* [syzbot] possible deadlock in hugetlb_fault
@ 2022-11-04 16:00 syzbot
2022-11-12 0:07 ` Mike Kravetz
0 siblings, 1 reply; 7+ messages in thread
From: syzbot @ 2022-11-04 16:00 UTC (permalink / raw)
To: akpm, linux-kernel, linux-mm, llvm, mike.kravetz, nathan,
ndesaulniers, songmuchun, syzkaller-bugs, trix
Hello,
syzbot found the following issue on:
HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
userspace arch: i386
Unfortunately, I don't have any reproducer for this issue yet.
Downloadable assets:
disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
======================================================
WARNING: possible circular locking dependency detected
6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
------------------------------------------------------
syz-executor.2/5665 is trying to acquire lock:
ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
but task is already holding lock:
ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&vma_lock->rw_sema){++++}-{3:3}:
down_write+0x90/0x220 kernel/locking/rwsem.c:1562
hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
__unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
unmap_vmas+0x21e/0x370 mm/memory.c:1733
exit_mmap+0x189/0x7a0 mm/mmap.c:3090
__mmput+0x128/0x4c0 kernel/fork.c:1185
mmput+0x5c/0x70 kernel/fork.c:1207
exit_mm kernel/exit.c:516 [inline]
do_exit+0xa39/0x2a20 kernel/exit.c:807
do_group_exit+0xd0/0x2a0 kernel/exit.c:950
get_signal+0x21a1/0x2430 kernel/signal.c:2858
arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
__syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
__do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
entry_SYSENTER_compat_after_hwframe+0x70/0x82
-> #0 (&mm->mmap_lock#2){++++}-{3:3}:
check_prev_add kernel/locking/lockdep.c:3097 [inline]
check_prevs_add kernel/locking/lockdep.c:3216 [inline]
validate_chain kernel/locking/lockdep.c:3831 [inline]
__lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
lock_acquire kernel/locking/lockdep.c:5668 [inline]
lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
__might_fault mm/memory.c:5646 [inline]
__might_fault+0x104/0x170 mm/memory.c:5639
_copy_from_user+0x25/0x170 lib/usercopy.c:13
copy_from_user include/linux/uaccess.h:161 [inline]
snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
vfs_write+0x2d7/0xdd0 fs/read_write.c:582
ksys_write+0x1e8/0x250 fs/read_write.c:637
do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
__do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
entry_SYSENTER_compat_after_hwframe+0x70/0x82
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&vma_lock->rw_sema);
lock(&mm->mmap_lock#2);
lock(&vma_lock->rw_sema);
lock(&mm->mmap_lock#2);
*** DEADLOCK ***
1 lock held by syz-executor.2/5665:
#0: ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
#0: ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
stack backtrace:
CPU: 1 PID: 5665 Comm: syz-executor.2 Not tainted 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/11/2022
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
check_noncircular+0x25f/0x2e0 kernel/locking/lockdep.c:2177
check_prev_add kernel/locking/lockdep.c:3097 [inline]
check_prevs_add kernel/locking/lockdep.c:3216 [inline]
validate_chain kernel/locking/lockdep.c:3831 [inline]
__lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
lock_acquire kernel/locking/lockdep.c:5668 [inline]
lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
__might_fault mm/memory.c:5646 [inline]
__might_fault+0x104/0x170 mm/memory.c:5639
_copy_from_user+0x25/0x170 lib/usercopy.c:13
copy_from_user include/linux/uaccess.h:161 [inline]
snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
vfs_write+0x2d7/0xdd0 fs/read_write.c:582
ksys_write+0x1e8/0x250 fs/read_write.c:637
do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
__do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
entry_SYSENTER_compat_after_hwframe+0x70/0x82
RIP: 0023:0xf7fad549
Code: 03 74 c0 01 10 05 03 74 b8 01 10 06 03 74 b4 01 10 07 03 74 b0 01 10 08 03 74 d8 01 00 00 00 00 00 51 52 55 89 e5 0f 34 cd 80 <5d> 5a 59 c3 90 90 90 90 8d b4 26 00 00 00 00 8d b4 26 00 00 00 00
RSP: 002b:00000000f7f875cc EFLAGS: 00000296 ORIG_RAX: 0000000000000004
RAX: ffffffffffffffda RBX: 0000000000000004 RCX: 0000000020000000
RDX: 00000000c86ade39 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000000
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
</TASK>
----------------
Code disassembly (best guess):
0: 03 74 c0 01 add 0x1(%rax,%rax,8),%esi
4: 10 05 03 74 b8 01 adc %al,0x1b87403(%rip) # 0x1b8740d
a: 10 06 adc %al,(%rsi)
c: 03 74 b4 01 add 0x1(%rsp,%rsi,4),%esi
10: 10 07 adc %al,(%rdi)
12: 03 74 b0 01 add 0x1(%rax,%rsi,4),%esi
16: 10 08 adc %cl,(%rax)
18: 03 74 d8 01 add 0x1(%rax,%rbx,8),%esi
1c: 00 00 add %al,(%rax)
1e: 00 00 add %al,(%rax)
20: 00 51 52 add %dl,0x52(%rcx)
23: 55 push %rbp
24: 89 e5 mov %esp,%ebp
26: 0f 34 sysenter
28: cd 80 int $0x80
* 2a: 5d pop %rbp <-- trapping instruction
2b: 5a pop %rdx
2c: 59 pop %rcx
2d: c3 retq
2e: 90 nop
2f: 90 nop
30: 90 nop
31: 90 nop
32: 8d b4 26 00 00 00 00 lea 0x0(%rsi,%riz,1),%esi
39: 8d b4 26 00 00 00 00 lea 0x0(%rsi,%riz,1),%esi
---
This report 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 syzkaller@googlegroups.com.
syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [syzbot] possible deadlock in hugetlb_fault
2022-11-04 16:00 [syzbot] possible deadlock in hugetlb_fault syzbot
@ 2022-11-12 0:07 ` Mike Kravetz
2022-11-12 4:01 ` Miaohe Lin
0 siblings, 1 reply; 7+ messages in thread
From: Mike Kravetz @ 2022-11-12 0:07 UTC (permalink / raw)
To: syzbot
Cc: Miaohe Lin, akpm, linux-kernel, linux-mm, llvm, nathan,
ndesaulniers, songmuchun, syzkaller-bugs, trix
On 11/04/22 09:00, syzbot wrote:
> Hello,
>
> syzbot found the following issue on:
>
> HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
> git tree: upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
> kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
> dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> userspace arch: i386
>
> Unfortunately, I don't have any reproducer for this issue yet.
