lkml.org 
[lkml]   [2022]   [Sep]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
Subjectnext: possible circular locking dependency: perf vs printk
Hello folks,

The lockdep splat below is pretty interesting. If I understand it
correctly:

a) perf invokes printk() under ctx lock. printk invokes up() on the
semapthore, calls into the scheduler, which takes rq and pi locks.

Then we have a reverse path:

b) scheduler takes rq lock and then invokes perf, which takes ctx lock.


[ 14.301661] WARNING: possible circular locking dependency detected
[ 14.301662] 6.0.0-rc5-next-20220914+ #79 Tainted: G N
[ 14.301666] ------------------------------------------------------
[ 14.301667] perf-exec/281 is trying to acquire lock:
[ 14.301669] ffffffff82340cf8 ((console_sem).lock){-.-.}-{2:2}, at: down_trylock+0xf/0x30
[ 14.301696]
[ 14.301696] but task is already holding lock:
[ 14.301697] ffff888100b43020 (&ctx->lock){....}-{2:2}, at: perf_event_exec+0xa9/0x590
[ 14.301714]
[ 14.301714] which lock already depends on the new lock.
[ 14.301714]
[ 14.301715]
[ 14.301715] the existing dependency chain (in reverse order) is:
[ 14.301717]
[ 14.301717] -> #3 (&ctx->lock){....}-{2:2}:
[ 14.301722] __lock_acquire+0x4b7/0xa00
[ 14.301728] lock_acquire+0xb1/0x290
[ 14.301733] _raw_spin_lock+0x2c/0x40
[ 14.301743] perf_event_context_sched_out+0x143/0x720
[ 14.301750] __perf_event_task_sched_out+0x50/0x140
[ 14.301757] prepare_task_switch+0x8a/0x1a0
[ 14.301765] __schedule+0x168/0x640
[ 14.301770] schedule+0x57/0xa0
[ 14.301775] io_schedule+0x42/0x70
[ 14.301781] folio_wait_bit_common+0x127/0x300
[ 14.301787] filemap_update_page+0x23b/0x260
[ 14.301793] filemap_get_pages+0x20a/0x340
[ 14.301799] filemap_read+0xc6/0x340
[ 14.301805] __kernel_read+0xe8/0x240
[ 14.301813] search_binary_handler+0x71/0x240
[ 14.301821] exec_binprm+0xac/0x2e0
[ 14.301827] bprm_execve+0x1ac/0x330
[ 14.301833] do_execveat_common.isra.0+0x1ad/0x220
[ 14.301839] __x64_sys_execve+0x32/0x40
[ 14.301846] do_syscall_64+0x34/0x80
[ 14.301854] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 14.301859]
[ 14.301859] -> #2 (&rq->__lock){-.-.}-{2:2}:
[ 14.301865] __lock_acquire+0x4b7/0xa00
[ 14.301869] lock_acquire+0xb1/0x290
[ 14.301873] _raw_spin_lock_nested+0x2a/0x40
[ 14.301880] raw_spin_rq_lock_nested+0x11/0x20
[ 14.301886] task_fork_fair+0x43/0x150
[ 14.301890] sched_cgroup_fork+0xe3/0x100
[ 14.301897] copy_process+0xf42/0x1730
[ 14.301905] kernel_clone+0x9b/0x410
[ 14.301910] user_mode_thread+0x5b/0x80
[ 14.301915] rest_init+0x1e/0x160
[ 14.301920] arch_call_rest_init+0xa/0x10
[ 14.301938] start_kernel+0x5f9/0x61e
[ 14.301943] secondary_startup_64_no_verify+0xce/0xdb
[ 14.301948]
[ 14.301948] -> #1 (&p->pi_lock){-.-.}-{2:2}:
[ 14.301954] __lock_acquire+0x4b7/0xa00
[ 14.301958] lock_acquire+0xb1/0x290
[ 14.301962] _raw_spin_lock_irqsave+0x38/0x50
[ 14.301969] try_to_wake_up+0x53/0x4a0
[ 14.301976] up+0x40/0x60
[ 14.301982] __up_console_sem+0x48/0x60
[ 14.301988] console_unlock+0x132/0x210
[ 14.301994] vprintk_emit+0x12e/0x310
[ 14.301999] _printk+0x58/0x6f
