[LU-13494] osp_invalidate_request() races with osp_update_fini()

Related — Tue, 21 Mar 2023 23:37:19 +0000

BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
...
RIP: 0010:osp_invalidate_request+0x74/0x3e0 [osp]
Code: 00 48 8b 43 08 48 8d 7c 24 20 4c 8d 6d 10 8b 70 18 e8 f0 51 39 ff 85 c0 0f 88 5d 01 00 00 4c 89 ef e8 60 bc 97 e0 48 8b 45 00 <48> 8b 30 4c 8d b8 60 ff ff ff 48 39 c5 49 89 c6 48 8d 8e 60 ff ff
RSP: 0018:ffff88023d453df8 EFLAGS: 00010286
RAX: 0000000000000000 RBX: ffff88021b36e000 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffffffffa0e1a4e0 RDI: 0000000000000282
RBP: ffff88024ccf6c00 R08: 0000000000000000 R09: 0000000000000001
R10: 0000000000000000 R11: 0000000000000000 R12: ffff88023d453e08
R13: ffff88024ccf6c10 R14: ffffffff00000100 R15: 00000000000007cf
FS:  0000000000000000(0000) GS:ffff88024f200000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000000 CR3: 000000024cfdd000 CR4: 00000000000006b0
Call Trace:
 osp_import_event+0x337/0x340 [osp]
 ptlrpc_invalidate_import+0x29b/0x980 [ptlrpc]
 ? wait_woken+0x90/0x90
 ptlrpc_invalidate_import_thread+0x3e/0x1c0 [ptlrpc]
 kthread+0x105/0x140
 ? ptlrpc_import_recovery_state_machine+0x960/0x960 [ptlrpc]
 ? kthread_flush_work_fn+0x10/0x10
 ret_from_fork+0x24/0x30

that's because import invalidation runs in a dedicated context and not synchronize with main cleanup thread calling into osp_update_fini()

Whamcloud Community JIRA

[LU-13494] osp_invalidate_request() races with osp_update_fini()