>
> Downloadable assets:
> disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
> vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
> kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
>
> ======================================================
> WARNING: possible circular locking dependency detected
> 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
> ------------------------------------------------------
> syz-executor.2/5665 is trying to acquire lock:
> ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
>
> but task is already holding lock:
> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
>
> which lock already depends on the new lock.
>
>
> the existing dependency chain (in reverse order) is:
>
> -> #1 (&vma_lock->rw_sema){++++}-{3:3}:
> down_write+0x90/0x220 kernel/locking/rwsem.c:1562
> hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
> __unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
> unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
> unmap_vmas+0x21e/0x370 mm/memory.c:1733
> exit_mmap+0x189/0x7a0 mm/mmap.c:3090
> __mmput+0x128/0x4c0 kernel/fork.c:1185
> mmput+0x5c/0x70 kernel/fork.c:1207
> exit_mm kernel/exit.c:516 [inline]
> do_exit+0xa39/0x2a20 kernel/exit.c:807
> do_group_exit+0xd0/0x2a0 kernel/exit.c:950
> get_signal+0x21a1/0x2430 kernel/signal.c:2858
> arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
> exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
> exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
> __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
> syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
> __do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> entry_SYSENTER_compat_after_hwframe+0x70/0x82
>
> -> #0 (&mm->mmap_lock#2){++++}-{3:3}:
> check_prev_add kernel/locking/lockdep.c:3097 [inline]
> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
> validate_chain kernel/locking/lockdep.c:3831 [inline]
> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
> lock_acquire kernel/locking/lockdep.c:5668 [inline]
> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
> __might_fault mm/memory.c:5646 [inline]
> __might_fault+0x104/0x170 mm/memory.c:5639
> _copy_from_user+0x25/0x170 lib/usercopy.c:13
> copy_from_user include/linux/uaccess.h:161 [inline]
> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
> ksys_write+0x1e8/0x250 fs/read_write.c:637
> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> entry_SYSENTER_compat_after_hwframe+0x70/0x82
>
> other info that might help us debug this:
>
> Possible unsafe locking scenario:
>
> CPU0 CPU1
> ---- ----
> lock(&vma_lock->rw_sema);
> lock(&mm->mmap_lock#2);
> lock(&vma_lock->rw_sema);
> lock(&mm->mmap_lock#2);
I may not be reading the report correctly, but I can not see how we acquire the
hugetlb vma_lock before trying to acquire mmap_lock in stack 0. We would not
acquire the vma_lock until we enter hugetlb fault processing (not in the stack).
Adding Miaohe Lin on Cc due to previous help with vma_lock potential deadlock
situations. Miaohe, does this make sense to you?
--
Mike Kravetz
>
>
> 1 lock held by syz-executor.2/5665:
> #0: ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
> #0: ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
>
> stack backtrace:
> CPU: 1 PID: 5665 Comm: syz-executor.2 Not tainted 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/11/2022
> Call Trace:
> <TASK>
> __dump_stack lib/dump_stack.c:88 [inline]
> dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
> check_noncircular+0x25f/0x2e0 kernel/locking/lockdep.c:2177
> check_prev_add kernel/locking/lockdep.c:3097 [inline]
> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
> validate_chain kernel/locking/lockdep.c:3831 [inline]
> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
> lock_acquire kernel/locking/lockdep.c:5668 [inline]
> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
> __might_fault mm/memory.c:5646 [inline]
> __might_fault+0x104/0x170 mm/memory.c:5639
> _copy_from_user+0x25/0x170 lib/usercopy.c:13
> copy_from_user include/linux/uaccess.h:161 [inline]
> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
> ksys_write+0x1e8/0x250 fs/read_write.c:637
> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> RIP: 0023:0xf7fad549
> Code: 03 74 c0 01 10 05 03 74 b8 01 10 06 03 74 b4 01 10 07 03 74 b0 01 10 08 03 74 d8 01 00 00 00 00 00 51 52 55 89 e5 0f 34 cd 80 <5d> 5a 59 c3 90 90 90 90 8d b4 26 00 00 00 00 8d b4 26 00 00 00 00
> RSP: 002b:00000000f7f875cc EFLAGS: 00000296 ORIG_RAX: 0000000000000004
> RAX: ffffffffffffffda RBX: 0000000000000004 RCX: 0000000020000000
> RDX: 00000000c86ade39 RSI: 0000000000000000 RDI: 0000000000000000
> RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000000
> R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
> </TASK>
> ----------------
> Code disassembly (best guess):
> 0: 03 74 c0 01 add 0x1(%rax,%rax,8),%esi
> 4: 10 05 03 74 b8 01 adc %al,0x1b87403(%rip) # 0x1b8740d
> a: 10 06 adc %al,(%rsi)
> c: 03 74 b4 01 add 0x1(%rsp,%rsi,4),%esi
> 10: 10 07 adc %al,(%rdi)
> 12: 03 74 b0 01 add 0x1(%rax,%rsi,4),%esi
> 16: 10 08 adc %cl,(%rax)
> 18: 03 74 d8 01 add 0x1(%rax,%rbx,8),%esi
> 1c: 00 00 add %al,(%rax)
> 1e: 00 00 add %al,(%rax)
> 20: 00 51 52 add %dl,0x52(%rcx)
> 23: 55 push %rbp
> 24: 89 e5 mov %esp,%ebp
> 26: 0f 34 sysenter
> 28: cd 80 int $0x80
> * 2a: 5d pop %rbp <-- trapping instruction
> 2b: 5a pop %rdx
> 2c: 59 pop %rcx
> 2d: c3 retq
> 2e: 90 nop
> 2f: 90 nop
> 30: 90 nop
> 31: 90 nop
> 32: 8d b4 26 00 00 00 00 lea 0x0(%rsi,%riz,1),%esi
> 39: 8d b4 26 00 00 00 00 lea 0x0(%rsi,%riz,1),%esi
>
>
> ---
> This report 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 syzkaller@googlegroups.com.
