possible deadlock in ext4_file_read_iter

From: Wei Chen
Date: Fri Nov 18 2022 - 02:44:07 EST

Next message: Tian, Kevin: "RE: [patch 04/20] genirq/irqdomain: Provide IRQ_DOMAIN_FLAG_MSI_PARENT"
Previous message: Tian, Kevin: "RE: [patch 00/39] genirq, PCI/MSI: Support for per device MSI and PCI/IMS - Part 1 cleanups"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Dear Linux Developer,

Recently when using our tool to fuzz kernel, the following crash was triggered:

HEAD commit: 4fe89d07 Linux v6.0
git tree: upstream
compiler: clang 12.0.0
console output:
https://drive.google.com/file/d/1xryzmZ5xIuxzLyGxjgIBEcCequ085c5y/view?usp=share_link
kernel config: https://drive.google.com/file/d/1ZHRxVTXHL9mENdAPmQYS1DtgbflZ9XsD/view?usp=sharing

Unfortunately, I didn't have a reproducer for this bug yet.

IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: Wei Chen <harperchen1110@xxxxxxxxx>

======================================================
WARNING: possible circular locking dependency detected
6.0.0+ #39 Not tainted
------------------------------------------------------
syz-executor.0/934 is trying to acquire lock:
ffff88804501ce28 (&mm->mmap_lock#2){++++}-{3:3}, at:
__might_fault+0x8f/0x110 mm/memory.c:5582

but task is already holding lock:
ffff88804673ca38 (&sb->s_type->i_mutex_key#8){++++}-{3:3}, at:
inode_lock_shared include/linux/fs.h:766 [inline]
ffff88804673ca38 (&sb->s_type->i_mutex_key#8){++++}-{3:3}, at:
ext4_dio_read_iter fs/ext4/file.c:63 [inline]
ffff88804673ca38 (&sb->s_type->i_mutex_key#8){++++}-{3:3}, at:
ext4_file_read_iter+0x26c/0x770 fs/ext4/file.c:130

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #3 (&sb->s_type->i_mutex_key#8){++++}-{3:3}:
lock_acquire+0x17f/0x430 kernel/locking/lockdep.c:5666
down_read+0x39/0x50 kernel/locking/rwsem.c:1499
inode_lock_shared include/linux/fs.h:766 [inline]
ext4_bmap+0x55/0x410 fs/ext4/inode.c:3157
bmap+0xa1/0xd0 fs/inode.c:1799
jbd2_journal_bmap fs/jbd2/journal.c:971 [inline]
__jbd2_journal_erase fs/jbd2/journal.c:1784 [inline]
jbd2_journal_flush+0x5bc/0xc70 fs/jbd2/journal.c:2490
ext4_ioctl_checkpoint fs/ext4/ioctl.c:1082 [inline]
__ext4_ioctl fs/ext4/ioctl.c:1586 [inline]
ext4_ioctl+0x3208/0x5230 fs/ext4/ioctl.c:1606
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:870 [inline]
__se_sys_ioctl+0xfb/0x170 fs/ioctl.c:856
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd

-> #2 (&journal->j_checkpoint_mutex){+.+.}-{3:3}:
lock_acquire+0x17f/0x430 kernel/locking/lockdep.c:5666
__mutex_lock_common+0x1b7/0x26d0 kernel/locking/mutex.c:603
mutex_lock_io_nested+0x43/0x60 kernel/locking/mutex.c:833
__jbd2_log_wait_for_space+0x21d/0x6f0 fs/jbd2/checkpoint.c:110
add_transaction_credits+0x916/0xbd0 fs/jbd2/transaction.c:298
start_this_handle+0x733/0x16f0 fs/jbd2/transaction.c:422
jbd2__journal_start+0x2ca/0x5b0 fs/jbd2/transaction.c:520
__ext4_journal_start_sb+0x111/0x1d0 fs/ext4/ext4_jbd2.c:105
__ext4_journal_start fs/ext4/ext4_jbd2.h:326 [inline]
ext4_dirty_inode+0x8a/0x100 fs/ext4/inode.c:5963
__mark_inode_dirty+0xb6/0x600 fs/fs-writeback.c:2381
generic_update_time+0x1c7/0x1e0 fs/inode.c:1860
inode_update_time fs/inode.c:1873 [inline]
__file_update_time fs/inode.c:2089 [inline]
file_update_time+0x3b3/0x430 fs/inode.c:2120
ext4_page_mkwrite+0x1c5/0x1410 fs/ext4/inode.c:6084
do_page_mkwrite+0x1a4/0x5f0 mm/memory.c:2971
do_shared_fault mm/memory.c:4589 [inline]
do_fault+0x441/0x9c0 mm/memory.c:4657
handle_pte_fault+0x6eb/0x1660 mm/memory.c:4917
__handle_mm_fault mm/memory.c:5059 [inline]
handle_mm_fault+0xdd3/0x1630 mm/memory.c:5157
do_user_addr_fault+0x896/0x10e0 arch/x86/mm/fault.c:1397
handle_page_fault arch/x86/mm/fault.c:1488 [inline]
exc_page_fault+0xa1/0x1e0 arch/x86/mm/fault.c:1544
asm_exc_page_fault+0x22/0x30 arch/x86/include/asm/idtentry.h:570

-> #1 (sb_pagefaults){.+.+}-{0:0}:
lock_acquire+0x17f/0x430 kernel/locking/lockdep.c:5666
percpu_down_read+0x44/0x190 include/linux/percpu-rwsem.h:51
__sb_start_write include/linux/fs.h:1826 [inline]
sb_start_pagefault include/linux/fs.h:1930 [inline]
ext4_page_mkwrite+0x1ae/0x1410 fs/ext4/inode.c:6083
do_page_mkwrite+0x1a4/0x5f0 mm/memory.c:2971
do_shared_fault mm/memory.c:4589 [inline]
do_fault+0x441/0x9c0 mm/memory.c:4657
handle_pte_fault+0x6eb/0x1660 mm/memory.c:4917
__handle_mm_fault mm/memory.c:5059 [inline]
handle_mm_fault+0xdd3/0x1630 mm/memory.c:5157
do_user_addr_fault+0x896/0x10e0 arch/x86/mm/fault.c:1397
handle_page_fault arch/x86/mm/fault.c:1488 [inline]
exc_page_fault+0xa1/0x1e0 arch/x86/mm/fault.c:1544
asm_exc_page_fault+0x22/0x30 arch/x86/include/asm/idtentry.h:570

-> #0 (&mm->mmap_lock#2){++++}-{3:3}:
check_prev_add kernel/locking/lockdep.c:3095 [inline]
check_prevs_add+0x4f5/0x5d30 kernel/locking/lockdep.c:3214
validate_chain kernel/locking/lockdep.c:3829 [inline]
__lock_acquire+0x4432/0x6080 kernel/locking/lockdep.c:5053
lock_acquire+0x17f/0x430 kernel/locking/lockdep.c:5666
__might_fault+0xb2/0x110 mm/memory.c:5583
__clear_user arch/x86/lib/usercopy_64.c:20 [inline]
clear_user+0xbf/0x150 arch/x86/lib/usercopy_64.c:52
iov_iter_zero+0x48a/0x1450 lib/iov_iter.c:795
iomap_dio_hole_iter fs/iomap/direct-io.c:390 [inline]
iomap_dio_iter fs/iomap/direct-io.c:438 [inline]
__iomap_dio_rw+0x14e1/0x2170 fs/iomap/direct-io.c:602
iomap_dio_rw+0x42/0xa0 fs/iomap/direct-io.c:690
ext4_dio_read_iter fs/ext4/file.c:79 [inline]
ext4_file_read_iter+0x5bc/0x770 fs/ext4/file.c:130
call_read_iter include/linux/fs.h:2181 [inline]
do_iter_readv_writev fs/read_write.c:733 [inline]
do_iter_read+0x710/0xc30 fs/read_write.c:796
vfs_readv fs/read_write.c:916 [inline]
do_readv+0x219/0x410 fs/read_write.c:953
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd

other info that might help us debug this:

