Re: [syzbot] possible deadlock in __fs_reclaim_acquire
From: syzbot
Date: Sun Jul 04 2021 - 07:51:36 EST
syzbot has found a reproducer for the following issue on:
HEAD commit: 3dbdb38e Merge branch 'for-5.14' of git://git.kernel.org/p..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=17e0b9d8300000
kernel config: https://syzkaller.appspot.com/x/.config?x=1700b0b2b41cd52c
dashboard link: https://syzkaller.appspot.com/bug?extid=127fd7828d6eeb611703
compiler: Debian clang version 11.0.1-2
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=10542f52300000
IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+127fd7828d6eeb611703@xxxxxxxxxxxxxxxxxxxxxxxxx
======================================================
WARNING: possible circular locking dependency detected
5.13.0-syzkaller #0 Not tainted
------------------------------------------------------
syz-execprog/8416 is trying to acquire lock:
ffffffff8cfd6720 (fs_reclaim){+.+.}-{0:0}, at: __fs_reclaim_acquire+0x0/0x30 mm/page_alloc.c:4222
but task is already holding lock:
ffff8880b9b31088 (lock#2){-.-.}-{2:2}, at: local_lock_acquire+0x7/0x130 include/linux/local_lock_internal.h:41
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (lock#2){-.-.}-{2:2}:
lock_acquire+0x182/0x4a0 kernel/locking/lockdep.c:5625
local_lock_acquire+0x23/0x130 include/linux/local_lock_internal.h:42
free_unref_page+0x242/0x550 mm/page_alloc.c:3439
mm_free_pgd kernel/fork.c:636 [inline]
__mmdrop+0xae/0x3f0 kernel/fork.c:687
mmdrop include/linux/sched/mm.h:49 [inline]
finish_task_switch+0x221/0x630 kernel/sched/core.c:4582
context_switch kernel/sched/core.c:4686 [inline]
__schedule+0xc0f/0x11f0 kernel/sched/core.c:5940
preempt_schedule_notrace+0x12c/0x170 kernel/sched/core.c:6179
preempt_schedule_notrace_thunk+0x16/0x18 arch/x86/entry/thunk_64.S:36
rcu_read_unlock_sched_notrace include/linux/rcupdate.h:809 [inline]
trace_lock_release+0x9f/0x140 include/trace/events/lock.h:58
lock_release+0x81/0x7b0 kernel/locking/lockdep.c:5636
might_alloc include/linux/sched/mm.h:199 [inline]
slab_pre_alloc_hook mm/slab.h:485 [inline]
slab_alloc_node mm/slub.c:2891 [inline]
slab_alloc mm/slub.c:2978 [inline]
kmem_cache_alloc+0x41/0x340 mm/slub.c:2983
kmem_cache_zalloc include/linux/slab.h:711 [inline]
attach_epitem fs/eventpoll.c:1414 [inline]
ep_insert fs/eventpoll.c:1468 [inline]
do_epoll_ctl+0x13a7/0x2f70 fs/eventpoll.c:2133
__do_sys_epoll_ctl fs/eventpoll.c:2184 [inline]
__se_sys_epoll_ctl fs/eventpoll.c:2175 [inline]
__x64_sys_epoll_ctl+0x14e/0x190 fs/eventpoll.c:2175
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
-> #0 (fs_reclaim){+.+.}-{0:0}:
check_prev_add kernel/locking/lockdep.c:3051 [inline]
check_prevs_add+0x4f9/0x5b30 kernel/locking/lockdep.c:3174
validate_chain kernel/locking/lockdep.c:3789 [inline]
__lock_acquire+0x4476/0x6100 kernel/locking/lockdep.c:5015
lock_acquire+0x182/0x4a0 kernel/locking/lockdep.c:5625
__fs_reclaim_acquire+0x20/0x30 mm/page_alloc.c:4564
fs_reclaim_acquire+0x59/0xf0 mm/page_alloc.c:4578
prepare_alloc_pages+0x151/0x5a0 mm/page_alloc.c:5176
__alloc_pages+0x14d/0x5f0 mm/page_alloc.c:5375
stack_depot_save+0x361/0x490 lib/stackdepot.c:303
save_stack+0xf9/0x1f0 mm/page_owner.c:120
__set_page_owner+0x42/0x2f0 mm/page_owner.c:181
prep_new_page mm/page_alloc.c:2445 [inline]