>
> syzbot will keep track of this issue. See:
> https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [syzbot] possible deadlock in hugetlb_fault
2022-11-12 0:07 ` Mike Kravetz
@ 2022-11-12 4:01 ` Miaohe Lin
2022-11-12 11:33 ` Dmitry Vyukov
0 siblings, 1 reply; 7+ messages in thread
From: Miaohe Lin @ 2022-11-12 4:01 UTC (permalink / raw)
To: Mike Kravetz, syzbot
Cc: akpm, linux-kernel, linux-mm, llvm, nathan, ndesaulniers,
songmuchun, syzkaller-bugs, trix
On 2022/11/12 8:07, Mike Kravetz wrote:
> On 11/04/22 09:00, syzbot wrote:
>> Hello,
>>
>> syzbot found the following issue on:
>>
>> HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
>> git tree: upstream
>> console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
>> kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
>> dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
>> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
>> userspace arch: i386
>>
>> Unfortunately, I don't have any reproducer for this issue yet.
>>
>> Downloadable assets:
>> disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
>> vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
>> kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
>>
>> IMPORTANT: if you fix the issue, please add the following tag to the commit:
>> Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
>>
>> ======================================================
>> WARNING: possible circular locking dependency detected
>> 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
>> ------------------------------------------------------
>> syz-executor.2/5665 is trying to acquire lock:
>> ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
>>
>> but task is already holding lock:
>> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
>> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
>>
>> which lock already depends on the new lock.
>>
>>
>> the existing dependency chain (in reverse order) is:
>>
>> -> #1 (&vma_lock->rw_sema){++++}-{3:3}:
>> down_write+0x90/0x220 kernel/locking/rwsem.c:1562
>> hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
>> __unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
>> unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
>> unmap_vmas+0x21e/0x370 mm/memory.c:1733
>> exit_mmap+0x189/0x7a0 mm/mmap.c:3090
>> __mmput+0x128/0x4c0 kernel/fork.c:1185
>> mmput+0x5c/0x70 kernel/fork.c:1207
>> exit_mm kernel/exit.c:516 [inline]
>> do_exit+0xa39/0x2a20 kernel/exit.c:807
>> do_group_exit+0xd0/0x2a0 kernel/exit.c:950
>> get_signal+0x21a1/0x2430 kernel/signal.c:2858
>> arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
>> exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
>> exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
>> __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
>> syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
>> __do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
>> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
>> entry_SYSENTER_compat_after_hwframe+0x70/0x82
>>
>> -> #0 (&mm->mmap_lock#2){++++}-{3:3}:
>> check_prev_add kernel/locking/lockdep.c:3097 [inline]
>> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
>> validate_chain kernel/locking/lockdep.c:3831 [inline]
>> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
>> lock_acquire kernel/locking/lockdep.c:5668 [inline]
>> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
>> __might_fault mm/memory.c:5646 [inline]
>> __might_fault+0x104/0x170 mm/memory.c:5639
>> _copy_from_user+0x25/0x170 lib/usercopy.c:13
>> copy_from_user include/linux/uaccess.h:161 [inline]
>> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
>> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
>> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
>> ksys_write+0x1e8/0x250 fs/read_write.c:637
>> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
>> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
>> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
>> entry_SYSENTER_compat_after_hwframe+0x70/0x82
>>
>> other info that might help us debug this:
>>
>> Possible unsafe locking scenario:
>>
>> CPU0 CPU1
>> ---- ----
>> lock(&vma_lock->rw_sema);
>> lock(&mm->mmap_lock#2);
>> lock(&vma_lock->rw_sema);
>> lock(&mm->mmap_lock#2);
>
> I may not be reading the report correctly, but I can not see how we acquire the
> hugetlb vma_lock before trying to acquire mmap_lock in stack 0. We would not
> acquire the vma_lock until we enter hugetlb fault processing (not in the stack).
>
> Adding Miaohe Lin on Cc due to previous help with vma_lock potential deadlock
> situations. Miaohe, does this make sense to you?
>
Hi Mike,
This doesn't make sense for me too. Stack #1 shows that syz-executor is releasing
its address space while stack #0 shows another thread is serving the write syscall.
In this case, mm->mm_users is 0 and all threads in this process should be serving
do_exit()? But I could be easily wrong. Also I can't see how vma_lock is locked before
trying to acquire mmap_lock in above stacks. Might this be a false positive?
Thanks,
Miaohe Lin
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [syzbot] possible deadlock in hugetlb_fault
2022-11-12 4:01 ` Miaohe Lin
@ 2022-11-12 11:33 ` Dmitry Vyukov
2022-11-12 14:14 ` Dmitry Vyukov
0 siblings, 1 reply; 7+ messages in thread
From: Dmitry Vyukov @ 2022-11-12 11:33 UTC (permalink / raw)
To: Miaohe Lin
Cc: Mike Kravetz, syzbot, akpm, linux-kernel, linux-mm, llvm, nathan,
ndesaulniers, songmuchun, syzkaller-bugs, trix
On Sat, 12 Nov 2022 at 05:02, 'Miaohe Lin' via syzkaller-bugs
<syzkaller-bugs@googlegroups.com> wrote:
>
> On 2022/11/12 8:07, Mike Kravetz wrote:
> > On 11/04/22 09:00, syzbot wrote:
> >> Hello,
> >>
> >> syzbot found the following issue on:
> >>
> >> HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
> >> git tree: upstream
> >> console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
> >> kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
> >> dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
> >> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> >> userspace arch: i386
> >>
> >> Unfortunately, I don't have any reproducer for this issue yet.
