[v6.6] possible deadlock in page_cache_ra_unbounded

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: cc1a1c5b404a Linux 6.6.103
git tree: linux-6.6.y
console output: https://syzkaller.appspot.com/x/log.txt?x=124d5262580000
kernel config: https://syzkaller.appspot.com/x/.config?x=412020841cf033b0
dashboard link: https://syzkaller.appspot.com/bug?extid=caa4095775a6b33cfc59
compiler: Debian clang version 20.1.7 (++20250616065708+6146a88f6049-1~exp1~20250616065826.132), Debian LLD 20.1.7

Unfortunately, I don't have any reproducer for this issue yet.

Downloadable assets:
disk image: https://storage.googleapis.com/syzbot-assets/46c06862a545/disk-cc1a1c5b.raw.xz
vmlinux: https://storage.googleapis.com/syzbot-assets/38ddabb6fc8b/vmlinux-cc1a1c5b.xz
kernel image: https://storage.googleapis.com/syzbot-assets/b130b7d031b5/bzImage-cc1a1c5b.xz

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+caa409...@syzkaller.appspotmail.com

EXT4-fs: Ignoring removed nomblk_io_submit option
EXT4-fs (loop3): Test dummy encryption mode enabled
EXT4-fs (loop3): mounted filesystem 00000000-0000-0000-0000-000000000000 r/w without journal. Quota mode: writeback.
======================================================
WARNING: possible circular locking dependency detected
syzkaller #0 Not tainted
------------------------------------------------------
syz.3.3157/17179 is trying to acquire lock:
ffff888148ddf0c0 (mapping.invalidate_lock#2){++++}-{3:3}, at: filemap_invalidate_lock_shared include/linux/fs.h:859 [inline]
ffff888148ddf0c0 (mapping.invalidate_lock#2){++++}-{3:3}, at: page_cache_ra_unbounded+0xdc/0x770 mm/readahead.c:225

but task is already holding lock:
ffff88802e969810 (&type->i_mutex_dir_key#3){++++}-{3:3}, at: iterate_dir+0xcb/0x580 fs/readdir.c:99

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #4 (&type->i_mutex_dir_key#3){++++}-{3:3}:
down_read+0x46/0x2e0 kernel/locking/rwsem.c:1520
inode_lock_shared include/linux/fs.h:814 [inline]
lookup_slow fs/namei.c:1710 [inline]
lookup_one_unlocked+0x171/0x290 fs/namei.c:2805
lookup_one_positive_unlocked fs/namei.c:2834 [inline]
lookup_positive_unlocked+0x2b/0xb0 fs/namei.c:2874
dquot_quota_on_mount+0x57/0xe0 fs/quota/dquot.c:2574
ext4_quota_on_mount fs/ext4/orphan.c:316 [inline]
ext4_orphan_cleanup+0x6e1/0x1400 fs/ext4/orphan.c:444
__ext4_fill_super fs/ext4/super.c:5608 [inline]
ext4_fill_super+0x5de7/0x66c0 fs/ext4/super.c:5731
get_tree_bdev+0x3e4/0x510 fs/super.c:1591
vfs_get_tree+0x8c/0x280 fs/super.c:1764
do_new_mount+0x24b/0xa40 fs/namespace.c:3377
do_mount fs/namespace.c:3717 [inline]
__do_sys_mount fs/namespace.c:3926 [inline]
__se_sys_mount+0x2da/0x3c0 fs/namespace.c:3903
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x55/0xb0 arch/x86/entry/common.c:81
entry_SYSCALL_64_after_hwframe+0x68/0xd2

-> #3 (&type->s_umount_key#31){++++}-{3:3}:
down_read+0x46/0x2e0 kernel/locking/rwsem.c:1520
__super_lock fs/super.c:58 [inline]
super_lock+0x167/0x360 fs/super.c:117
super_lock_shared fs/super.c:146 [inline]
super_lock_shared_active fs/super.c:1442 [inline]
fs_bdev_sync+0xa4/0x170 fs/super.c:1477
blkdev_flushbuf block/ioctl.c:375 [inline]
blkdev_common_ioctl+0x880/0x23d0 block/ioctl.c:505
blkdev_ioctl+0x4eb/0x6f0 block/ioctl.c:627
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:871 [inline]
__se_sys_ioctl+0xfd/0x170 fs/ioctl.c:857
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x55/0xb0 arch/x86/entry/common.c:81
entry_SYSCALL_64_after_hwframe+0x68/0xd2

-> #2 (&bdev->bd_holder_lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:603 [inline]
__mutex_lock+0x129/0xcc0 kernel/locking/mutex.c:747
bd_finish_claiming+0x22f/0x3f0 block/bdev.c:568
blkdev_get_by_dev+0x45c/0x600 block/bdev.c:801
bdev_open_by_dev+0x77/0x100 block/bdev.c:842
setup_bdev_super+0x59/0x660 fs/super.c:1496
mount_bdev+0x1dd/0x2d0 fs/super.c:1640
legacy_get_tree+0xea/0x180 fs/fs_context.c:662
vfs_get_tree+0x8c/0x280 fs/super.c:1764
do_new_mount+0x24b/0xa40 fs/namespace.c:3377
init_mount+0xd2/0x120 fs/init.c:25
do_mount_root+0x97/0x230 init/do_mounts.c:166
mount_root_generic+0x195/0x3c0 init/do_mounts.c:205
prepare_namespace+0xc2/0x100 init/do_mounts.c:489
kernel_init_freeable+0x413/0x570 init/main.c:1566
kernel_init+0x1d/0x1c0 init/main.c:1443
ret_from_fork+0x48/0x80 arch/x86/kernel/process.c:152
ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:293

