Hi Jens,
I just realised that I'm still seeing a warning about circular locking
involving q->elevator_lock in -rc6. Do you know if there's a fix for this
yet?
David
---
======================================================
WARNING: possible circular locking dependency detected
6.15.0-rc6-build2+ #1293 Not tainted
------------------------------------------------------
(udev-worker)/4602 is trying to acquire lock:
ffff88810c5585c0 (&q->elevator_lock){+.+.}-{4:4}, at: elv_iosched_store+0xf9/0x210
but task is already holding lock:
ffff88810c5580a8 (&q->q_usage_counter(io)#9){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0xf/0x20
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #2 (&q->q_usage_counter(io)#9){++++}-{0:0}:
validate_chain+0x1dc/0x280
__lock_acquire+0x5b6/0x720
lock_acquire.part.0+0xb4/0x1f0
blk_alloc_queue+0x38b/0x420
blk_mq_alloc_queue+0xd4/0x150
scsi_alloc_sdev+0x47e/0x5e0
scsi_probe_and_add_lun+0x15c/0x370
__scsi_add_device+0x123/0x190
ata_scsi_scan_host+0x94/0x1d0
async_run_entry_fn+0x4b/0x130
process_one_work+0x485/0x7b0
process_scheduled_works+0x73/0x90
worker_thread+0x1c8/0x2a0
kthread+0x2f9/0x310
ret_from_fork+0x24/0x40
ret_from_fork_asm+0x1a/0x30
-> #1 (fs_reclaim){+.+.}-{0:0}:
validate_chain+0x1dc/0x280
__lock_acquire+0x5b6/0x720
lock_acquire.part.0+0xb4/0x1f0
__fs_reclaim_acquire+0x21/0x30
fs_reclaim_acquire+0x2d/0x70
might_alloc+0x8/0x40
kmem_cache_alloc_noprof+0x42/0x230
__kernfs_new_node+0xc6/0x3e0
kernfs_new_node+0x89/0xc0
kernfs_create_dir_ns+0x27/0xa0
sysfs_create_dir_ns+0xf5/0x170
kobject_add_internal+0x141/0x2c0
kobject_add+0xfd/0x140
elv_register_queue+0x74/0x100
blk_register_queue+0x16a/0x240
add_disk_fwnode+0x371/0x710
sd_probe+0x50d/0x620
really_probe+0x167/0x320
__driver_probe_device+0x121/0x160
driver_probe_device+0x4a/0xd0
__device_attach_driver+0x99/0xd0
bus_for_each_drv+0x104/0x140
__device_attach_async_helper+0xdb/0x140
async_run_entry_fn+0x4b/0x130
process_one_work+0x485/0x7b0
process_scheduled_works+0x73/0x90
worker_thread+0x1c8/0x2a0
kthread+0x2f9/0x310
ret_from_fork+0x24/0x40
ret_from_fork_asm+0x1a/0x30
-> #0 (&q->elevator_lock){+.+.}-{4:4}:
check_noncircular+0x96/0xc0
check_prev_add+0x115/0x2f0
validate_chain+0x1dc/0x280
__lock_acquire+0x5b6/0x720
lock_acquire.part.0+0xb4/0x1f0
__mutex_lock+0x16d/0x5e0
elv_iosched_store+0xf9/0x210
queue_attr_store+0xcb/0x1c0
kernfs_fop_write_iter+0x194/0x210
vfs_write+0x220/0x310
ksys_write+0xb8/0x120
do_syscall_64+0x9f/0x100
entry_SYSCALL_64_after_hwframe+0x76/0x7e
other info that might help us debug this:
Chain exists of:
&q->elevator_lock --> fs_reclaim --> &q->q_usage_counter(io)#9
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&q->q_usage_counter(io)#9);
lock(fs_reclaim);
lock(&q->q_usage_counter(io)#9);
lock(&q->elevator_lock);
