possible deadlock in userfaultfd_release

[syzbot]

Hello,

syzbot found the following crash on:

HEAD commit: 7aa823a9 Linux 4.19.51
git tree: linux-4.19.y
console output: https://syzkaller.appspot.com/x/log.txt?x=15e5e42aa00000
kernel config: https://syzkaller.appspot.com/x/.config?x=57222bc62f28e3a6
dashboard link: https://syzkaller.appspot.com/bug?extid=0b8608e3d25d48fc4f4c
compiler: gcc (GCC) 9.0.0 20181231 (experimental)
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=10d9f956a00000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=10737eaea00000

IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+0b8608...@syzkaller.appspotmail.com

audit: type=1400 audit(1560715658.504:36): avc: denied { map } for
pid=8029 comm="syz-executor068" path="/root/syz-executor068989011"
dev="sda1" ino=16484 scontext=unconfined_u:system_r:insmod_t:s0-s0:c0.c1023
tcontext=unconfined_u:object_r:user_home_t:s0 tclass=file permissive=1
========================================================
WARNING: possible irq lock inversion dependency detected
4.19.51 #23 Not tainted
--------------------------------------------------------
syz-executor068/8031 just changed the state of lock:
000000004926c8ee (&ctx->fault_pending_wqh){+.+.}, at: spin_lock
include/linux/spinlock.h:329 [inline]
000000004926c8ee (&ctx->fault_pending_wqh){+.+.}, at:
userfaultfd_release+0x4d6/0x720 fs/userfaultfd.c:922
but this lock was taken by another, SOFTIRQ-safe lock in the past:
(&(&ctx->ctx_lock)->rlock){..-.}


and interrupts could create inverse lock ordering between them.


other info that might help us debug this:
Chain exists of:
&(&ctx->ctx_lock)->rlock --> &ctx->fd_wqh --> &ctx->fault_pending_wqh

Possible interrupt unsafe locking scenario:

CPU0 CPU1
---- ----
lock(&ctx->fault_pending_wqh);
local_irq_disable();
lock(&(&ctx->ctx_lock)->rlock);
lock(&ctx->fd_wqh);
<Interrupt>
lock(&(&ctx->ctx_lock)->rlock);

*** DEADLOCK ***

no locks held by syz-executor068/8031.