[ 14.302006] addrconf_notify.cold+0x21/0x26
[ 14.302014] raw_notifier_call_chain+0x41/0x60
[ 14.302021] netdev_state_change+0x67/0x90
[ 14.302028] linkwatch_do_dev+0x54/0x70
[ 14.302038] __linkwatch_run_queue+0xe9/0x200
[ 14.302043] linkwatch_event+0x21/0x30
[ 14.302047] process_one_work+0x294/0x560
[ 14.302052] worker_thread+0x4c/0x3c0
[ 14.302056] kthread+0xd4/0x100
[ 14.302063] ret_from_fork+0x1f/0x30
[ 14.302067]
[ 14.302067] -> #0 ((console_sem).lock){-.-.}-{2:2}:
[ 14.302072] check_prev_add+0x8f/0xbd0
[ 14.302076] validate_chain+0x571/0x830
[ 14.302080] __lock_acquire+0x4b7/0xa00
[ 14.302084] lock_acquire+0xb1/0x290
[ 14.302088] _raw_spin_lock_irqsave+0x38/0x50
[ 14.302095] down_trylock+0xf/0x30
[ 14.302102] __down_trylock_console_sem+0x23/0x90
[ 14.302108] vprintk_emit+0x107/0x310
[ 14.302114] _printk+0x58/0x6f
[ 14.302119] report_bug.cold+0x10/0x4b
[ 14.302126] handle_bug+0x3c/0x60
[ 14.302133] exc_invalid_op+0x13/0x60
[ 14.302137] asm_exc_invalid_op+0x16/0x20
[ 14.302141] perf_event_update_sibling_time+0x2d/0x100
[ 14.302147] perf_event_set_state+0x8e/0xa0
[ 14.302152] perf_event_exec+0x101/0x590
[ 14.302159] begin_new_exec+0x2d8/0x410
[ 14.302165] load_elf_binary+0x6c4/0xec0
[ 14.302171] search_binary_handler+0xbb/0x240
[ 14.302176] exec_binprm+0xac/0x2e0
[ 14.302183] bprm_execve+0x1ac/0x330
[ 14.302188] do_execveat_common.isra.0+0x1ad/0x220
[ 14.302195] __x64_sys_execve+0x32/0x40
[ 14.302201] do_syscall_64+0x34/0x80
[ 14.302208] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 14.302217]
[ 14.302217] other info that might help us debug this:
[ 14.302217]
[ 14.302218] Chain exists of:
[ 14.302218] (console_sem).lock --> &rq->__lock --> &ctx->lock
[ 14.302218]
[ 14.302224] Possible unsafe locking scenario:
[ 14.302224]
[ 14.302225] CPU0 CPU1
[ 14.302226] ---- ----
[ 14.302227] lock(&ctx->lock);
[ 14.302229] lock(&rq->__lock);
[ 14.302232] lock(&ctx->lock);
[ 14.302234] lock((console_sem).lock);
[ 14.302236]
[ 14.302236] *** DEADLOCK ***
[ 14.302236]
[ 14.302237] 4 locks held by perf-exec/281:
[ 14.302241] #0: ffff888105179808 (&sig->cred_guard_mutex){+.+.}-{3:3}, at: bprm_execve+0x38/0x330
[ 14.302253] #1: ffff8881051798a0 (&sig->exec_update_lock){++++}-{3:3}, at: exec_mmap+0x55/0x220
[ 14.302264] #2: ffff88813bbf0a78 (&cpuctx_lock){....}-{2:2}, at: perf_event_exec+0x98/0x590
[ 14.302276] #3: ffff888100b43020 (&ctx->lock){....}-{2:2}, at: perf_event_exec+0xa9/0x590
[ 14.302287]
[ 14.302287] stack backtrace:
[ 14.302289] CPU: 7 PID: 281 Comm: perf-exec Tainted: G N 6.0.0-rc5-next-20220914+ #79
[ 14.302296] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-debian-1.16.0-4 04/01/2014
[ 14.302303] Call Trace:
[ 14.302306] <TASK>
[ 14.302309] dump_stack_lvl+0x4c/0x5f
[ 14.302318] check_noncircular+0xfe/0x110
[ 14.302325] ? lock_acquire+0xb1/0x290
[ 14.302332] check_prev_add+0x8f/0xbd0
[ 14.302336] ? rcu_read_lock_sched_held+0x3b/0x70
[ 14.302346] validate_chain+0x571/0x830
[ 14.302353] __lock_acquire+0x4b7/0xa00
[ 14.302361] lock_acquire+0xb1/0x290
[ 14.302366] ? down_trylock+0xf/0x30
