| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <20250519155149.2382-1-superman.xpt@gmail.com>
Date: Mon, 19 May 2025 08:51:49 -0700
From: Penglei Jiang <superman.xpt@...il.com>
To: syzbot+3a88c590edd72179657c@...kaller.appspotmail.com,
clm@...com,
dsterba@...e.com,
josef@...icpanda.com
Cc: linux-btrfs@...r.kernel.org,
linux-kernel@...r.kernel.org,
syzkaller-bugs@...glegroups.com
Subject: Re: [syzbot] [btrfs?] possible deadlock in btrfs_quota_enable (2)
On Sat, 10 May 2025 13:51:24 -0700, syzbot wrote:
> syz.4.219/7436 is trying to acquire lock:
> ffff88806ce71918 (&fs_info->qgroup_ioctl_lock){+.+.}-{4:4}, at: btrfs_quota_enable+0x2be/0x1d50 fs/btrfs/qgroup.c:1059
>
> but task is already holding lock:
> ffff88806ce724f8 (btrfs_trans_num_extwriters){++++}-{0:0}, at: join_transaction+0x164/0xd70 fs/btrfs/transaction.c:320
>
> which lock already depends on the new lock.
>
>
> the existing dependency chain (in reverse order) is:
>
> -> #5 (btrfs_trans_num_extwriters){++++}-{0:0}:
> lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5866
> join_transaction+0x1a4/0xd70 fs/btrfs/transaction.c:321
> start_transaction+0x6ae/0x1620 fs/btrfs/transaction.c:705
> btrfs_dirty_inode+0x9f/0x190 fs/btrfs/inode.c:6129
> inode_update_time fs/inode.c:2076 [inline]
> touch_atime+0x2f6/0x6d0 fs/inode.c:2149
> file_accessed include/linux/fs.h:2599 [inline]
> filemap_read+0x1024/0x11d0 mm/filemap.c:2774
> __kernel_read+0x469/0x8c0 fs/read_write.c:528
> integrity_kernel_read+0x89/0xd0 security/integrity/iint.c:28
> ima_calc_file_hash_tfm security/integrity/ima/ima_crypto.c:480 [inline]
> ima_calc_file_shash security/integrity/ima/ima_crypto.c:511 [inline]
> ima_calc_file_hash+0x152c/0x18d0 security/integrity/ima/ima_crypto.c:568
> ima_collect_measurement+0x42e/0x8e0 security/integrity/ima/ima_api.c:293
> process_measurement+0x1121/0x1a40 security/integrity/ima/ima_main.c:385
> ima_file_check+0xd7/0x120 security/integrity/ima/ima_main.c:613
> security_file_post_open+0xbb/0x290 security/security.c:3130
> do_open fs/namei.c:3882 [inline]
> path_openat+0x2f26/0x3830 fs/namei.c:4039
> do_filp_open+0x1fa/0x410 fs/namei.c:4066
> do_sys_openat2+0x121/0x1c0 fs/open.c:1429
> do_sys_open fs/open.c:1444 [inline]
> __do_sys_openat fs/open.c:1460 [inline]
> __se_sys_openat fs/open.c:1455 [inline]
> __x64_sys_openat+0x138/0x170 fs/open.c:1455
> do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
> do_syscall_64+0xf6/0x210 arch/x86/entry/syscall_64.c:94
> entry_SYSCALL_64_after_hwframe+0x77/0x7f
>
> -> #4 (btrfs_trans_num_writers){++++}-{0:0}:
> reacquire_held_locks+0x127/0x1d0 kernel/locking/lockdep.c:5383
> __lock_release kernel/locking/lockdep.c:5572 [inline]
> lock_release+0x1b4/0x3e0 kernel/locking/lockdep.c:5887
> percpu_up_read include/linux/percpu-rwsem.h:100 [inline]
> __sb_end_write include/linux/fs.h:1778 [inline]
> sb_end_intwrite+0x26/0x1c0 include/linux/fs.h:1895
> __btrfs_end_transaction+0x248/0x640 fs/btrfs/transaction.c:1075
> btrfs_dirty_inode+0x14c/0x190 fs/btrfs/inode.c:6143
> inode_update_time fs/inode.c:2076 [inline]
> __file_update_time fs/inode.c:2305 [inline]
> file_update_time+0x344/0x490 fs/inode.c:2335
> btrfs_page_mkwrite+0x634/0x16a0 fs/btrfs/file.c:1814
> do_page_mkwrite+0x14a/0x310 mm/memory.c:3287
> wp_page_shared mm/memory.c:3688 [inline]
> do_wp_page+0x2626/0x5760 mm/memory.c:3907
> handle_pte_fault mm/memory.c:6013 [inline]
> __handle_mm_fault+0x1028/0x5380 mm/memory.c:6140