-> #1 (bdev_lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:603 [inline]
__mutex_lock+0x129/0xcc0 kernel/locking/mutex.c:747
bd_prepare_to_claim+0x1ba/0x480 block/bdev.c:510
truncate_bdev_range+0x4e/0x260 block/bdev.c:105
blkdev_fallocate+0x4e0/0x670 block/fops.c:783
vfs_fallocate+0x58e/0x700 fs/open.c:324
ksys_fallocate fs/open.c:347 [inline]
__do_sys_fallocate fs/open.c:355 [inline]
__se_sys_fallocate fs/open.c:353 [inline]
__x64_sys_fallocate+0xc1/0x110 fs/open.c:353
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x55/0xb0 arch/x86/entry/common.c:81
entry_SYSCALL_64_after_hwframe+0x68/0xd2

-> #0 (mapping.invalidate_lock#2){++++}-{3:3}:
check_prev_add kernel/locking/lockdep.c:3134 [inline]
check_prevs_add kernel/locking/lockdep.c:3253 [inline]
validate_chain kernel/locking/lockdep.c:3869 [inline]
__lock_acquire+0x2ddb/0x7c80 kernel/locking/lockdep.c:5137
lock_acquire+0x197/0x410 kernel/locking/lockdep.c:5754
down_read+0x46/0x2e0 kernel/locking/rwsem.c:1520
filemap_invalidate_lock_shared include/linux/fs.h:859 [inline]
page_cache_ra_unbounded+0xdc/0x770 mm/readahead.c:225
page_cache_sync_readahead include/linux/pagemap.h:1325 [inline]
ext4_readdir+0xb71/0x39d0 fs/ext4/dir.c:197
iterate_dir+0x1c2/0x580 fs/readdir.c:106
__do_sys_getdents64 fs/readdir.c:405 [inline]
__se_sys_getdents64+0xe9/0x260 fs/readdir.c:390
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x55/0xb0 arch/x86/entry/common.c:81
entry_SYSCALL_64_after_hwframe+0x68/0xd2

other info that might help us debug this:

Chain exists of:
mapping.invalidate_lock#2 --> &type->s_umount_key#31 --> &type->i_mutex_dir_key#3

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
rlock(&type->i_mutex_dir_key#3);
lock(&type->s_umount_key#31);
lock(&type->i_mutex_dir_key#3);
rlock(mapping.invalidate_lock#2);

*** DEADLOCK ***

2 locks held by syz.3.3157/17179:
#0: ffff8880777d00c8 (&f->f_pos_lock){+.+.}-{3:3}, at: __fdget_pos+0x2a3/0x330 fs/file.c:1040
#1: ffff88802e969810 (&type->i_mutex_dir_key#3){++++}-{3:3}, at: iterate_dir+0xcb/0x580 fs/readdir.c:99

stack backtrace:
CPU: 1 PID: 17179 Comm: syz.3.3157 Not tainted syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/12/2025
Call Trace:
<TASK>
dump_stack_lvl+0x16c/0x230 lib/dump_stack.c:106
check_noncircular+0x2bd/0x3c0 kernel/locking/lockdep.c:2187
check_prev_add kernel/locking/lockdep.c:3134 [inline]
check_prevs_add kernel/locking/lockdep.c:3253 [inline]
validate_chain kernel/locking/lockdep.c:3869 [inline]
__lock_acquire+0x2ddb/0x7c80 kernel/locking/lockdep.c:5137
lock_acquire+0x197/0x410 kernel/locking/lockdep.c:5754
down_read+0x46/0x2e0 kernel/locking/rwsem.c:1520
filemap_invalidate_lock_shared include/linux/fs.h:859 [inline]
page_cache_ra_unbounded+0xdc/0x770 mm/readahead.c:225
page_cache_sync_readahead include/linux/pagemap.h:1325 [inline]
ext4_readdir+0xb71/0x39d0 fs/ext4/dir.c:197
iterate_dir+0x1c2/0x580 fs/readdir.c:106
__do_sys_getdents64 fs/readdir.c:405 [inline]
__se_sys_getdents64+0xe9/0x260 fs/readdir.c:390
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x55/0xb0 arch/x86/entry/common.c:81
entry_SYSCALL_64_after_hwframe+0x68/0xd2
RIP: 0033:0x7fd085b8ebe9
Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 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 a8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007fd083df6038 EFLAGS: 00000246 ORIG_RAX: 00000000000000d9
RAX: ffffffffffffffda RBX: 00007fd085dc5fa0 RCX: 00007fd085b8ebe9
RDX: 00000000000000b3 RSI: 0000200000000200 RDI: 0000000000000005
RBP: 00007fd085c11e19 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007fd085dc6038 R14: 00007fd085dc5fa0 R15: 00007ffca9bdc688
</TASK>
EXT4-fs error (device loop3): ext4_readdir:263: inode #12: block 80: comm syz.3.3157: path /81/bus/file0: bad entry in directory: rec_len is smaller than minimal - offset=12, inode=6, rec_len=0, size=4096 fake=0
EXT4-fs (loop3): Remounting filesystem read-only


---
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 syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

If the report is already addressed, let syzbot know by replying with:
#syz fix: exact-commit-title

If you want to overwrite report's subsystems, reply with:
#syz set subsystems: new-subsystem
(See the list of subsystem names on the web dashboard)

If the report is a duplicate of another one, reply with:
#syz dup: exact-subject-of-another-report

If you want to undo deduplication, reply with:
#syz undup