[ 14.302377] _raw_spin_lock_irqsave+0x38/0x50
[ 14.302385] ? down_trylock+0xf/0x30
[ 14.302392] down_trylock+0xf/0x30
[ 14.302399] ? _printk+0x58/0x6f
[ 14.302404] __down_trylock_console_sem+0x23/0x90
[ 14.302411] vprintk_emit+0x107/0x310
[ 14.302418] ? perf_event_update_sibling_time+0x2d/0x100
[ 14.302425] _printk+0x58/0x6f
[ 14.302432] ? restore_regs_and_return_to_kernel+0x22/0x22
[ 14.302439] report_bug.cold+0x10/0x4b
[ 14.302448] handle_bug+0x3c/0x60
[ 14.302456] exc_invalid_op+0x13/0x60
[ 14.302461] asm_exc_invalid_op+0x16/0x20
[ 14.302466] RIP: 0010:perf_event_update_sibling_time+0x2d/0x100
[ 14.302472] Code: 00 00 8b 0d 61 25 2c 01 53 48 89 fb 85 c9 74 1c 65 8b 05 16 92 e8 7e 85 c0 75 11 65 8b 05 27 90 e8 7e 85 c0 0f 84 a0 00 00 00 <0f> 0b 48 39 9b 90 00 00 00 74 02 5b c3 48 8b 43 10 48 83 e8 10 48
[ 14.302477] RSP: 0018:ffffc90000defca8 EFLAGS: 00010097
[ 14.302482] RAX: 0000000000000000 RBX: ffff888109602fb0 RCX: 0000000000000000
[ 14.302486] RDX: 0000000000000000 RSI: ffff888100b430b0 RDI: ffff888105174688
[ 14.302489] RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000002
[ 14.302492] R10: ffffc90000defb28 R11: 0000000000033c20 R12: ffff888109602fb0
[ 14.302495] R13: 0000000000000000 R14: ffff888100b43108 R15: ffff888105173e00
[ 14.302505] ? perf_event_update_sibling_time+0xe5/0x100
[ 14.302511] perf_event_set_state+0x8e/0xa0
[ 14.302517] perf_event_exec+0x101/0x590
[ 14.302529] begin_new_exec+0x2d8/0x410
[ 14.302535] ? load_elf_phdrs+0x6c/0xb0
[ 14.302541] load_elf_binary+0x6c4/0xec0
[ 14.302547] ? find_held_lock+0x2b/0x80
[ 14.302555] ? search_binary_handler+0xb5/0x240
[ 14.302561] ? __lock_release.isra.0+0x61/0x150
[ 14.302566] ? search_binary_handler+0xb5/0x240
[ 14.302572] ? search_binary_handler+0xb5/0x240
[ 14.302581] search_binary_handler+0xbb/0x240
[ 14.302588] exec_binprm+0xac/0x2e0
[ 14.302596] bprm_execve+0x1ac/0x330
[ 14.302604] do_execveat_common.isra.0+0x1ad/0x220
[ 14.302613] __x64_sys_execve+0x32/0x40
[ 14.302621] do_syscall_64+0x34/0x80
[ 14.302629] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 14.302633] RIP: 0033:0x7f3d4310c3c7
[ 14.302663] Code: Unable to access opcode bytes at RIP 0x7f3d4310c39d.
[ 14.302665] RSP: 002b:00007ffd90454038 EFLAGS: 00000202 ORIG_RAX: 000000000000003b
[ 14.302671] RAX: ffffffffffffffda RBX: 0000000000000005 RCX: 00007f3d4310c3c7
[ 14.302674] RDX: 00005644b06bfc70 RSI: 00007ffd904592c0 RDI: 00007ffd90454040
[ 14.302677] RBP: 00007ffd90454120 R08: 0000000000000fff R09: 00005644b06c00ce
[ 14.302680] R10: 00007f3d430466d0 R11: 0000000000000202 R12: 00007ffd904592c0
[ 14.302683] R13: 00005644b06bfc70 R14: 00007ffd9045ae4e R15: 00005644b06c00c6
[ 14.302693] </TASK>
[ 14.512851] WARNING: CPU: 7 PID: 281 at kernel/events/core.c:655 perf_event_update_sibling_time+0x2d/0x100
[ 14.515326] Modules linked in:
[ 14.516128] CPU: 7 PID: 281 Comm: perf-exec Tainted: G N 6.0.0-rc5-next-20220914+ #79
[ 14.518499] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-debian-1.16.0-4 04/01/2014
[ 14.520964] RIP: 0010:perf_event_update_sibling_time+0x2d/0x100
