在 2022/04/21 17:30, Ming Lei 写道:
On Thu, Apr 21, 2022 at 04:51:28PM +0800, yukuai (C) wrote:
在 2022/04/20 20:42, Shinichiro Kawasaki 写道:
Hello,
Jens has merged Yu Kuai's fix[1], so I think it won't be triggered now.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git/commit/?h=block-5.18&id=a87c29e1a85e64b28445bb1e80505230bf2e3b4b
Hi Ming, I applied the patch above on top of v5.18-rc3 and ran block/002.
Unfortunately, it failed with a new symptom with KASAN use-after-free [2]. I
ran block/002 with linux-block/block-5.18 branch tip with git hash a87c29e1a85e
and got the same KASAN uaf. Reverting the patch from the linux-block/block-5.18
branch, the KASAN uaf disappears (Still block/002 fails). Regarding block/002,
it looks the patch made the failure symptom worse.
[2]
[ 466.424358] run blktests block/002 at 2022-04-20 19:44:02
[ 466.508847] scsi_debug:sdebug_driver_probe: scsi_debug: trim poll_queues to 0. poll_q/nr_hw = (0/1)
[ 466.518617] scsi host7: scsi_debug: version 0191 [20210520]
dev_size_mb=8, opts=0x0, submit_queues=1, statistics=0
[ 466.535080] scsi 7:0:0:0: Direct-Access Linux scsi_debug 0191 PQ: 0 ANSI: 7
[ 466.548701] sd 7:0:0:0: Power-on or device reset occurred
[ 466.549819] sd 7:0:0:0: Attached scsi generic sg9 type 0
[ 466.557985] sd 7:0:0:0: [sdi] 16384 512-byte logical blocks: (8.39 MB/8.00 MiB)
[ 466.570116] sd 7:0:0:0: [sdi] Write Protect is off
[ 466.575644] sd 7:0:0:0: [sdi] Mode Sense: 73 00 10 08
[ 466.577821] sd 7:0:0:0: [sdi] Write cache: enabled, read cache: enabled, supports DPO and FUA
[ 466.590343] sd 7:0:0:0: [sdi] Optimal transfer size 524288 bytes
[ 466.645516] sd 7:0:0:0: [sdi] Attached SCSI disk
[ 467.438285] sd 7:0:0:0: [sdi] Synchronizing SCSI cache
[ 467.458790] ==================================================================
[ 467.466714] BUG: KASAN: use-after-free in __lock_acquire+0x396b/0x5030
[ 467.473951] Read of size 8 at addr ffff888104e05248 by task check/1549
[ 467.483373] CPU: 1 PID: 1549 Comm: check Not tainted 5.18.0-rc3+ #24
[ 467.490426] Hardware name: Supermicro X10SLL-F/X10SLL-F, BIOS 3.0 04/24/2015
[ 467.498164] Call Trace:
[ 467.501313] <TASK>
[ 467.504120] dump_stack_lvl+0x56/0x6f
[ 467.508488] print_report.cold+0x5e/0x5db
[ 467.513205] ? __lock_acquire+0x396b/0x5030
[ 467.518092] kasan_report+0xbf/0xf0
[ 467.522288] ? lockdep_lock+0x30/0x1a0
[ 467.526738] ? __lock_acquire+0x396b/0x5030
[ 467.531630] __lock_acquire+0x396b/0x5030
[ 467.536346] ? lockdep_unlock+0xf2/0x240
[ 467.540970] ? __lock_acquire+0x23db/0x5030
[ 467.545861] ? lockdep_hardirqs_on_prepare+0x410/0x410
[ 467.551705] lock_acquire+0x19a/0x4b0
[ 467.556068] ? lockref_get+0x9/0x40
[ 467.560264] ? lock_release+0x6c0/0x6c0
[ 467.564806] ? lock_is_held_type+0xe2/0x140
[ 467.569693] ? find_held_lock+0x2c/0x110
