[Cluster-devel] [PATCH v2] fs: gfs2: add validation checks for size of superblock

Anant Thazhemadam anant.thazhemadam at gmail.com
Wed Oct 14 16:31:09 UTC 2020


In gfs2_check_sb(), no validation checks are performed with regards to
the size of the superblock.
syzkaller detected a slab-out-of-bounds bug that was primarily caused
because the block size for a superblock was set to zero.
A valid size for a superblock is a power of 2 between 512 and PAGE_SIZE.
Performing validation checks and ensuring that the size of the superblock
is valid fixes this bug.

Reported-by: syzbot+af90d47a37376844e731 at syzkaller.appspotmail.com
Tested-by: syzbot+af90d47a37376844e731 at syzkaller.appspotmail.com
Suggested-by: Andrew Price <anprice at redhat.com>
Signed-off-by: Anant Thazhemadam <anant.thazhemadam at gmail.com>
---

Changes in v2:

	* Completely dropped the changes proposed in v1. Instead,
	  validity checks for superblock size have been introduced. 
	  (Suggested by Andrew Price<anprice at redhat.com>)

	* Addded a "Suggested-by" tag accrediting the patch idea to
	  Andrew. If there's any issue with that, please let me know.

	* Changed the commit header and commit message appropriately.

	* Updated "Reported-by" and "Tested-by" tags to the same instance
	  of the bug that was detected earlier (non consequential change).


 fs/gfs2/ops_fstype.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index 6d18d2c91add..f0605fae2c4c 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -169,6 +169,13 @@ static int gfs2_check_sb(struct gfs2_sbd *sdp, int silent)
 		return -EINVAL;
 	}
 
+	/* Check if the size of the block is valid - a power of 2 between 512 and  PAGE_SIZE */
+	if (sb->sb_bsize < 512 || sb->sb_bsize > PAGE_SIZE || (sb->sb_bsize & (sb->sb_bsize - 1))) {
+		if (!silent)
+			pr_warn("Invalid superblock size\n");
+		return -EINVAL;
+	}
+
 	/*  If format numbers match exactly, we're done.  */
 
 	if (sb->sb_fs_format == GFS2_FORMAT_FS &&
-- 
2.25.1




More information about the Cluster-devel mailing list