From: Andreas Gruenbacher <agruenba@redhat.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>,
Edward Adam Davis <eadavis@qq.com>,
gfs2@lists.linux.dev,
syzbot <syzbot+9d20c3ad7d29227de28d@syzkaller.appspotmail.com>
Subject: Re: [syzbot] [mm?] [gfs2?] WARNING in read_rindex_entry
Date: Fri, 21 Aug 2026 14:49:43 +0200 [thread overview]
Message-ID: <20260821124944.1648686-1-agruenba@redhat.com> (raw)
In-Reply-To: <20260819171409.e30eed94c3f7eb44b38e5ec7@linux-foundation.org>
On Thu, Aug 20, 2026 at 2:24 AM Andrew Morton <akpm@linux-foundation.org> wrote:
> On Wed, 19 Aug 2026 15:49:42 -0700 syzbot <syzbot+9d20c3ad7d29227de28d@syzkaller.appspotmail.com> wrote:
> > Hello,
> >
> > syzbot found the following issue on:
> >
> > HEAD commit: 15ef2f78c49d Merge tag 'input-for-v7.2-rc7' of git://git.k..
> > git tree: upstream
> > console+strace: https://syzkaller.appspot.com/x/log.txt?x=10871279580000
> > kernel config: https://syzkaller.appspot.com/x/.config?x=a0fdc3b566746377
> > dashboard link: https://syzkaller.appspot.com/bug?extid=9d20c3ad7d29227de28d
> > compiler: gcc (Debian 14.2.0-19) 14.2.0, GNU ld (GNU Binutils for Debian) 2.44
> > syz repro: https://syzkaller.appspot.com/x/repro.syz?x=14871279580000
> > C reproducer: https://syzkaller.appspot.com/x/repro.c?x=100d2a25580000
> >
> > Downloadable assets:
> > disk image: https://storage.googleapis.com/syzbot-assets/c2a8c7d3a889/disk-15ef2f78.raw.xz
> > vmlinux: https://storage.googleapis.com/syzbot-assets/2da78afed345/vmlinux-15ef2f78.xz
> > kernel image: https://storage.googleapis.com/syzbot-assets/c4ffd97e4708/bzImage-15ef2f78.xz
> > mounted in repro: https://storage.googleapis.com/syzbot-assets/eaa4120812b6/mount_0.gz
> > fsck result: failed (log: https://syzkaller.appspot.com/x/fsck.log?x=17f52a25580000)
> >
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: syzbot+9d20c3ad7d29227de28d@syzkaller.appspotmail.com
>
> Thanks. Mounting an intentionally corrupted fs image, I assume.
>
> It appears that GFS2 forgot to validate the `ri_length' which it read
> from disk. That gets used in compute_bitstructs()->kzalloc_objs() and
> the memory allocator didn't like the excessively-sized allocation
> attempt.
Thanks for looking into this. This commit (attached below) from the
for-later branch should prevent that from happening:
https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-9/-/merge_requests/8684#note_3716004816
Thanks,
Andreas
--
gfs2: Improve resource group validation and error handling
Validate the resource group boundaries, check against the device size, simplify
the initialization logic, and check for resource group overlaps.
Reported-by: syzbot+9d20c3ad7d29227de28d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=9d20c3ad7d29227de28d
Tested-by: syzbot+9d20c3ad7d29227de28d@syzkaller.appspotmail.com
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
---
fs/gfs2/rgrp.c | 123 ++++++++++++++++++++++++++-----------------------
1 file changed, 65 insertions(+), 58 deletions(-)
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index f6048a73e5c3..3ac751eea055 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -745,6 +745,7 @@ void gfs2_clear_rgrpd(struct gfs2_sbd *sdp)
/**
* compute_bitstructs - Compute the bitmap sizes
+ * @sb: The superblock
* @rgd: The resource group descriptor
*
* Calculates bitmap descriptors, one for each block that contains bitmap data
@@ -752,84 +753,74 @@ void gfs2_clear_rgrpd(struct gfs2_sbd *sdp)
* Returns: errno
*/
-static int compute_bitstructs(struct gfs2_rgrpd *rgd)
+static int compute_bitstructs(struct super_block *sb, struct gfs2_rgrpd *rgd)
{
struct gfs2_sbd *sdp = rgd->rd_sbd;
struct gfs2_bitmap *bi;
- u32 length = rgd->rd_length; /* # blocks in hdr & bitmap */
+ u32 expected_length;
u32 bytes_left, bytes;
+ u64 data_end;
int x;
- if (!length)
- return -EINVAL;
+ /*
+ * The first resource group block has a gfs2_rgrp header; the remaining
+ * blocks have a gfs2_meta_header header. The rest of each block is
+ * filled with bitmap data.
