[LU-14441] Oops in obd_set_max_rpcs_in_flight()

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.16.0
• Affects Version/s: Upstream
• Severity: 3
• Assignee: Alex Zhuravlev
• Reporter: Alex Zhuravlev

Description

BUG: unable to handle kernel NULL pointer dereference at 00000000000000e0
PGD 12785f067 P4D 12785f067 PUD 11da1c067 PMD 0 
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
CPU: 0 PID: 15555 Comm: lctl Tainted: G        W  O     --------- ---  4.18.0 #34
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
RIP: 0010:obd_set_max_rpcs_in_flight+0x29/0x270 [obdclass]
Code: ff 8d 46 ff 3d ff 01 00 00 0f 87 f6 01 00 00 41 56 f6 05 85 ba e7 ff 40 41 55 41 54 55 89 f5 53 48 8b 87 c8 00 00 00 48 89 fb <48> 8b 90 e0 00 00 00 0f 85 1a 01 00 00 48 8b 02 b9 04 00 00 00 48
RSP: 0018:ffff88014acafde0 EFLAGS: 00010202
RAX: 0000000000000000 RBX: ffff8801469110e0 RCX: 0000000000000008
RDX: 0000000000000008 RSI: 0000000000000008 RDI: ffff8801469110e0
RBP: 0000000000000008 R08: 0000000000000008 R09: 0000000000000001
R10: 000000000000000a R11: f000000000000000 R12: 0000000000000001
R13: fffffffffffffff2 R14: ffff88014acaff10 R15: ffff88015c071e20
FS:  00007f2d019ea740(0000) GS:ffff88016ae00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000000000e0 CR3: 000000015c1e7000 CR4: 00000000000006b0
Call Trace:
 max_rpcs_in_flight_store+0x3b/0x50 [mdc]
 kernfs_fop_write+0x10d/0x190
 __vfs_write+0x1f/0x160
 ? rcu_sync_lockdep_assert+0x9/0x50
 ? __sb_start_write+0x13f/0x1a0
 ? vfs_write+0x183/0x1b0
 vfs_write+0xba/0x1b0
 ksys_write+0x3d/0xa0
 do_syscall_64+0x4b/0x1a0
 entry_SYSCALL_64_after_hwframe+0x6a/0xdf

that's because cli->cl_import=NULL as a concurrent umount was shutting the client down.

Issue Links

• is related to: LU-15501 obd_set_max_rpcs_in_flight store is racy