the shortest dependencies between 2nd lock and 1st lock:
-> (&(&ctx->ctx_lock)->rlock){..-.} ops: 2 {
IN-SOFTIRQ-W at:
lock_acquire+0x16f/0x3f0
kernel/locking/lockdep.c:3900
__raw_spin_lock_irq
include/linux/spinlock_api_smp.h:128 [inline]
_raw_spin_lock_irq+0x60/0x80
kernel/locking/spinlock.c:160
spin_lock_irq include/linux/spinlock.h:354 [inline]
free_ioctx_users+0x2d/0x490 fs/aio.c:614
percpu_ref_put_many
include/linux/percpu-refcount.h:284 [inline]
percpu_ref_put include/linux/percpu-refcount.h:300
[inline]
percpu_ref_call_confirm_rcu
lib/percpu-refcount.c:123 [inline]
percpu_ref_switch_to_atomic_rcu+0x407/0x540
lib/percpu-refcount.c:158
__rcu_reclaim kernel/rcu/rcu.h:236 [inline]
rcu_do_batch kernel/rcu/tree.c:2584 [inline]
invoke_rcu_callbacks kernel/rcu/tree.c:2897 [inline]
__rcu_process_callbacks kernel/rcu/tree.c:2864
[inline]
rcu_process_callbacks+0xba0/0x1a30
kernel/rcu/tree.c:2881
__do_softirq+0x25c/0x921 kernel/softirq.c:292
invoke_softirq kernel/softirq.c:372 [inline]
irq_exit+0x180/0x1d0 kernel/softirq.c:412
exiting_irq arch/x86/include/asm/apic.h:536 [inline]
smp_apic_timer_interrupt+0x13b/0x550
arch/x86/kernel/apic/apic.c:1056
apic_timer_interrupt+0xf/0x20
arch/x86/entry/entry_64.S:869
native_safe_halt+0xe/0x10
arch/x86/include/asm/irqflags.h:60
arch_cpu_idle+0xa/0x10 arch/x86/kernel/process.c:556
default_idle_call+0x36/0x90 kernel/sched/idle.c:93
cpuidle_idle_call kernel/sched/idle.c:153 [inline]
do_idle+0x377/0x560 kernel/sched/idle.c:262
cpu_startup_entry+0xc8/0xe0 kernel/sched/idle.c:368
rest_init+0xf1/0xf6 init/main.c:442
start_kernel+0x88c/0x8c5 init/main.c:738
x86_64_start_reservations+0x29/0x2b
arch/x86/kernel/head64.c:470
x86_64_start_kernel+0x77/0x7b
arch/x86/kernel/head64.c:451
secondary_startup_64+0xa4/0xb0
arch/x86/kernel/head_64.S:243
INITIAL USE at:
lock_acquire+0x16f/0x3f0
kernel/locking/lockdep.c:3900
__raw_spin_lock_irq
include/linux/spinlock_api_smp.h:128 [inline]
_raw_spin_lock_irq+0x60/0x80
kernel/locking/spinlock.c:160
spin_lock_irq include/linux/spinlock.h:354 [inline]
aio_poll fs/aio.c:1739 [inline]
__io_submit_one fs/aio.c:1849 [inline]
io_submit_one+0xead/0x2eb0 fs/aio.c:1885
__do_sys_io_submit fs/aio.c:1929 [inline]
__se_sys_io_submit fs/aio.c:1900 [inline]
__x64_sys_io_submit+0x1aa/0x520 fs/aio.c:1900
do_syscall_64+0xfd/0x620 arch/x86/entry/common.c:293
entry_SYSCALL_64_after_hwframe+0x49/0xbe
}
... key at: [<ffffffff8a3813a0>] __key.50192+0x0/0x40
... acquired at:
__raw_spin_lock include/linux/spinlock_api_smp.h:142 [inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
aio_poll fs/aio.c:1741 [inline]
__io_submit_one fs/aio.c:1849 [inline]
io_submit_one+0xef2/0x2eb0 fs/aio.c:1885
__do_sys_io_submit fs/aio.c:1929 [inline]
__se_sys_io_submit fs/aio.c:1900 [inline]
__x64_sys_io_submit+0x1aa/0x520 fs/aio.c:1900
do_syscall_64+0xfd/0x620 arch/x86/entry/common.c:293
entry_SYSCALL_64_after_hwframe+0x49/0xbe

-> (&ctx->fd_wqh){....} ops: 4 {
INITIAL USE at:
lock_acquire+0x16f/0x3f0 kernel/locking/lockdep.c:3900
__raw_spin_lock_irq
include/linux/spinlock_api_smp.h:128 [inline]
_raw_spin_lock_irq+0x60/0x80
kernel/locking/spinlock.c:160
spin_lock_irq include/linux/spinlock.h:354 [inline]
userfaultfd_ctx_read fs/userfaultfd.c:1042 [inline]
userfaultfd_read+0x262/0x18c0 fs/userfaultfd.c:1204
do_loop_readv_writev fs/read_write.c:701 [inline]
do_loop_readv_writev fs/read_write.c:688 [inline]
do_iter_read+0x490/0x640 fs/read_write.c:925
vfs_readv+0xf0/0x160 fs/read_write.c:987
do_readv+0x15e/0x370 fs/read_write.c:1020
__do_sys_readv fs/read_write.c:1107 [inline]
__se_sys_readv fs/read_write.c:1104 [inline]
__x64_sys_readv+0x75/0xb0 fs/read_write.c:1104
do_syscall_64+0xfd/0x620 arch/x86/entry/common.c:293
entry_SYSCALL_64_after_hwframe+0x49/0xbe
}
... key at: [<ffffffff8a381120>] __key.43729+0x0/0x40
... acquired at:
__raw_spin_lock include/linux/spinlock_api_smp.h:142 [inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
userfaultfd_ctx_read fs/userfaultfd.c:1046 [inline]
userfaultfd_read+0x394/0x18c0 fs/userfaultfd.c:1204
do_loop_readv_writev fs/read_write.c:701 [inline]
do_loop_readv_writev fs/read_write.c:688 [inline]
do_iter_read+0x490/0x640 fs/read_write.c:925
vfs_readv+0xf0/0x160 fs/read_write.c:987
do_readv+0x15e/0x370 fs/read_write.c:1020
__do_sys_readv fs/read_write.c:1107 [inline]
__se_sys_readv fs/read_write.c:1104 [inline]
__x64_sys_readv+0x75/0xb0 fs/read_write.c:1104
do_syscall_64+0xfd/0x620 arch/x86/entry/common.c:293
entry_SYSCALL_64_after_hwframe+0x49/0xbe