__alloc_pages_bulk+0x9f2/0x1090 mm/page_alloc.c:5313
alloc_pages_bulk_array_node include/linux/gfp.h:557 [inline]
vm_area_alloc_pages mm/vmalloc.c:2775 [inline]
__vmalloc_area_node mm/vmalloc.c:2845 [inline]
__vmalloc_node_range+0x3ad/0x7f0 mm/vmalloc.c:2947
vmalloc_user+0x70/0x80 mm/vmalloc.c:3082
kcov_mmap+0x28/0x130 kernel/kcov.c:465
call_mmap include/linux/fs.h:2119 [inline]
mmap_region+0x1410/0x1df0 mm/mmap.c:1809
do_mmap+0x930/0x11a0 mm/mmap.c:1585
vm_mmap_pgoff+0x19e/0x2b0 mm/util.c:519
ksys_mmap_pgoff+0x504/0x7b0 mm/mmap.c:1636
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(lock#2);
lock(fs_reclaim);
lock(lock#2);
lock(fs_reclaim);
*** DEADLOCK ***
2 locks held by syz-execprog/8416:
#0: ffff8880161e0128 (&mm->mmap_lock#2){++++}-{3:3}, at: mmap_write_lock_killable include/linux/mmap_lock.h:87 [inline]
#0: ffff8880161e0128 (&mm->mmap_lock#2){++++}-{3:3}, at: vm_mmap_pgoff+0x14d/0x2b0 mm/util.c:517
#1: ffff8880b9b31088 (lock#2){-.-.}-{2:2}, at: local_lock_acquire+0x7/0x130 include/linux/local_lock_internal.h:41
stack backtrace:
CPU: 1 PID: 8416 Comm: syz-execprog Not tainted 5.13.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:79 [inline]
dump_stack_lvl+0x1ae/0x29f lib/dump_stack.c:96
print_circular_bug+0xb17/0xdc0 kernel/locking/lockdep.c:2009
check_noncircular+0x2cc/0x390 kernel/locking/lockdep.c:2131
check_prev_add kernel/locking/lockdep.c:3051 [inline]
check_prevs_add+0x4f9/0x5b30 kernel/locking/lockdep.c:3174
validate_chain kernel/locking/lockdep.c:3789 [inline]
__lock_acquire+0x4476/0x6100 kernel/locking/lockdep.c:5015
lock_acquire+0x182/0x4a0 kernel/locking/lockdep.c:5625
__fs_reclaim_acquire+0x20/0x30 mm/page_alloc.c:4564
fs_reclaim_acquire+0x59/0xf0 mm/page_alloc.c:4578
prepare_alloc_pages+0x151/0x5a0 mm/page_alloc.c:5176
__alloc_pages+0x14d/0x5f0 mm/page_alloc.c:5375
stack_depot_save+0x361/0x490 lib/stackdepot.c:303
save_stack+0xf9/0x1f0 mm/page_owner.c:120
__set_page_owner+0x42/0x2f0 mm/page_owner.c:181
prep_new_page mm/page_alloc.c:2445 [inline]
__alloc_pages_bulk+0x9f2/0x1090 mm/page_alloc.c:5313
alloc_pages_bulk_array_node include/linux/gfp.h:557 [inline]
vm_area_alloc_pages mm/vmalloc.c:2775 [inline]
__vmalloc_area_node mm/vmalloc.c:2845 [inline]
__vmalloc_node_range+0x3ad/0x7f0 mm/vmalloc.c:2947
vmalloc_user+0x70/0x80 mm/vmalloc.c:3082
kcov_mmap+0x28/0x130 kernel/kcov.c:465
call_mmap include/linux/fs.h:2119 [inline]
mmap_region+0x1410/0x1df0 mm/mmap.c:1809
do_mmap+0x930/0x11a0 mm/mmap.c:1585
vm_mmap_pgoff+0x19e/0x2b0 mm/util.c:519
ksys_mmap_pgoff+0x504/0x7b0 mm/mmap.c:1636
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x4b132a
Code: e8 db 57 fb ff 48 8b 7c 24 10 48 8b 74 24 18 48 8b 54 24 20 4c 8b 54 24 28 4c 8b 44 24 30 4c 8b 4c 24 38 48 8b 44 24 08 0f 05 <48> 3d 01 f0 ff ff 76 20 48 c7 44 24 40 ff ff ff ff 48 c7 44 24 48
RSP: 002b:000000c000173a10 EFLAGS: 00000202 ORIG_RAX: 0000000000000009
RAX: ffffffffffffffda RBX: 000000c000020800 RCX: 00000000004b132a
RDX: 0000000000000003 RSI: 0000000000080000 RDI: 0000000000000000
RBP: 000000c000173a70 R08: 0000000000000006 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000202 R12: 0000000000727f1a
R13: 00000000000001f6 R14: 0000000000000200 R15: 0000000000000100
BUG: sleeping function called from invalid context at mm/page_alloc.c:5179
in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 8416, name: syz-execprog
INFO: lockdep is turned off.