[ 467.574316] ? lock_release+0x3a7/0x6c0
[ 467.578856] _raw_spin_lock+0x2f/0x40
[ 467.583222] ? lockref_get+0x9/0x40
[ 467.587414] lockref_get+0x9/0x40
[ 467.591439] simple_recursive_removal+0x36/0x7e0
[ 467.596758] ? debugfs_remove+0x60/0x60
[ 467.601300] ? do_raw_spin_unlock+0x55/0x1f0
[ 467.606278] debugfs_remove+0x40/0x60
[ 467.610643] blk_mq_debugfs_unregister_queue_rqos+0x34/0x70
[ 467.616919] rq_qos_exit+0x1b/0xf0
[ 467.621028] ? sysfs_file_ops+0x170/0x170
[ 467.625740] blk_cleanup_queue+0xfd/0x1f0
[ 467.630449] __scsi_remove_device+0xdd/0x2b0
[ 467.635422] sdev_store_delete+0x83/0x120
[ 467.640137] kernfs_fop_write_iter+0x353/0x520
[ 467.645287] new_sync_write+0x2d9/0x500
[ 467.649827] ? new_sync_read+0x500/0x500
[ 467.654455] ? perf_msr_probe+0x1f0/0x280
[ 467.659170] ? lock_release+0x6c0/0x6c0
[ 467.663709] ? inode_security+0x54/0xf0
[ 467.668253] ? lock_is_held_type+0xe2/0x140
[ 467.673144] vfs_write+0x61c/0x910
[ 467.677250] ksys_write+0xe3/0x1a0
[ 467.681355] ? __ia32_sys_read+0xa0/0xa0
[ 467.685982] ? syscall_enter_from_user_mode+0x21/0x70
[ 467.691740] do_syscall_64+0x3b/0x90
[ 467.696018] entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 467.701772] RIP: 0033:0x7f2d0b701817
[ 467.706046] Code: 0f 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 48 89 54 24 18 48 89 74 24
[ 467.725492] RSP: 002b:00007ffd37a645e8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
[ 467.733762] RAX: ffffffffffffffda RBX: 0000000000000002 RCX: 00007f2d0b701817
[ 467.741596] RDX: 0000000000000002 RSI: 000055a23ecf4630 RDI: 0000000000000001
[ 467.749431] RBP: 000055a23ecf4630 R08: 0000000000000000 R09: 00007f2d0b7b64e0
[ 467.757267] R10: 00007f2d0b7b63e0 R11: 0000000000000246 R12: 0000000000000002
[ 467.765101] R13: 00007f2d0b7fb5a0 R14: 0000000000000002 R15: 00007f2d0b7fb7a0
[ 467.772945] </TASK>
[ 467.778033] Allocated by task 111:
[ 467.782141] kasan_save_stack+0x1e/0x40
[ 467.786681] __kasan_slab_alloc+0x90/0xc0
[ 467.791395] kmem_cache_alloc_lru+0x258/0x720
[ 467.796457] __d_alloc+0x31/0x960
[ 467.800477] d_alloc+0x44/0x200
[ 467.804326] d_alloc_parallel+0xca/0x1490
[ 467.809041] __lookup_slow+0x17f/0x3d0
[ 467.813495] lookup_one_len+0x10b/0x130
[ 467.818038] start_creating.part.0+0xf0/0x220
[ 467.823098] debugfs_create_dir+0x2f/0x460
[ 467.827901] blk_mq_debugfs_register_rqos+0x1fe/0x330
Hi,
Sorry that I missed the 'q->rqos_debugfs_dir' which is created under
'q->debugfs_dir'.
Ming, do you think move blk_mq_debugfs_unregister_queue_rqos() to
blk_unregister_queue() is okay?
Hi Yukuai,
blktrace still may work for passthrough req trace after disk is deleted,
so I think removing q->debugfs_dir in blk_unregister_queue() may not a
good idea.
Thanks for the explanation, that make sense.
Kuai