> handle_mm_fault+0x2d5/0x7f0 mm/memory.c:6309
> do_user_addr_fault+0xa81/0x1390 arch/x86/mm/fault.c:1337
> handle_page_fault arch/x86/mm/fault.c:1480 [inline]
> exc_page_fault+0x68/0x110 arch/x86/mm/fault.c:1538
> asm_exc_page_fault+0x26/0x30 arch/x86/include/asm/idtentry.h:623
>
> -> #3 (sb_pagefaults#2){.+.+}-{0:0}:
> lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5866
> percpu_down_read include/linux/percpu-rwsem.h:52 [inline]
> __sb_start_write include/linux/fs.h:1783 [inline]
> sb_start_pagefault include/linux/fs.h:1948 [inline]
> btrfs_page_mkwrite+0x3b2/0x16a0 fs/btrfs/file.c:1798
> do_page_mkwrite+0x14a/0x310 mm/memory.c:3287
> do_shared_fault mm/memory.c:5594 [inline]
> do_fault mm/memory.c:5656 [inline]
> do_pte_missing mm/memory.c:4160 [inline]
> handle_pte_fault mm/memory.c:5997 [inline]
> __handle_mm_fault+0x18d2/0x5380 mm/memory.c:6140
> handle_mm_fault+0x2d5/0x7f0 mm/memory.c:6309
> do_user_addr_fault+0x764/0x1390 arch/x86/mm/fault.c:1388
> handle_page_fault arch/x86/mm/fault.c:1480 [inline]
> exc_page_fault+0x68/0x110 arch/x86/mm/fault.c:1538
> asm_exc_page_fault+0x26/0x30 arch/x86/include/asm/idtentry.h:623
>
> -> #2 (&mm->mmap_lock){++++}-{4:4}:
> lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5866
> down_read_killable+0x50/0x350 kernel/locking/rwsem.c:1547
> mmap_read_lock_killable+0x1d/0x70 include/linux/mmap_lock.h:193
> get_mmap_lock_carefully mm/memory.c:6355 [inline]
> lock_mm_and_find_vma+0x2a8/0x300 mm/memory.c:6406
> do_user_addr_fault+0x331/0x1390 arch/x86/mm/fault.c:1360
> handle_page_fault arch/x86/mm/fault.c:1480 [inline]
> exc_page_fault+0x68/0x110 arch/x86/mm/fault.c:1538
> asm_exc_page_fault+0x26/0x30 arch/x86/include/asm/idtentry.h:623
> filldir64+0x2b3/0x690 fs/readdir.c:371
> dir_emit include/linux/fs.h:3861 [inline]
> kernfs_fop_readdir+0x534/0x870 fs/kernfs/dir.c:1907
> iterate_dir+0x5ac/0x770 fs/readdir.c:108
> __do_sys_getdents64 fs/readdir.c:403 [inline]
> __se_sys_getdents64+0xe4/0x260 fs/readdir.c:389
> do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
> do_syscall_64+0xf6/0x210 arch/x86/entry/syscall_64.c:94
> entry_SYSCALL_64_after_hwframe+0x77/0x7f
>
> -> #1 (&root->kernfs_rwsem){++++}-{4:4}:
> lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5866
> down_write+0x96/0x1f0 kernel/locking/rwsem.c:1577
> kernfs_add_one+0x41/0x520 fs/kernfs/dir.c:791
> kernfs_create_dir_ns+0xde/0x130 fs/kernfs/dir.c:1091
> sysfs_create_dir_ns+0x123/0x280 fs/sysfs/dir.c:59
> create_dir lib/kobject.c:73 [inline]
> kobject_add_internal+0x59f/0xb40 lib/kobject.c:240
> kobject_add_varg lib/kobject.c:374 [inline]
> kobject_init_and_add+0x125/0x190 lib/kobject.c:457
> btrfs_sysfs_add_qgroups+0x111/0x2b0 fs/btrfs/sysfs.c:2616
> btrfs_quota_enable+0x278/0x1d50 fs/btrfs/qgroup.c:1030
> btrfs_ioctl_quota_ctl+0x183/0x1c0 fs/btrfs/ioctl.c:3676
> vfs_ioctl fs/ioctl.c:51 [inline]
> __do_sys_ioctl fs/ioctl.c:906 [inline]
> __se_sys_ioctl+0xf9/0x170 fs/ioctl.c:892
> do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
> do_syscall_64+0xf6/0x210 arch/x86/entry/syscall_64.c:94
> entry_SYSCALL_64_after_hwframe+0x77/0x7f
>
> -> #0 (&fs_info->qgroup_ioctl_lock){+.+.}-{4:4}:
> check_prev_add kernel/locking/lockdep.c:3166 [inline]
> check_prevs_add kernel/locking/lockdep.c:3285 [inline]
> validate_chain+0xb9b/0x2140 kernel/locking/lockdep.c:3909
> __lock_acquire+0xaac/0xd20 kernel/locking/lockdep.c:5235
> lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5866
