]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
ext4: handle wraparound when searching for blocks for indirect mapped blocks
authorTheodore Ts'o <tytso@mit.edu>
Fri, 3 Apr 2026 13:35:36 +0000 (09:35 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 11 Apr 2026 12:16:36 +0000 (14:16 +0200)
[ Upstream commit bb81702370fad22c06ca12b6e1648754dbc37e0f ]

Commit 4865c768b563 ("ext4: always allocate blocks only from groups
inode can use") restricts what blocks will be allocated for indirect
block based files to block numbers that fit within 32-bit block
numbers.

However, when using a review bot running on the latest Gemini LLM to
check this commit when backporting into an LTS based kernel, it raised
this concern:

   If ac->ac_g_ex.fe_group is >= ngroups (for instance, if the goal
   group was populated via stream allocation from s_mb_last_groups),
   then start will be >= ngroups.

   Does this allow allocating blocks beyond the 32-bit limit for
   indirect block mapped files? The commit message mentions that
   ext4_mb_scan_groups_linear() takes care to not select unsupported
   groups. However, its loop uses group = *start, and the very first
   iteration will call ext4_mb_scan_group() with this unsupported
   group because next_linear_group() is only called at the end of the
   iteration.

After reviewing the code paths involved and considering the LLM
review, I determined that this can happen when there is a file system
where some files/directories are extent-mapped and others are
indirect-block mapped.  To address this, add a safety clamp in
ext4_mb_scan_groups().

Fixes: 4865c768b563 ("ext4: always allocate blocks only from groups inode can use")
Cc: Jan Kara <jack@suse.cz>
Reviewed-by: Baokun Li <libaokun@linux.alibaba.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Link: https://patch.msgid.link/20260326045834.1175822-1-tytso@mit.edu
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org
[ adapted fix from ext4_mb_scan_groups() to inline equivalent in ext4_mb_regular_allocator() ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/ext4/mballoc.c

index 3b8d35a98d4d16bf03d21ff56d06c3fda096c769..df6fa0a568559eb37bdb652b767d0930fe9fb143 100644 (file)
@@ -2733,6 +2733,8 @@ repeat:
                 * from the goal value specified
                 */
                group = ac->ac_g_ex.fe_group;
+               if (group >= ngroups)
+                       group = 0;
                ac->ac_groups_linear_remaining = sbi->s_mb_max_linear_groups;
                prefetch_grp = group;