+ */
+
+ if (rgd->rd_addr <= (GFS2_SB_ADDR >> sdp->sd_fsb2bb_shift)) {
+ gfs2_consist_rgrpd(rgd);
+ return -EIO;
+ }
+ if (check_add_overflow(rgd->rd_data0, rgd->rd_data, &data_end) ||
+ rgd->rd_data == 0 || data_end > sb_bdev_nr_blocks(sb)) {
+ gfs2_consist_rgrpd(rgd);
+ return -EIO;
+ }
+ if (rgd->rd_bitbytes != DIV_ROUND_UP(rgd->rd_data, GFS2_NBBY)) {
+ gfs2_consist_rgrpd(rgd);
+ return -EIO;
+ }
+ expected_length = DIV_ROUND_UP(rgd->rd_bitbytes +
+ sizeof(struct gfs2_rgrp) - sizeof(struct gfs2_meta_header),
+ sdp->sd_sb.sb_bsize - sizeof(struct gfs2_meta_header));
+ if (rgd->rd_length != expected_length) {
+ gfs2_consist_rgrpd(rgd);
+ return -EIO;
+ }
+ if (rgd->rd_data0 < rgd->rd_addr + rgd->rd_length) {
+ gfs2_consist_rgrpd(rgd);
+ return -EIO;
+ }
- rgd->rd_bits = kzalloc_objs(struct gfs2_bitmap, length, GFP_NOFS);
+ rgd->rd_bits = kzalloc_objs(struct gfs2_bitmap, rgd->rd_length, GFP_NOFS);
if (!rgd->rd_bits)
return -ENOMEM;
bytes_left = rgd->rd_bitbytes;
- for (x = 0; x < length; x++) {
+ for (x = 0; x < rgd->rd_length; x++) {
bi = rgd->rd_bits + x;
bi->bi_flags = 0;
- /* small rgrp; bitmap stored completely in header block */
- if (length == 1) {
- bytes = bytes_left;
- bi->bi_offset = sizeof(struct gfs2_rgrp);
+ if (x == 0) {
+ /* header block */
bi->bi_start = 0;
- bi->bi_bytes = bytes;
- bi->bi_blocks = bytes * GFS2_NBBY;
- /* header block */
- } else if (x == 0) {
- bytes = sdp->sd_sb.sb_bsize - sizeof(struct gfs2_rgrp);
bi->bi_offset = sizeof(struct gfs2_rgrp);
- bi->bi_start = 0;
- bi->bi_bytes = bytes;
- bi->bi_blocks = bytes * GFS2_NBBY;
- /* last block */
- } else if (x + 1 == length) {
- bytes = bytes_left;
- bi->bi_offset = sizeof(struct gfs2_meta_header);
- bi->bi_start = rgd->rd_bitbytes - bytes_left;
- bi->bi_bytes = bytes;
- bi->bi_blocks = bytes * GFS2_NBBY;
- /* other blocks */
} else {
- bytes = sdp->sd_sb.sb_bsize -
- sizeof(struct gfs2_meta_header);
+ /* bitmap-only block */
+ struct gfs2_bitmap *prev = bi - 1;
+
+ bi->bi_start = prev->bi_start + prev->bi_bytes;
bi->bi_offset = sizeof(struct gfs2_meta_header);
- bi->bi_start = rgd->rd_bitbytes - bytes_left;
- bi->bi_bytes = bytes;
- bi->bi_blocks = bytes * GFS2_NBBY;
}
-
+ bytes = sdp->sd_sb.sb_bsize - bi->bi_offset;
+ if (bytes > bytes_left)
+ bytes = bytes_left;
+ bi->bi_bytes = bytes;
+ bi->bi_blocks = bytes * GFS2_NBBY;
bytes_left -= bytes;
}
-
- if (bytes_left) {
- gfs2_consist_rgrpd(rgd);
- return -EIO;
- }