*** DEADLOCK ***
5 locks held by (udev-worker)/4602:
#0: ffff88810e4a4408 (sb_writers#4){.+.+}-{0:0}, at: vfs_write+0xfc/0x310
#1: ffff888112e2b888 (&of->mutex#2){+.+.}-{4:4}, at: kernfs_fop_write_iter+0x125/0x210
#2: ffff88810c8062d8 (kn->active#55){.+.+}-{0:0}, at: kernfs_fop_write_iter+0x135/0x210
#3: ffff88810c5580a8 (&q->q_usage_counter(io)#9){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0xf/0x20
#4: ffff88810c5580e0 (&q->q_usage_counter(queue)){++++}-{0:0}, at: blk_mq_freeze_queue_nomemsave+0xf/0x20
stack backtrace:
CPU: 3 UID: 0 PID: 4602 Comm: (udev-worker) Not tainted 6.15.0-rc6-build2+ #1293 PREEMPT(voluntary)
Hardware name: ASUS All Series/H97-PLUS, BIOS 2306 10/09/2014
Call Trace:
<TASK>
dump_stack_lvl+0x57/0x80
print_circular_bug+0xb5/0xd0
check_noncircular+0x96/0xc0
? _find_first_zero_bit+0x1e/0x50
check_prev_add+0x115/0x2f0
validate_chain+0x1dc/0x280
__lock_acquire+0x5b6/0x720
lock_acquire.part.0+0xb4/0x1f0
? elv_iosched_store+0xf9/0x210
? rcu_is_watching+0x34/0x60
? lock_acquire+0x88/0xf0
__mutex_lock+0x16d/0x5e0
? elv_iosched_store+0xf9/0x210
? preempt_count_sub+0x18/0xc0
? elv_iosched_store+0xf9/0x210
? __pfx___mutex_lock+0x10/0x10
? blk_mq_freeze_queue_wait+0xe6/0x130
? lock_acquire.part.0+0xc4/0x1f0
? __pfx_autoremove_wake_function+0x10/0x10
? lock_acquire+0x88/0xf0
? elv_iosched_store+0xf9/0x210
elv_iosched_store+0xf9/0x210
? __pfx_elv_iosched_store+0x10/0x10
? __pfx___mutex_trylock_common+0x10/0x10
queue_attr_store+0xcb/0x1c0
? mark_lock+0x2e/0x110
? __pfx_queue_attr_store+0x10/0x10
? __lock_acquire+0x5b6/0x720
? lock_is_held_type+0xbe/0x110
? find_held_lock+0x2b/0x80
? sysfs_file_kobj+0x129/0x140
? __lock_release.isra.0+0x5a/0x150
? sysfs_file_kobj+0x129/0x140
? lock_release+0xe3/0x120
? __rcu_read_unlock+0x4c/0x70
? sysfs_file_kobj+0x133/0x140
? __pfx_sysfs_kf_write+0x10/0x10
kernfs_fop_write_iter+0x194/0x210
vfs_write+0x220/0x310
? __pfx_vfs_write+0x10/0x10
? ktime_get_coarse_real_ts64+0x19/0x70
? files_lookup_fd_raw+0x40/0x50
? __fget_light+0x5b/0x90
ksys_write+0xb8/0x120
? __pfx_ksys_write+0x10/0x10
? syscall_trace_enter+0x10c/0x150
do_syscall_64+0x9f/0x100
entry_SYSCALL_64_after_hwframe+0x76/0x7e
RIP: 0033:0x7f033d80e044
Code: c7 00 16 00 00 00 b8 ff ff ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 f3 0f 1e fa 80 3d 65 a0 10 00 00 74 13 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 54 c3 0f 1f 00 55 48 89 e5 48 83 ec 20 48 89
RSP: 002b:00007ffe41add9f8 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 00007f033d80e044
RDX: 0000000000000003 RSI: 00007ffe41addd00 RDI: 0000000000000027
RBP: 00007ffe41adda20 R08: 00007f033d90f1c8 R09: 00007ffe41addad0
R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000003
R13: 00007ffe41addd00 R14: 00005626a9b65980 R15: 00007f033d90ee80
</TASK>