> >>
> >> Downloadable assets:
> >> disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
> >> vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
> >> kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
> >>
> >> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> >> Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
> >>
> >> ======================================================
> >> WARNING: possible circular locking dependency detected
> >> 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
> >> ------------------------------------------------------
> >> syz-executor.2/5665 is trying to acquire lock:
> >> ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
> >>
> >> but task is already holding lock:
> >> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
> >> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
> >>
> >> which lock already depends on the new lock.
> >>
> >>
> >> the existing dependency chain (in reverse order) is:
> >>
> >> -> #1 (&vma_lock->rw_sema){++++}-{3:3}:
> >> down_write+0x90/0x220 kernel/locking/rwsem.c:1562
> >> hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
> >> __unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
> >> unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
> >> unmap_vmas+0x21e/0x370 mm/memory.c:1733
> >> exit_mmap+0x189/0x7a0 mm/mmap.c:3090
> >> __mmput+0x128/0x4c0 kernel/fork.c:1185
> >> mmput+0x5c/0x70 kernel/fork.c:1207
> >> exit_mm kernel/exit.c:516 [inline]
> >> do_exit+0xa39/0x2a20 kernel/exit.c:807
> >> do_group_exit+0xd0/0x2a0 kernel/exit.c:950
> >> get_signal+0x21a1/0x2430 kernel/signal.c:2858
> >> arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
> >> exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
> >> exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
> >> __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
> >> syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
> >> __do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
> >> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> >> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> >>
> >> -> #0 (&mm->mmap_lock#2){++++}-{3:3}:
> >> check_prev_add kernel/locking/lockdep.c:3097 [inline]
> >> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
> >> validate_chain kernel/locking/lockdep.c:3831 [inline]
> >> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
> >> lock_acquire kernel/locking/lockdep.c:5668 [inline]
> >> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
> >> __might_fault mm/memory.c:5646 [inline]
> >> __might_fault+0x104/0x170 mm/memory.c:5639
> >> _copy_from_user+0x25/0x170 lib/usercopy.c:13
> >> copy_from_user include/linux/uaccess.h:161 [inline]
> >> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
> >> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
> >> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
> >> ksys_write+0x1e8/0x250 fs/read_write.c:637
> >> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
> >> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
> >> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> >> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> >>
> >> other info that might help us debug this:
> >>
> >> Possible unsafe locking scenario:
> >>
> >> CPU0 CPU1
> >> ---- ----
> >> lock(&vma_lock->rw_sema);
> >> lock(&mm->mmap_lock#2);
> >> lock(&vma_lock->rw_sema);
> >> lock(&mm->mmap_lock#2);
> >
> > I may not be reading the report correctly, but I can not see how we acquire the
> > hugetlb vma_lock before trying to acquire mmap_lock in stack 0. We would not
> > acquire the vma_lock until we enter hugetlb fault processing (not in the stack).
The unlock of vma_lock is conditional:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/hugetlb.c?id=f2f32f8af2b0ca9d619e5183eae3eed431793baf#n6840
and the condition is:
static bool __vma_shareable_flags_pmd(struct vm_area_struct *vma)
{
return vma->vm_flags & (VM_MAYSHARE | VM_SHARED) &&
vma->vm_private_data;
}
Is it possible that the condition has changed between vma lock and unlock?
What mutexes protect vma->vm_flags/vm_private_data?
That would make the report perfectly sensible.
FWIW the test case that was running is this, that's the syscalls that
were running concurrently:
07:56:56 executing program 2:
r0 = syz_open_dev$sndmidi(&(0x7f0000000040), 0x2, 0x141101)
r1 = dup(r0)
setsockopt$inet_sctp_SCTP_I_WANT_MAPPED_V4_ADDR(r1, 0x84, 0xc,
&(0x7f0000000080), 0x4) (async)
write$6lowpan_enable(r1, &(0x7f0000000000)='0', 0xc86ade39) (async)
mmap(&(0x7f0000000000/0xb36000)=nil, 0xb36000, 0x3, 0x68831,
0xffffffffffffffff, 0x0) (async)
madvise(&(0x7f0000000000/0x600000)=nil, 0x600003, 0x4) (async, rerun: 32)
mremap(&(0x7f00007a0000/0x3000)=nil, 0x3000, 0x2000, 0x7,
&(0x7f0000835000/0x2000)=nil) (rerun: 32)
> > Adding Miaohe Lin on Cc due to previous help with vma_lock potential deadlock
> > situations. Miaohe, does this make sense to you?
> >
>
> Hi Mike,
> This doesn't make sense for me too. Stack #1 shows that syz-executor is releasing
> its address space while stack #0 shows another thread is serving the write syscall.
> In this case, mm->mm_users is 0 and all threads in this process should be serving
> do_exit()? But I could be easily wrong. Also I can't see how vma_lock is locked before
> trying to acquire mmap_lock in above stacks. Might this be a false positive?
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [syzbot] possible deadlock in hugetlb_fault
2022-11-12 11:33 ` Dmitry Vyukov
@ 2022-11-12 14:14 ` Dmitry Vyukov
2022-11-14 2:35 ` Miaohe Lin
0 siblings, 1 reply; 7+ messages in thread
From: Dmitry Vyukov @ 2022-11-12 14:14 UTC (permalink / raw)
To: Miaohe Lin
Cc: Mike Kravetz, syzbot, akpm, linux-kernel, linux-mm, llvm, nathan,
ndesaulniers, songmuchun, syzkaller-bugs, trix
On Sat, 12 Nov 2022 at 12:33, Dmitry Vyukov <dvyukov@google.com> wrote:
>
> On Sat, 12 Nov 2022 at 05:02, 'Miaohe Lin' via syzkaller-bugs
> <syzkaller-bugs@googlegroups.com> wrote:
> >
> > On 2022/11/12 8:07, Mike Kravetz wrote:
> > > On 11/04/22 09:00, syzbot wrote:
> > >> Hello,
> > >>
> > >> syzbot found the following issue on:
> > >>
> > >> HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
> > >> git tree: upstream
> > >> console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
> > >> kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
> > >> dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
> > >> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> > >> userspace arch: i386
> > >>
> > >> Unfortunately, I don't have any reproducer for this issue yet.