- bi = rgd->rd_bits + (length - 1);
- if ((bi->bi_start + bi->bi_bytes) * GFS2_NBBY != rgd->rd_data) {
- gfs2_lm(sdp,
- "ri_addr=%llu "
- "ri_length=%u "
- "ri_data0=%llu "
- "ri_data=%u "
- "ri_bitbytes=%u "
- "start=%u len=%u offset=%u\n",
- (unsigned long long)rgd->rd_addr,
- rgd->rd_length,
- (unsigned long long)rgd->rd_data0,
- rgd->rd_data,
- rgd->rd_bitbytes,
- bi->bi_start, bi->bi_bytes, bi->bi_offset);
- gfs2_consist_rgrpd(rgd);
- return -EIO;
- }
-
return 0;
}
@@ -864,6 +855,7 @@ static int rgd_insert(struct gfs2_rgrpd *rgd)
{
struct gfs2_sbd *sdp = rgd->rd_sbd;
struct rb_node **newn = &sdp->sd_rindex_tree.rb_node, *parent = NULL;
+ struct rb_node *prevn;
/* Figure out where to put new node */
while (*newn) {
@@ -882,6 +874,19 @@ static int rgd_insert(struct gfs2_rgrpd *rgd)
rb_link_node(&rgd->rd_node, parent, newn);
rb_insert_color(&rgd->rd_node, &sdp->sd_rindex_tree);
sdp->sd_rgrps++;
+
+ prevn = rb_prev(&rgd->rd_node);
+ if (prevn) {
+ struct gfs2_rgrpd *prev =
+ rb_entry(prevn, struct gfs2_rgrpd, rd_node);
+
+ if (prev->rd_data0 + prev->rd_data > rgd->rd_addr) {
+ fs_err(sdp, "overlapping resource groups.\n");
+ rb_erase(&rgd->rd_node, &sdp->sd_rindex_tree);
+ return -ENOENT;
+ }
+ }
+
return 0;
}
@@ -928,7 +933,7 @@ static int read_rindex_entry(struct gfs2_inode *ip)
if (error)
goto fail;
- error = compute_bitstructs(rgd);
+ error = compute_bitstructs(sdp->sd_vfs, rgd);
if (error)
goto fail_glock;
@@ -944,7 +949,9 @@ static int read_rindex_entry(struct gfs2_inode *ip)
return 0;
}
- error = 0; /* someone else read in the rgrp; free it and ignore it */
+ /* If someone else read in the rgrp, free it and ignore it. */
+ if (error == -EEXIST)
+ error = 0;
fail_glock:
gfs2_glock_put(rgd->rd_gl);
--
2.55.0
next prev parent reply other threads:[~2026-08-21 12:49 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-19 22:49 [syzbot] [mm?] [gfs2?] WARNING in read_rindex_entry syzbot
2026-08-20 0:14 ` Andrew Morton
2026-08-20 1:35 ` Zi Yan
2026-08-21 12:49 ` Andreas Gruenbacher [this message]
2026-08-20 4:21 ` [PATCH] gfs2: valid the length of rgrp header Edward Adam Davis
2026-08-21 12:54 ` Andreas Gruenbacher
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260821124944.1648686-1-agruenba@redhat.com \
--to=agruenba@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=eadavis@qq.com \
--cc=gfs2@lists.linux.dev \
--cc=syzbot+9d20c3ad7d29227de28d@syzkaller.appspotmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox