67 08 48 8d 50 18 49 39 d4 74 66 <48> 8b 70 18 31 db 4c 8d 7e e8 eb 17 49 8b 47 18 48 8d 50 e8 49 8d [ 2360.875211] RSP: 0000:ffffc9000097bd88 EFLAGS: 00010082 [ 2360.876594] RAX: ffffffffffffffe8 RBX: ffff888454cd9f90 RCX: 0000000000000000 [ 2360.878261] RDX: 0000000000000000 RSI: 0000000000000003 RDI: ffff888454cd9f90 [ 2360.879930] RBP: ffffc9000097bdd0 R08: 0000000000000000 R09: ffffc9000097bdd0 [ 2360.881593] R10: 0000000000000000 R11: 0000000000000001 R12: ffff888454cd9f98 [ 2360.883291] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000003 [ 2360.884970] FS: 0000000000000000(0000) GS:ffff88846fd00000(0000) knlGS:0000000000000000 [ 2360.887231] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2360.888656] CR2: 0000000000000000 CR3: 000000044c19e002 CR4: 0000000000760ee0 [ 2360.890269] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 2360.891881] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7:
Quoted 1 line
0000000000000400
Do not word wrap oops dumps, and please omit the valueless timestamps as well.
Quoted 12 lines
[ 2360.893489] PKRU: 55555554 [ 2360.894504] Call Trace: [ 2360.895484] __wake_up_common_lock+0x7a/0xc0 [ 2360.896728] destroy_prefetch_work+0x5a/0x60 [mlx5_ib] [ 2360.898106] mlx5_ib_prefetch_mr_work+0x64/0x80 [mlx5_ib] [ 2360.899525] process_one_work+0x15b/0x360 [ 2360.900726] worker_thread+0x49/0x3d0 [ 2360.901879] kthread+0xf5/0x130 [ 2360.902961] ? rescuer_thread+0x310/0x310 [ 2360.904172] ? kthread_bind+0x10/0x10 [ 2360.905334] ret_from_fork+0x1f/0x30
Quoted 5 lines
Fixes de5ed007a03d ("IB/mlx5: Fix implicit ODP race")Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>drivers/infiniband/hw/mlx5/mr.c | 1 + 1 file changed, 1 insertion(+)
Applied to for-rc, thanks
I added a cc stable
Jason