irq event stamp: 70646
hardirqs last enabled at (70645): [<ffffffff89cf038b>] __raw_spin_unlock_irqrestore include/linux/spinlock_api_smp.h:160 [inline]
hardirqs last enabled at (70645): [<ffffffff89cf038b>] _raw_spin_unlock_irqrestore+0x8b/0x120 kernel/locking/spinlock.c:191
hardirqs last disabled at (70646): [<ffffffff81be4351>] __alloc_pages_bulk+0x801/0x1090 mm/page_alloc.c:5291
softirqs last enabled at (69738): [<ffffffff814d4fbb>] invoke_softirq kernel/softirq.c:432 [inline]
softirqs last enabled at (69738): [<ffffffff814d4fbb>] __irq_exit_rcu+0x21b/0x260 kernel/softirq.c:636
softirqs last disabled at (69687): [<ffffffff814d4fbb>] invoke_softirq kernel/softirq.c:432 [inline]
softirqs last disabled at (69687): [<ffffffff814d4fbb>] __irq_exit_rcu+0x21b/0x260 kernel/softirq.c:636
CPU: 1 PID: 8416 Comm: syz-execprog Not tainted 5.13.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:79 [inline]
dump_stack_lvl+0x1ae/0x29f lib/dump_stack.c:96
___might_sleep+0x4e5/0x6b0 kernel/sched/core.c:9153
prepare_alloc_pages+0x1c0/0x5a0 mm/page_alloc.c:5179
__alloc_pages+0x14d/0x5f0 mm/page_alloc.c:5375
stack_depot_save+0x361/0x490 lib/stackdepot.c:303
save_stack+0xf9/0x1f0 mm/page_owner.c:120
__set_page_owner+0x42/0x2f0 mm/page_owner.c:181
prep_new_page mm/page_alloc.c:2445 [inline]
__alloc_pages_bulk+0x9f2/0x1090 mm/page_alloc.c:5313
alloc_pages_bulk_array_node include/linux/gfp.h:557 [inline]
vm_area_alloc_pages mm/vmalloc.c:2775 [inline]
__vmalloc_area_node mm/vmalloc.c:2845 [inline]
__vmalloc_node_range+0x3ad/0x7f0 mm/vmalloc.c:2947
vmalloc_user+0x70/0x80 mm/vmalloc.c:3082
kcov_mmap+0x28/0x130 kernel/kcov.c:465
call_mmap include/linux/fs.h:2119 [inline]
mmap_region+0x1410/0x1df0 mm/mmap.c:1809
do_mmap+0x930/0x11a0 mm/mmap.c:1585
vm_mmap_pgoff+0x19e/0x2b0 mm/util.c:519
ksys_mmap_pgoff+0x504/0x7b0 mm/mmap.c:1636
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x4b132a
Code: e8 db 57 fb ff 48 8b 7c 24 10 48 8b 74 24 18 48 8b 54 24 20 4c 8b 54 24 28 4c 8b 44 24 30 4c 8b 4c 24 38 48 8b 44 24 08 0f 05 <48> 3d 01 f0 ff ff 76 20 48 c7 44 24 40 ff ff ff ff 48 c7 44 24 48
RSP: 002b:000000c000173a10 EFLAGS: 00000202 ORIG_RAX: 0000000000000009
RAX: ffffffffffffffda RBX: 000000c000020800 RCX: 00000000004b132a
RDX: 0000000000000003 RSI: 0000000000080000 RDI: 0000000000000000
RBP: 000000c000173a70 R08: 0000000000000006 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000202 R12: 0000000000727f1a
R13: 00000000000001f6 R14: 0000000000000200 R15: 0000000000000100