Hi Greg and Stable folks.
We've noticed regression in raid1 due to following commits:
79dabfd00a2b ("md/raid1: hold the barrier until handle_read_error() finishes")
caeed0b9f1ce ("md/raid1: free the r1bio before waiting for blocked rdev")
Kernel crash during io tests like below:
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007040] RIP:
0010:call_bio_endio+0x1a/0x60 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007147] Code: 00 5b e9 d9 79
b3 f0 66 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 53 48 8b 47 18 48 8b
5f 30 a8 01 75 04 c6 43 1a 0a 48 8b 53 08 <48> 8b 82 40 02 00 00 48 8b
40 50 48 8b 40 60 a8 80 74 12 48 8b 47
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007347] RSP:
0018:ffffb3300f627b90 EFLAGS: 00010202
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007448] RAX:
0000000000000025 RBX: ffff8d2cab013210 RCX: 0000000000000000
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007582] RDX:
00000001ab000000 RSI: ffff8d2cab013210 RDI: ffff8cfdb1e1d100
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007688] RBP:
ffff8d34c2ee2800 R08: 000000000000c0d4 R09: 0000000000073f2c
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007795] R10:
0000000000073f34 R11: ffff8cfdb1e1d100 R12: ffff8d2cab013200
Sep 11 23:03:15 ps401a-901 kernel: [ 449.007901] R13:
0000000000000000 R14: 0000000000000000 R15: ffff8d34c2ee2800
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008011] FS:
0000000000000000(0000) GS:ffff8d3487c40000(0000)
knlGS:0000000000000000
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008146] CS: 0010 DS: 0000
ES: 0000 CR0: 0000000080050033
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008248] CR2:
00000001ab000240 CR3: 000000038360a000 CR4: 00000000000406e0
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008355] Call Trace:
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008448] <TASK>
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008539] ? __die_body+0x1a/0x60
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008638] ? page_fault_oops+0x136/0x2a0
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008754] ? exc_page_fault+0x5f/0x110
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008853] ?
asm_exc_page_fault+0x22/0x30
Sep 11 23:03:15 ps401a-901 kernel: [ 449.008955] ?
call_bio_endio+0x1a/0x60 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009055]
raid_end_bio_io+0x28/0x90 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009158]
raid1_end_write_request+0x10b/0x340 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009263] submit_bio_checks+0x84/0x450
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009364] ? __wake_up_common+0x77/0x140
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009463] __submit_bio+0x106/0x190
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009560] ? __queue_work+0x136/0x3b0
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009659] submit_bio_noacct+0x268/0x2c0
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009758]
flush_bio_list+0x60/0x100 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009859]
flush_pending_writes+0x71/0xb0 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.009976] raid1d+0xa6/0x1280 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010076] ? psi_task_switch+0xde/0x200
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010175] ? __switch_to_asm+0x3a/0x60
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010274] ?
finish_task_switch+0x7d/0x280
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010373] ?
try_to_del_timer_sync+0x4d/0x80
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010475] ?
md_thread+0x137/0x170 [md_mod]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010586] ?
process_checks+0x4c0/0x4c0 [raid1]
Sep 11 23:03:15 ps401a-901 kernel: [ 449.010688]
md_thread+0x137/0x170 [md_mod]
Reverting both patches locally I can no longer reproduce the crash.
Please drop both patches from all the stable queues.
Thx!
Jnipu Wang @ IONOS cloud