-> (&ctx->fault_pending_wqh){+.+.} ops: 3 {
HARDIRQ-ON-W at:
lock_acquire+0x16f/0x3f0 kernel/locking/lockdep.c:3900
__raw_spin_lock include/linux/spinlock_api_smp.h:142
[inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
userfaultfd_release+0x4d6/0x720 fs/userfaultfd.c:922
__fput+0x2dd/0x8b0 fs/file_table.c:278
____fput+0x16/0x20 fs/file_table.c:309
task_work_run+0x145/0x1c0 kernel/task_work.c:113
exit_task_work include/linux/task_work.h:22 [inline]
do_exit+0x933/0x2fa0 kernel/exit.c:876
do_group_exit+0x135/0x370 kernel/exit.c:979
get_signal+0x3ec/0x1fc0 kernel/signal.c:2574
do_signal+0x95/0x1960 arch/x86/kernel/signal.c:821
exit_to_usermode_loop+0x244/0x2c0
arch/x86/entry/common.c:163
prepare_exit_to_usermode arch/x86/entry/common.c:198
[inline]
syscall_return_slowpath arch/x86/entry/common.c:271
[inline]
do_syscall_64+0x53d/0x620 arch/x86/entry/common.c:296
entry_SYSCALL_64_after_hwframe+0x49/0xbe
SOFTIRQ-ON-W at:
lock_acquire+0x16f/0x3f0 kernel/locking/lockdep.c:3900
__raw_spin_lock include/linux/spinlock_api_smp.h:142
[inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
userfaultfd_release+0x4d6/0x720 fs/userfaultfd.c:922
__fput+0x2dd/0x8b0 fs/file_table.c:278
____fput+0x16/0x20 fs/file_table.c:309
task_work_run+0x145/0x1c0 kernel/task_work.c:113
exit_task_work include/linux/task_work.h:22 [inline]
do_exit+0x933/0x2fa0 kernel/exit.c:876
do_group_exit+0x135/0x370 kernel/exit.c:979
get_signal+0x3ec/0x1fc0 kernel/signal.c:2574
do_signal+0x95/0x1960 arch/x86/kernel/signal.c:821
exit_to_usermode_loop+0x244/0x2c0
arch/x86/entry/common.c:163
prepare_exit_to_usermode arch/x86/entry/common.c:198
[inline]
syscall_return_slowpath arch/x86/entry/common.c:271
[inline]
do_syscall_64+0x53d/0x620 arch/x86/entry/common.c:296
entry_SYSCALL_64_after_hwframe+0x49/0xbe
INITIAL USE at:
lock_acquire+0x16f/0x3f0 kernel/locking/lockdep.c:3900
__raw_spin_lock include/linux/spinlock_api_smp.h:142
[inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
userfaultfd_ctx_read fs/userfaultfd.c:1046 [inline]
userfaultfd_read+0x394/0x18c0 fs/userfaultfd.c:1204
do_loop_readv_writev fs/read_write.c:701 [inline]
do_loop_readv_writev fs/read_write.c:688 [inline]
do_iter_read+0x490/0x640 fs/read_write.c:925
vfs_readv+0xf0/0x160 fs/read_write.c:987
do_readv+0x15e/0x370 fs/read_write.c:1020
__do_sys_readv fs/read_write.c:1107 [inline]
__se_sys_readv fs/read_write.c:1104 [inline]
__x64_sys_readv+0x75/0xb0 fs/read_write.c:1104
do_syscall_64+0xfd/0x620 arch/x86/entry/common.c:293
entry_SYSCALL_64_after_hwframe+0x49/0xbe
}
... key at: [<ffffffff8a3811e0>] __key.43726+0x0/0x40
... acquired at:
mark_lock_irq kernel/locking/lockdep.c:2755 [inline]
mark_lock+0x420/0x1370 kernel/locking/lockdep.c:3127
mark_irqflags kernel/locking/lockdep.c:3023 [inline]
__lock_acquire+0x6b5/0x48f0 kernel/locking/lockdep.c:3368
lock_acquire+0x16f/0x3f0 kernel/locking/lockdep.c:3900
__raw_spin_lock include/linux/spinlock_api_smp.h:142 [inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
userfaultfd_release+0x4d6/0x720 fs/userfaultfd.c:922
__fput+0x2dd/0x8b0 fs/file_table.c:278
____fput+0x16/0x20 fs/file_table.c:309
task_work_run+0x145/0x1c0 kernel/task_work.c:113
exit_task_work include/linux/task_work.h:22 [inline]
do_exit+0x933/0x2fa0 kernel/exit.c:876
do_group_exit+0x135/0x370 kernel/exit.c:979
get_signal+0x3ec/0x1fc0 kernel/signal.c:2574
do_signal+0x95/0x1960 arch/x86/kernel/signal.c:821
exit_to_usermode_loop+0x244/0x2c0 arch/x86/entry/common.c:163
prepare_exit_to_usermode arch/x86/entry/common.c:198 [inline]
syscall_return_slowpath arch/x86/entry/common.c:271 [inline]
do_syscall_64+0x53d/0x620 arch/x86/entry/common.c:296
entry_SYSCALL_64_after_hwframe+0x49/0xbe