> __mutex_lock_common kernel/locking/mutex.c:601 [inline]
> __mutex_lock+0x182/0xe80 kernel/locking/mutex.c:746
> btrfs_quota_enable+0x2be/0x1d50 fs/btrfs/qgroup.c:1059
> btrfs_ioctl_quota_ctl+0x183/0x1c0 fs/btrfs/ioctl.c:3676
> vfs_ioctl fs/ioctl.c:51 [inline]
> __do_sys_ioctl fs/ioctl.c:906 [inline]
> __se_sys_ioctl+0xf9/0x170 fs/ioctl.c:892
> do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
> do_syscall_64+0xf6/0x210 arch/x86/entry/syscall_64.c:94
> entry_SYSCALL_64_after_hwframe+0x77/0x7f
>
> other info that might help us debug this:
>
> Chain exists of:
> &fs_info->qgroup_ioctl_lock --> btrfs_trans_num_writers --> btrfs_trans_num_extwriters
>
> Possible unsafe locking scenario:
>
> CPU0 CPU1
> ---- ----
> rlock(btrfs_trans_num_extwriters);
> lock(btrfs_trans_num_writers);
> lock(btrfs_trans_num_extwriters);
> lock(&fs_info->qgroup_ioctl_lock);
>
> *** DEADLOCK ***
>
> 5 locks held by syz.4.219/7436:
> #0: ffff8880259ac420 (sb_writers#17){.+.+}-{0:0}, at: mnt_want_write_file+0x60/0x200 fs/namespace.c:600
> #1: ffff88806ce70bd0 (&fs_info->subvol_sem){+.+.}-{4:4}, at: btrfs_ioctl_quota_ctl+0x178/0x1c0 fs/btrfs/ioctl.c:3675
> #2: ffff8880259ac610 (sb_internal#3){.+.+}-{0:0}, at: btrfs_quota_enable+0x2b1/0x1d50 fs/btrfs/qgroup.c:1057
> #3: ffff88806ce724d0 (btrfs_trans_num_writers){++++}-{0:0}, at: join_transaction+0x164/0xd70 fs/btrfs/transaction.c:320
> #4: ffff88806ce724f8 (btrfs_trans_num_extwriters){++++}-{0:0}, at: join_transaction+0x164/0xd70 fs/btrfs/transaction.c:320
We assign a sequence number to each lock to represent the order in which
they are held:
0: vfs freeze semaphores
1: qgroup_ioctl_lock
2: kernfs_rwsem
3: mmap_lock
4: btrfs_trans_num_writers
5: btrfs_trans_num_extwriters
cpu-0
=====
0: vfs freeze semaphores
4: btrfs_trans_num_writers
5: btrfs_trans_num_extwriters
1: qgroup_ioctl_lock
cpu-1
=====
1: qgroup_ioctl_lock
2: kernfs_rwsem
cpu-2
=====
2: kernfs_rwsem
3: mmap_lock
cpu-3
=====
3: mmap_lock
4: btrfs_trans_num_writers
5: btrfs_trans_num_extwriters
I believe we should adjust the order of locks in the CPU-0 call stack by moving
the acquisition of qgroup_ioctl_lock inside the start_transaction() function.
After the adjustment, it becomes as follows:
0: vfs freeze semaphores
1: qgroup_ioctl_lock
4: btrfs_trans_num_writers
5: btrfs_trans_num_extwriters
static struct btrfs_trans_handle *
start_transaction(struct btrfs_root *root, unsigned int num_items,
unsigned int type, enum btrfs_reserve_flush_enum flush,
bool enforce_qgroups)
{
...
/*
* If we are JOIN_NOLOCK we're already committing a transaction and
* waiting on this guy, so we don't need to do the sb_start_intwrite
* because we're already holding a ref. We need this because we could
* have raced in and did an fsync() on a file which can kick a commit
* and then we deadlock with somebody doing a freeze.
*
* If we are ATTACH, it means we just want to catch the current
* transaction and commit it, so we needn't do sb_start_intwrite().
*/
if (type & __TRANS_FREEZABLE)
sb_start_intwrite(fs_info->sb);
mutex_lock(&fs_info->qgroup_ioctl_lock);
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
if (may_wait_transaction(fs_info, type))
wait_current_trans(fs_info);
do {
ret = join_transaction(fs_info, type);
if (ret == -EBUSY) {
wait_current_trans(fs_info);
if (unlikely(type == TRANS_ATTACH ||
type == TRANS_JOIN_NOSTART))
ret = -ENOENT;
}
} while (ret == -EBUSY);
...
}
Powered by blists - more mailing lists