> > >>
> > >> Downloadable assets:
> > >> disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
> > >> vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
> > >> kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
> > >>
> > >> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > >> Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
> > >>
> > >> ======================================================
> > >> WARNING: possible circular locking dependency detected
> > >> 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
> > >> ------------------------------------------------------
> > >> syz-executor.2/5665 is trying to acquire lock:
> > >> ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
> > >>
> > >> but task is already holding lock:
> > >> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
> > >> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
> > >>
> > >> which lock already depends on the new lock.
> > >>
> > >>
> > >> the existing dependency chain (in reverse order) is:
> > >>
> > >> -> #1 (&vma_lock->rw_sema){++++}-{3:3}:
> > >> down_write+0x90/0x220 kernel/locking/rwsem.c:1562
> > >> hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
> > >> __unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
> > >> unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
> > >> unmap_vmas+0x21e/0x370 mm/memory.c:1733
> > >> exit_mmap+0x189/0x7a0 mm/mmap.c:3090
> > >> __mmput+0x128/0x4c0 kernel/fork.c:1185
> > >> mmput+0x5c/0x70 kernel/fork.c:1207
> > >> exit_mm kernel/exit.c:516 [inline]
> > >> do_exit+0xa39/0x2a20 kernel/exit.c:807
> > >> do_group_exit+0xd0/0x2a0 kernel/exit.c:950
> > >> get_signal+0x21a1/0x2430 kernel/signal.c:2858
> > >> arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
> > >> exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
> > >> exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
> > >> __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
> > >> syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
> > >> __do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
> > >> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> > >> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> > >>
> > >> -> #0 (&mm->mmap_lock#2){++++}-{3:3}:
> > >> check_prev_add kernel/locking/lockdep.c:3097 [inline]
> > >> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
> > >> validate_chain kernel/locking/lockdep.c:3831 [inline]
> > >> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
> > >> lock_acquire kernel/locking/lockdep.c:5668 [inline]
> > >> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
> > >> __might_fault mm/memory.c:5646 [inline]
> > >> __might_fault+0x104/0x170 mm/memory.c:5639
> > >> _copy_from_user+0x25/0x170 lib/usercopy.c:13
> > >> copy_from_user include/linux/uaccess.h:161 [inline]
> > >> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
> > >> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
> > >> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
> > >> ksys_write+0x1e8/0x250 fs/read_write.c:637
> > >> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
> > >> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
> > >> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> > >> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> > >>
> > >> other info that might help us debug this:
> > >>
> > >> Possible unsafe locking scenario:
> > >>
> > >> CPU0 CPU1
> > >> ---- ----
> > >> lock(&vma_lock->rw_sema);
> > >> lock(&mm->mmap_lock#2);
> > >> lock(&vma_lock->rw_sema);
> > >> lock(&mm->mmap_lock#2);
> > >
> > > I may not be reading the report correctly, but I can not see how we acquire the
> > > hugetlb vma_lock before trying to acquire mmap_lock in stack 0. We would not
> > > acquire the vma_lock until we enter hugetlb fault processing (not in the stack).
>
> The unlock of vma_lock is conditional:
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/hugetlb.c?id=f2f32f8af2b0ca9d619e5183eae3eed431793baf#n6840
>
> and the condition is:
>
> static bool __vma_shareable_flags_pmd(struct vm_area_struct *vma)
> {
> return vma->vm_flags & (VM_MAYSHARE | VM_SHARED) &&
> vma->vm_private_data;
> }
>
> Is it possible that the condition has changed between vma lock and unlock?
> What mutexes protect vma->vm_flags/vm_private_data?
>
> That would make the report perfectly sensible.
>
> FWIW the test case that was running is this, that's the syscalls that
> were running concurrently:
>
> 07:56:56 executing program 2:
> r0 = syz_open_dev$sndmidi(&(0x7f0000000040), 0x2, 0x141101)
> r1 = dup(r0)
> setsockopt$inet_sctp_SCTP_I_WANT_MAPPED_V4_ADDR(r1, 0x84, 0xc,
> &(0x7f0000000080), 0x4) (async)
> write$6lowpan_enable(r1, &(0x7f0000000000)='0', 0xc86ade39) (async)
> mmap(&(0x7f0000000000/0xb36000)=nil, 0xb36000, 0x3, 0x68831,
> 0xffffffffffffffff, 0x0) (async)
> madvise(&(0x7f0000000000/0x600000)=nil, 0x600003, 0x4) (async, rerun: 32)
> mremap(&(0x7f00007a0000/0x3000)=nil, 0x3000, 0x2000, 0x7,
> &(0x7f0000835000/0x2000)=nil) (rerun: 32)
This new bug report seems to confirm the hypothesis:
WARNING: locking bug in hugetlb_no_page
https://syzkaller.appspot.com/bug?extid=d07c65298d2c15eafcb0
> > > Adding Miaohe Lin on Cc due to previous help with vma_lock potential deadlock
> > > situations. Miaohe, does this make sense to you?
> > >
> >
> > Hi Mike,
> > This doesn't make sense for me too. Stack #1 shows that syz-executor is releasing
> > its address space while stack #0 shows another thread is serving the write syscall.
> > In this case, mm->mm_users is 0 and all threads in this process should be serving
> > do_exit()? But I could be easily wrong. Also I can't see how vma_lock is locked before
> > trying to acquire mmap_lock in above stacks. Might this be a false positive?