stack backtrace:
CPU: 0 PID: 8031 Comm: syz-executor068 Not tainted 4.19.51 #23
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0x172/0x1f0 lib/dump_stack.c:113
print_irq_inversion_bug.part.0+0x2c0/0x2cd kernel/locking/lockdep.c:2621
print_irq_inversion_bug kernel/locking/lockdep.c:2624 [inline]
check_usage_backwards.cold+0x1d/0x26 kernel/locking/lockdep.c:2670
mark_lock_irq kernel/locking/lockdep.c:2755 [inline]
mark_lock+0x420/0x1370 kernel/locking/lockdep.c:3127
mark_irqflags kernel/locking/lockdep.c:3023 [inline]
__lock_acquire+0x6b5/0x48f0 kernel/locking/lockdep.c:3368
lock_acquire+0x16f/0x3f0 kernel/locking/lockdep.c:3900
__raw_spin_lock include/linux/spinlock_api_smp.h:142 [inline]
_raw_spin_lock+0x2f/0x40 kernel/locking/spinlock.c:144
spin_lock include/linux/spinlock.h:329 [inline]
userfaultfd_release+0x4d6/0x720 fs/userfaultfd.c:922
__fput+0x2dd/0x8b0 fs/file_table.c:278
____fput+0x16/0x20 fs/file_table.c:309
task_work_run+0x145/0x1c0 kernel/task_work.c:113
exit_task_work include/linux/task_work.h:22 [inline]
do_exit+0x933/0x2fa0 kernel/exit.c:876
do_group_exit+0x135/0x370 kernel/exit.c:979
get_signal+0x3ec/0x1fc0 kernel/signal.c:2574
do_signal+0x95/0x1960 arch/x86/kernel/signal.c:821
exit_to_usermode_loop+0x244/0x2c0 arch/x86/entry/common.c:163
prepare_exit_to_usermode arch/x86/entry/common.c:198 [inline]
syscall_return_slowpath arch/x86/entry/common.c:271 [inline]
do_syscall_64+0x53d/0x620 arch/x86/entry/common.c:296
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x445919
C


---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches

[syzbot]

syzbot suspects this bug was fixed by commit:

commit 052b318100856aa86f4e0c03cfe43a1bb6bfb487
Author: Eric Biggers <ebig...@google.com>
Date: Thu Jul 4 22:14:39 2019 +0000

fs/userfaultfd.c: disable irqs for fault_pending and event locks

bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=16919696e00000
start commit: 7aa823a9 Linux 4.19.51
git tree: linux-4.19.y

kernel config: https://syzkaller.appspot.com/x/.config?x=57222bc62f28e3a6
dashboard link: https://syzkaller.appspot.com/bug?extid=0b8608e3d25d48fc4f4c

syz repro: https://syzkaller.appspot.com/x/repro.syz?x=10d9f956a00000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=10737eaea00000

If the result looks correct, please mark the bug fixed by replying with:

#syz fix: fs/userfaultfd.c: disable irqs for fault_pending and event locks

For information about bisection process see: https://goo.gl/tpsmEJ#bisection

[Eric Biggers]