[ 14.522497] Code: 00 00 8b 0d 61 25 2c 01 53 48 89 fb 85 c9 74 1c 65 8b 05 16 92 e8 7e 85 c0 75 11 65 8b 05 27 90 e8 7e 85 c0 0f 84 a0 00 00 00 <0f> 0b 48 39 9b 90 00 00 00 74 02 5b c3 48 8b 43 10 48 83 e8 10 48
[ 14.527303] RSP: 0018:ffffc90000defca8 EFLAGS: 00010097
[ 14.528668] RAX: 0000000000000000 RBX: ffff888109602fb0 RCX: 0000000000000000
[ 14.530499] RDX: 0000000000000000 RSI: ffff888100b430b0 RDI: ffff888105174688
[ 14.532345] RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000002
[ 14.534168] R10: ffffc90000defb28 R11: 0000000000033c20 R12: ffff888109602fb0
[ 14.536021] R13: 0000000000000000 R14: ffff888100b43108 R15: ffff888105173e00
[ 14.537858] FS: 0000000000000000(0000) GS:ffff88813bbc0000(0000) knlGS:0000000000000000
[ 14.539950] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 14.541429] CR2: 00007f3d4310c39d CR3: 0000000106080002 CR4: 0000000000770ee0
[ 14.543269] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 14.545097] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 14.546927] PKRU: 55555554
[ 14.547668] Call Trace:
[ 14.548291] <TASK>
[ 14.548851] perf_event_set_state+0x8e/0xa0
[ 14.549939] perf_event_exec+0x101/0x590
[ 14.550953] begin_new_exec+0x2d8/0x410
[ 14.551968] ? load_elf_phdrs+0x6c/0xb0
[ 14.552975] load_elf_binary+0x6c4/0xec0
[ 14.553997] ? find_held_lock+0x2b/0x80
[ 14.555000] ? search_binary_handler+0xb5/0x240
[ 14.556188] ? __lock_release.isra.0+0x61/0x150
[ 14.557371] ? search_binary_handler+0xb5/0x240
[ 14.558540] ? search_binary_handler+0xb5/0x240
[ 14.559725] search_binary_handler+0xbb/0x240
[ 14.560869] exec_binprm+0xac/0x2e0
[ 14.561778] bprm_execve+0x1ac/0x330
[ 14.562705] do_execveat_common.isra.0+0x1ad/0x220
[ 14.563973] __x64_sys_execve+0x32/0x40
[ 14.564964] do_syscall_64+0x34/0x80
[ 14.565902] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 14.567214] RIP: 0033:0x7f3d4310c3c7
[ 14.568168] Code: Unable to access opcode bytes at RIP 0x7f3d4310c39d.
[ 14.569857] RSP: 002b:00007ffd90454038 EFLAGS: 00000202 ORIG_RAX: 000000000000003b
[ 14.571817] RAX: ffffffffffffffda RBX: 0000000000000005 RCX: 00007f3d4310c3c7
[ 14.573656] RDX: 00005644b06bfc70 RSI: 00007ffd904592c0 RDI: 00007ffd90454040
[ 14.575510] RBP: 00007ffd90454120 R08: 0000000000000fff R09: 00005644b06c00ce
[ 14.577334] R10: 00007f3d430466d0 R11: 0000000000000202 R12: 00007ffd904592c0
[ 14.579152] R13: 00005644b06bfc70 R14: 00007ffd9045ae4e R15: 00005644b06c00c6
[ 14.580994] </TASK>
[ 14.581573] irq event stamp: 692
[ 14.582416] hardirqs last enabled at (691): [<ffffffff8106e474>] __local_bh_enable_ip+0x74/0xc0
[ 14.584704] hardirqs last disabled at (692): [<ffffffff811a120e>] perf_event_exec+0x3fe/0x590
[ 14.586886] softirqs last enabled at (690): [<ffffffff8103018b>] fpu_flush_thread+0xbb/0x100
[ 14.589082] softirqs last disabled at (688): [<ffffffff810300d5>] fpu_flush_thread+0x5/0x100
[ 14.591246] ---[ end trace 0000000000000000 ]---

\
 
 \ /
  Last update: 2022-09-14 12:23    [W:0.051 / U:0.284 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site