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [syzbot] possible deadlock in hugetlb_fault
2022-11-12 14:14 ` Dmitry Vyukov
@ 2022-11-14 2:35 ` Miaohe Lin
2022-11-14 9:57 ` Dmitry Vyukov
0 siblings, 1 reply; 7+ messages in thread
From: Miaohe Lin @ 2022-11-14 2:35 UTC (permalink / raw)
To: Dmitry Vyukov
Cc: Mike Kravetz, syzbot, akpm, linux-kernel, linux-mm, llvm, nathan,
ndesaulniers, songmuchun, syzkaller-bugs, trix
On 2022/11/12 22:14, Dmitry Vyukov wrote:
> On Sat, 12 Nov 2022 at 12:33, Dmitry Vyukov <dvyukov@google.com> wrote:
>>
>> On Sat, 12 Nov 2022 at 05:02, 'Miaohe Lin' via syzkaller-bugs
>> <syzkaller-bugs@googlegroups.com> wrote:
>>>
>>> On 2022/11/12 8:07, Mike Kravetz wrote:
>>>> On 11/04/22 09:00, syzbot wrote:
>>>>> Hello,
>>>>>
>>>>> syzbot found the following issue on:
>>>>>
>>>>> HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
>>>>> git tree: upstream
>>>>> console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
>>>>> kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
>>>>> dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
>>>>> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
>>>>> userspace arch: i386
>>>>>
>>>>> Unfortunately, I don't have any reproducer for this issue yet.
>>>>>
>>>>> Downloadable assets:
>>>>> disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
>>>>> vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
>>>>> kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
>>>>>
>>>>> IMPORTANT: if you fix the issue, please add the following tag to the commit:
>>>>> Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
>>>>>
>>>>> ======================================================
>>>>> WARNING: possible circular locking dependency detected
>>>>> 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
>>>>> ------------------------------------------------------
>>>>> syz-executor.2/5665 is trying to acquire lock:
>>>>> ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
>>>>>
>>>>> but task is already holding lock:
>>>>> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
>>>>> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
>>>>>
>>>>> which lock already depends on the new lock.
>>>>>
>>>>>
>>>>> the existing dependency chain (in reverse order) is:
>>>>>
>>>>> -> #1 (&vma_lock->rw_sema){++++}-{3:3}:
>>>>> down_write+0x90/0x220 kernel/locking/rwsem.c:1562
>>>>> hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
>>>>> __unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
>>>>> unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
>>>>> unmap_vmas+0x21e/0x370 mm/memory.c:1733
>>>>> exit_mmap+0x189/0x7a0 mm/mmap.c:3090
>>>>> __mmput+0x128/0x4c0 kernel/fork.c:1185
>>>>> mmput+0x5c/0x70 kernel/fork.c:1207
>>>>> exit_mm kernel/exit.c:516 [inline]
>>>>> do_exit+0xa39/0x2a20 kernel/exit.c:807
>>>>> do_group_exit+0xd0/0x2a0 kernel/exit.c:950
>>>>> get_signal+0x21a1/0x2430 kernel/signal.c:2858
>>>>> arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
>>>>> exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
>>>>> exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
>>>>> __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
>>>>> syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
>>>>> __do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
>>>>> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
>>>>> entry_SYSENTER_compat_after_hwframe+0x70/0x82
>>>>>
>>>>> -> #0 (&mm->mmap_lock#2){++++}-{3:3}:
>>>>> check_prev_add kernel/locking/lockdep.c:3097 [inline]
>>>>> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
>>>>> validate_chain kernel/locking/lockdep.c:3831 [inline]
>>>>> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
>>>>> lock_acquire kernel/locking/lockdep.c:5668 [inline]
>>>>> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
>>>>> __might_fault mm/memory.c:5646 [inline]
>>>>> __might_fault+0x104/0x170 mm/memory.c:5639
>>>>> _copy_from_user+0x25/0x170 lib/usercopy.c:13
>>>>> copy_from_user include/linux/uaccess.h:161 [inline]
>>>>> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
>>>>> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
>>>>> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
>>>>> ksys_write+0x1e8/0x250 fs/read_write.c:637
>>>>> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
>>>>> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
>>>>> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
>>>>> entry_SYSENTER_compat_after_hwframe+0x70/0x82
>>>>>
>>>>> other info that might help us debug this:
>>>>>
>>>>> Possible unsafe locking scenario:
>>>>>
>>>>> CPU0 CPU1
>>>>> ---- ----
>>>>> lock(&vma_lock->rw_sema);
>>>>> lock(&mm->mmap_lock#2);
>>>>> lock(&vma_lock->rw_sema);
>>>>> lock(&mm->mmap_lock#2);
>>>>
>>>> I may not be reading the report correctly, but I can not see how we acquire the
>>>> hugetlb vma_lock before trying to acquire mmap_lock in stack 0. We would not
>>>> acquire the vma_lock until we enter hugetlb fault processing (not in the stack).
>>
>> The unlock of vma_lock is conditional:
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/hugetlb.c?id=f2f32f8af2b0ca9d619e5183eae3eed431793baf#n6840
>>
>> and the condition is:
>>
>> static bool __vma_shareable_flags_pmd(struct vm_area_struct *vma)
>> {
>> return vma->vm_flags & (VM_MAYSHARE | VM_SHARED) &&
>> vma->vm_private_data;
>> }
>>
>> Is it possible that the condition has changed between vma lock and unlock?
>> What mutexes protect vma->vm_flags/vm_private_data?
>>
>> That would make the report perfectly sensible.
>>
>> FWIW the test case that was running is this, that's the syscalls that
>> were running concurrently:
>>
>> 07:56:56 executing program 2:
>> r0 = syz_open_dev$sndmidi(&(0x7f0000000040), 0x2, 0x141101)
>> r1 = dup(r0)
>> setsockopt$inet_sctp_SCTP_I_WANT_MAPPED_V4_ADDR(r1, 0x84, 0xc,
>> &(0x7f0000000080), 0x4) (async)
>> write$6lowpan_enable(r1, &(0x7f0000000000)='0', 0xc86ade39) (async)
>> mmap(&(0x7f0000000000/0xb36000)=nil, 0xb36000, 0x3, 0x68831,
>> 0xffffffffffffffff, 0x0) (async)
>> madvise(&(0x7f0000000000/0x600000)=nil, 0x600003, 0x4) (async, rerun: 32)
>> mremap(&(0x7f00007a0000/0x3000)=nil, 0x3000, 0x2000, 0x7,
>> &(0x7f0000835000/0x2000)=nil) (rerun: 32)
>
>
> This new bug report seems to confirm the hypothesis:
>
> WARNING: locking bug in hugetlb_no_page
> https://syzkaller.appspot.com/bug?extid=d07c65298d2c15eafcb0
Thanks Dmitry!
I believe Mike found the root cause and his proposed patches in next-20221111 will fix the problem:
https://lore.kernel.org/linux-mm/20221111232628.290160-1-mike.kravetz@oracle.com/
Let's keep looking. :) Thanks for your work!
Thanks,
Miaohe Lin
>
>
>>>> Adding Miaohe Lin on Cc due to previous help with vma_lock potential deadlock
>>>> situations. Miaohe, does this make sense to you?
>>>>
>>>
>>> Hi Mike,
>>> This doesn't make sense for me too. Stack #1 shows that syz-executor is releasing
>>> its address space while stack #0 shows another thread is serving the write syscall.
>>> In this case, mm->mm_users is 0 and all threads in this process should be serving
>>> do_exit()? But I could be easily wrong. Also I can't see how vma_lock is locked before
>>> trying to acquire mmap_lock in above stacks. Might this be a false positive?
> .
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [syzbot] possible deadlock in hugetlb_fault
2022-11-14 2:35 ` Miaohe Lin
@ 2022-11-14 9:57 ` Dmitry Vyukov
0 siblings, 0 replies; 7+ messages in thread
From: Dmitry Vyukov @ 2022-11-14 9:57 UTC (permalink / raw)
To: Miaohe Lin
Cc: Mike Kravetz, syzbot, akpm, linux-kernel, linux-mm, llvm, nathan,
ndesaulniers, songmuchun, syzkaller-bugs, trix
On Mon, 14 Nov 2022 at 03:35, 'Miaohe Lin' via syzkaller-bugs
<syzkaller-bugs@googlegroups.com> wrote:
>
> On 2022/11/12 22:14, Dmitry Vyukov wrote:
> > On Sat, 12 Nov 2022 at 12:33, Dmitry Vyukov <dvyukov@google.com> wrote:
> >>
> >> On Sat, 12 Nov 2022 at 05:02, 'Miaohe Lin' via syzkaller-bugs
> >> <syzkaller-bugs@googlegroups.com> wrote:
> >>>
> >>> On 2022/11/12 8:07, Mike Kravetz wrote:
> >>>> On 11/04/22 09:00, syzbot wrote:
> >>>>> Hello,
> >>>>>
> >>>>> syzbot found the following issue on:
> >>>>>
> >>>>> HEAD commit: f2f32f8af2b0 Merge tag 'for-6.1-rc3-tag' of git://git.kern..
> >>>>> git tree: upstream
> >>>>> console output: https://syzkaller.appspot.com/x/log.txt?x=137d52ca880000
> >>>>> kernel config: https://syzkaller.appspot.com/x/.config?x=d080a4bd239918dd
> >>>>> dashboard link: https://syzkaller.appspot.com/bug?extid=ca56f14c500045350f93
> >>>>> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> >>>>> userspace arch: i386
> >>>>>
> >>>>> Unfortunately, I don't have any reproducer for this issue yet.
> >>>>>
> >>>>> Downloadable assets:
> >>>>> disk image: https://storage.googleapis.com/syzbot-assets/b4f72e7a4c11/disk-f2f32f8a.raw.xz
> >>>>> vmlinux: https://storage.googleapis.com/syzbot-assets/3f88997ad7c9/vmlinux-f2f32f8a.xz
> >>>>> kernel image: https://storage.googleapis.com/syzbot-assets/b4b5b3963e2d/bzImage-f2f32f8a.xz
> >>>>>
> >>>>> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> >>>>> Reported-by: syzbot+ca56f14c500045350f93@syzkaller.appspotmail.com
> >>>>>
> >>>>> ======================================================
> >>>>> WARNING: possible circular locking dependency detected
> >>>>> 6.1.0-rc3-syzkaller-00152-gf2f32f8af2b0 #0 Not tainted
> >>>>> ------------------------------------------------------
> >>>>> syz-executor.2/5665 is trying to acquire lock:
> >>>>> ffff88801c74c298 (&mm->mmap_lock#2){++++}-{3:3}, at: __might_fault+0xa1/0x170 mm/memory.c:5645
> >>>>>
> >>>>> but task is already holding lock:
> >>>>> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_vma_lock_read mm/hugetlb.c:6816 [inline]
> >>>>> ffff88801c4f3078 (&vma_lock->rw_sema){++++}-{3:3}, at: hugetlb_fault+0x40a/0x2060 mm/hugetlb.c:5859
> >>>>>
> >>>>> which lock already depends on the new lock.
> >>>>>
> >>>>>
> >>>>> the existing dependency chain (in reverse order) is:
> >>>>>
> >>>>> -> #1 (&vma_lock->rw_sema){++++}-{3:3}:
> >>>>> down_write+0x90/0x220 kernel/locking/rwsem.c:1562
> >>>>> hugetlb_vma_lock_write mm/hugetlb.c:6834 [inline]
> >>>>> __unmap_hugepage_range_final+0x97/0x340 mm/hugetlb.c:5202
> >>>>> unmap_single_vma+0x23d/0x2a0 mm/memory.c:1690
> >>>>> unmap_vmas+0x21e/0x370 mm/memory.c:1733
> >>>>> exit_mmap+0x189/0x7a0 mm/mmap.c:3090
> >>>>> __mmput+0x128/0x4c0 kernel/fork.c:1185
> >>>>> mmput+0x5c/0x70 kernel/fork.c:1207
> >>>>> exit_mm kernel/exit.c:516 [inline]
> >>>>> do_exit+0xa39/0x2a20 kernel/exit.c:807
> >>>>> do_group_exit+0xd0/0x2a0 kernel/exit.c:950
> >>>>> get_signal+0x21a1/0x2430 kernel/signal.c:2858
> >>>>> arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869
> >>>>> exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
> >>>>> exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
> >>>>> __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
> >>>>> syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296
> >>>>> __do_fast_syscall_32+0x72/0xf0 arch/x86/entry/common.c:181
> >>>>> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> >>>>> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> >>>>>
> >>>>> -> #0 (&mm->mmap_lock#2){++++}-{3:3}:
> >>>>> check_prev_add kernel/locking/lockdep.c:3097 [inline]
> >>>>> check_prevs_add kernel/locking/lockdep.c:3216 [inline]
> >>>>> validate_chain kernel/locking/lockdep.c:3831 [inline]
> >>>>> __lock_acquire+0x2a43/0x56d0 kernel/locking/lockdep.c:5055
> >>>>> lock_acquire kernel/locking/lockdep.c:5668 [inline]
> >>>>> lock_acquire+0x1df/0x630 kernel/locking/lockdep.c:5633
> >>>>> __might_fault mm/memory.c:5646 [inline]
> >>>>> __might_fault+0x104/0x170 mm/memory.c:5639
> >>>>> _copy_from_user+0x25/0x170 lib/usercopy.c:13
> >>>>> copy_from_user include/linux/uaccess.h:161 [inline]
> >>>>> snd_rawmidi_kernel_write1+0x366/0x880 sound/core/rawmidi.c:1549
> >>>>> snd_rawmidi_write+0x273/0xbb0 sound/core/rawmidi.c:1618
> >>>>> vfs_write+0x2d7/0xdd0 fs/read_write.c:582
> >>>>> ksys_write+0x1e8/0x250 fs/read_write.c:637
> >>>>> do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
> >>>>> __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
> >>>>> do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
> >>>>> entry_SYSENTER_compat_after_hwframe+0x70/0x82
> >>>>>
> >>>>> other info that might help us debug this:
> >>>>>
> >>>>> Possible unsafe locking scenario:
> >>>>>
> >>>>> CPU0 CPU1
> >>>>> ---- ----
> >>>>> lock(&vma_lock->rw_sema);
> >>>>> lock(&mm->mmap_lock#2);
> >>>>> lock(&vma_lock->rw_sema);
> >>>>> lock(&mm->mmap_lock#2);
> >>>>
> >>>> I may not be reading the report correctly, but I can not see how we acquire the
> >>>> hugetlb vma_lock before trying to acquire mmap_lock in stack 0. We would not
> >>>> acquire the vma_lock until we enter hugetlb fault processing (not in the stack).
> >>
> >> The unlock of vma_lock is conditional:
> >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/hugetlb.c?id=f2f32f8af2b0ca9d619e5183eae3eed431793baf#n6840
> >>
> >> and the condition is:
> >>
> >> static bool __vma_shareable_flags_pmd(struct vm_area_struct *vma)
> >> {
> >> return vma->vm_flags & (VM_MAYSHARE | VM_SHARED) &&
> >> vma->vm_private_data;
> >> }
> >>
> >> Is it possible that the condition has changed between vma lock and unlock?
> >> What mutexes protect vma->vm_flags/vm_private_data?
> >>
> >> That would make the report perfectly sensible.
> >>
> >> FWIW the test case that was running is this, that's the syscalls that
> >> were running concurrently:
> >>
> >> 07:56:56 executing program 2:
> >> r0 = syz_open_dev$sndmidi(&(0x7f0000000040), 0x2, 0x141101)
> >> r1 = dup(r0)
> >> setsockopt$inet_sctp_SCTP_I_WANT_MAPPED_V4_ADDR(r1, 0x84, 0xc,
> >> &(0x7f0000000080), 0x4) (async)
> >> write$6lowpan_enable(r1, &(0x7f0000000000)='0', 0xc86ade39) (async)
> >> mmap(&(0x7f0000000000/0xb36000)=nil, 0xb36000, 0x3, 0x68831,
> >> 0xffffffffffffffff, 0x0) (async)
> >> madvise(&(0x7f0000000000/0x600000)=nil, 0x600003, 0x4) (async, rerun: 32)
> >> mremap(&(0x7f00007a0000/0x3000)=nil, 0x3000, 0x2000, 0x7,
> >> &(0x7f0000835000/0x2000)=nil) (rerun: 32)
> >
> >
> > This new bug report seems to confirm the hypothesis:
> >
> > WARNING: locking bug in hugetlb_no_page
> > https://syzkaller.appspot.com/bug?extid=d07c65298d2c15eafcb0
>
> Thanks Dmitry!
>
> I believe Mike found the root cause and his proposed patches in next-20221111 will fix the problem:
> https://lore.kernel.org/linux-mm/20221111232628.290160-1-mike.kravetz@oracle.com/
>
> Let's keep looking. :) Thanks for your work!
Let's tell syzbot about the fix so that it reports similar issues in future:
#syz fix:
hugetlb: don't delete vma_lock in hugetlb MADV_DONTNEED processing
> Thanks,
> Miaohe Lin
>
>
> >
> >
> >>>> Adding Miaohe Lin on Cc due to previous help with vma_lock potential deadlock
> >>>> situations. Miaohe, does this make sense to you?
> >>>>
> >>>
> >>> Hi Mike,
> >>> This doesn't make sense for me too. Stack #1 shows that syz-executor is releasing
> >>> its address space while stack #0 shows another thread is serving the write syscall.
> >>> In this case, mm->mm_users is 0 and all threads in this process should be serving
> >>> do_exit()? But I could be easily wrong. Also I can't see how vma_lock is locked before
> >>> trying to acquire mmap_lock in above stacks. Might this be a false positive?
> > .
> >
>
> --
> You received this message because you are subscribed to the Google Groups "syzkaller-bugs" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to syzkaller-bugs+unsubscribe@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/syzkaller-bugs/4c5fa6b4-6278-ce2b-73c2-883eb734bdc7%40huawei.com.
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2022-11-14 9:57 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-04 16:00 [syzbot] possible deadlock in hugetlb_fault syzbot
2022-11-12 0:07 ` Mike Kravetz
2022-11-12 4:01 ` Miaohe Lin
2022-11-12 11:33 ` Dmitry Vyukov
2022-11-12 14:14 ` Dmitry Vyukov
2022-11-14 2:35 ` Miaohe Lin
2022-11-14 9:57 ` Dmitry Vyukov
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).