]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
ext4: move ext4_percpu_param_init() before ext4_mb_init()
authorBaokun Li <libaokun1@huawei.com>
Tue, 9 Dec 2025 13:31:16 +0000 (21:31 +0800)
committerSasha Levin <sashal@kernel.org>
Wed, 4 Mar 2026 12:20:25 +0000 (07:20 -0500)
commitaec095f3cc6cf209effd93278ce35be27db81d73
tree35bc423ace2a836a296e8c8a8a1e87192da2f852
parentc25de5a8846ca37dc2d8ac30fc227500919b6d62
ext4: move ext4_percpu_param_init() before ext4_mb_init()

[ Upstream commit 270564513489d98b721a1e4a10017978d5213bff ]

When running `kvm-xfstests -c ext4/1k -C 1 generic/383` with the
`DOUBLE_CHECK` macro defined, the following panic is triggered:

==================================================================
EXT4-fs error (device vdc): ext4_validate_block_bitmap:423:
                        comm mount: bg 0: bad block bitmap checksum
BUG: unable to handle page fault for address: ff110000fa2cc000
PGD 3e01067 P4D 3e02067 PUD 0
Oops: Oops: 0000 [#1] SMP NOPTI
CPU: 0 UID: 0 PID: 2386 Comm: mount Tainted: G W
                        6.18.0-gba65a4e7120a-dirty #1152 PREEMPT(none)
RIP: 0010:percpu_counter_add_batch+0x13/0xa0
Call Trace:
 <TASK>
 ext4_mark_group_bitmap_corrupted+0xcb/0xe0
 ext4_validate_block_bitmap+0x2a1/0x2f0
 ext4_read_block_bitmap+0x33/0x50
 mb_group_bb_bitmap_alloc+0x33/0x80
 ext4_mb_add_groupinfo+0x190/0x250
 ext4_mb_init_backend+0x87/0x290
 ext4_mb_init+0x456/0x640
 __ext4_fill_super+0x1072/0x1680
 ext4_fill_super+0xd3/0x280
 get_tree_bdev_flags+0x132/0x1d0
 vfs_get_tree+0x29/0xd0
 vfs_cmd_create+0x59/0xe0
 __do_sys_fsconfig+0x4f6/0x6b0
 do_syscall_64+0x50/0x1f0
 entry_SYSCALL_64_after_hwframe+0x76/0x7e
==================================================================

This issue can be reproduced using the following commands:
        mkfs.ext4 -F -q -b 1024 /dev/sda 5G
        tune2fs -O quota,project /dev/sda
        mount /dev/sda /tmp/test

With DOUBLE_CHECK defined, mb_group_bb_bitmap_alloc() reads
and validates the block bitmap. When the validation fails,
ext4_mark_group_bitmap_corrupted() attempts to update
sbi->s_freeclusters_counter. However, this percpu_counter has not been
initialized yet at this point, which leads to the panic described above.

Fix this by moving the execution of ext4_percpu_param_init() to occur
before ext4_mb_init(), ensuring the per-CPU counters are initialized
before they are used.

Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://patch.msgid.link/20251209133116.731350-1-libaokun@huaweicloud.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>
fs/ext4/super.c