]> git.ipfire.org Git - thirdparty/linux.git/commit
romfs: check sb_set_blocksize() return value
authorDeepanshu Kartikey <kartikey406@gmail.com>
Tue, 13 Jan 2026 08:40:37 +0000 (14:10 +0530)
committerChristian Brauner <brauner@kernel.org>
Tue, 13 Jan 2026 08:56:58 +0000 (09:56 +0100)
commitab7ad7abb3660c58ffffdf07ff3bb976e7e0afa0
treed257f7ed8d48643aef4a7f183aab8cd142bda5ae
parente16688084df76e719b329a1a56b338ec491002e0
romfs: check sb_set_blocksize() return value

romfs_fill_super() ignores the return value of sb_set_blocksize(), which
can fail if the requested block size is incompatible with the block
device's configuration.

This can be triggered by setting a loop device's block size larger than
PAGE_SIZE using ioctl(LOOP_SET_BLOCK_SIZE, 32768), then mounting a romfs
filesystem on that device.

When sb_set_blocksize(sb, ROMBSIZE) is called with ROMBSIZE=4096 but the
device has logical_block_size=32768, bdev_validate_blocksize() fails
because the requested size is smaller than the device's logical block
size. sb_set_blocksize() returns 0 (failure), but romfs ignores this and
continues mounting.

The superblock's block size remains at the device's logical block size
(32768). Later, when sb_bread() attempts I/O with this oversized block
size, it triggers a kernel BUG in folio_set_bh():

    kernel BUG at fs/buffer.c:1582!
    BUG_ON(size > PAGE_SIZE);

Fix by checking the return value of sb_set_blocksize() and failing the
mount with -EINVAL if it returns 0.

Reported-by: syzbot+9c4e33e12283d9437c25@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=9c4e33e12283d9437c25
Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
Link: https://patch.msgid.link/20260113084037.1167887-1-kartikey406@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
fs/romfs/super.c