Chain exists of:
&mm->mmap_lock#2 --> &journal->j_checkpoint_mutex -->
&sb->s_type->i_mutex_key#8

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
lock(&sb->s_type->i_mutex_key#8);
lock(&journal->j_checkpoint_mutex);
lock(&sb->s_type->i_mutex_key#8);
lock(&mm->mmap_lock#2);

*** DEADLOCK ***

2 locks held by syz-executor.0/934:
#0: ffff8880188000e8 (&f->f_pos_lock){+.+.}-{3:3}, at:
__fdget_pos+0x23a/0x2d0 fs/file.c:1036
#1: ffff88804673ca38 (&sb->s_type->i_mutex_key#8){++++}-{3:3}, at:
inode_lock_shared include/linux/fs.h:766 [inline]
#1: ffff88804673ca38 (&sb->s_type->i_mutex_key#8){++++}-{3:3}, at:
ext4_dio_read_iter fs/ext4/file.c:63 [inline]
#1: ffff88804673ca38 (&sb->s_type->i_mutex_key#8){++++}-{3:3}, at:
ext4_file_read_iter+0x26c/0x770 fs/ext4/file.c:130

stack backtrace:
CPU: 1 PID: 934 Comm: syz-executor.0 Not tainted 6.0.0+ #39
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.13.0-1ubuntu1.1 04/01/2014
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1b1/0x28e lib/dump_stack.c:106
print_circular_bug+0xa95/0xd40 kernel/locking/lockdep.c:2053
check_noncircular+0x2cc/0x390 kernel/locking/lockdep.c:2175
check_prev_add kernel/locking/lockdep.c:3095 [inline]
check_prevs_add+0x4f5/0x5d30 kernel/locking/lockdep.c:3214
validate_chain kernel/locking/lockdep.c:3829 [inline]
__lock_acquire+0x4432/0x6080 kernel/locking/lockdep.c:5053
lock_acquire+0x17f/0x430 kernel/locking/lockdep.c:5666
__might_fault+0xb2/0x110 mm/memory.c:5583
__clear_user arch/x86/lib/usercopy_64.c:20 [inline]
clear_user+0xbf/0x150 arch/x86/lib/usercopy_64.c:52
iov_iter_zero+0x48a/0x1450 lib/iov_iter.c:795
iomap_dio_hole_iter fs/iomap/direct-io.c:390 [inline]
iomap_dio_iter fs/iomap/direct-io.c:438 [inline]
__iomap_dio_rw+0x14e1/0x2170 fs/iomap/direct-io.c:602
iomap_dio_rw+0x42/0xa0 fs/iomap/direct-io.c:690
ext4_dio_read_iter fs/ext4/file.c:79 [inline]
ext4_file_read_iter+0x5bc/0x770 fs/ext4/file.c:130
call_read_iter include/linux/fs.h:2181 [inline]
do_iter_readv_writev fs/read_write.c:733 [inline]
do_iter_read+0x710/0xc30 fs/read_write.c:796
vfs_readv fs/read_write.c:916 [inline]
do_readv+0x219/0x410 fs/read_write.c:953
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x7f020068bded
Code: 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48
89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d
01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f020182cc58 EFLAGS: 00000246 ORIG_RAX: 0000000000000013
RAX: ffffffffffffffda RBX: 00007f02007abf80 RCX: 00007f020068bded
RDX: 0000000000000001 RSI: 0000000020001440 RDI: 0000000000000004
RBP: 00007f02006f8ce0 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007f02007abf80
R13: 00007ffcd43c130f R14: 00007ffcd43c14b0 R15: 00007f020182cdc0
</TASK>

Best,
Wei

Next message: Tian, Kevin: "RE: [patch 04/20] genirq/irqdomain: Provide IRQ_DOMAIN_FLAG_MSI_PARENT"
Previous message: Tian, Kevin: "RE: [patch 00/39] genirq, PCI/MSI: Support for per device MSI and PCI/IMS - Part 1 cleanups"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]