linux-kernel - Re: [PATCH v3] ocfs2: Invalidate inode if i

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aS5GDxVwTtefzxCI@eray-kasa>
Date: Tue, 2 Dec 2025 04:51:11 +0300
From: Ahmet Eray Karadag <eraykrdg1@...il.com>
To: Joseph Qi <joseph.qi@...ux.alibaba.com>
Cc: Heming Zhao <heming.zhao@...e.com>, mark@...heh.com, jlbec@...lplan.org,
	ocfs2-devel@...ts.linux.dev, linux-kernel@...r.kernel.org,
	david.hunter.linux@...il.com, skhan@...uxfoundation.org,
	syzbot+55c40ae8a0e5f3659f2b@...kaller.appspotmail.com,
	Albin Babu Varghese <albinbabuvarghese20@...il.com>
Subject: Re: [PATCH v3] ocfs2: Invalidate inode if i_mode is zero after block
 read

On Tue, Dec 02, 2025 at 09:25:14AM +0800, Joseph Qi wrote:
> 
> 
> On 2025/11/20 10:32, Heming Zhao wrote:
> > On Thu, Nov 20, 2025 at 01:17:24AM +0300, Ahmet Eray Karadag wrote:
> >> A panic occurs in ocfs2_unlink due to WARN_ON(inode->i_nlink == 0) when
> >> handling a corrupted inode with i_mode=0 and i_nlink=0 in memory.
> >>
> >> This "zombie" inode is created because ocfs2_read_locked_inode proceeds
> >> even after ocfs2_validate_inode_block successfully validates a block
> >> that structurally looks okay (passes checksum, signature etc.) but
> >> contains semantically invalid data (specifically i_mode=0). The current
> >> validation function doesn't check for i_mode being zero.
> >>
> >> This results in an in-memory inode with i_mode=0 being added to the VFS
> >> cache, which later triggers the panic during unlink.
> >>
> >> Prevent this by adding an explicit check for (i_mode == 0, i_nlink == 0, non-orphan)
> >> within ocfs2_validate_inode_block. If the check is true, return -EFSCORRUPTED to signal
> >> corruption. This causes the caller (ocfs2_read_locked_inode) to invoke
> >> make_bad_inode(), correctly preventing the zombie inode from entering
> >> the cache.
> >>
> >> Reported-by: syzbot+55c40ae8a0e5f3659f2b@...kaller.appspotmail.com
> >> Fixes: https://syzkaller.appspot.com/bug?extid=55c40ae8a0e5f3659f2b
> >> Co-developed-by: Albin Babu Varghese <albinbabuvarghese20@...il.com>
> >> Signed-off-by: Albin Babu Varghese <albinbabuvarghese20@...il.com>
> >> Signed-off-by: Ahmet Eray Karadag <eraykrdg1@...il.com>
> >> Previous link: https://lore.kernel.org/all/20251022222752.46758-2-eraykrdg1@gmail.com/T/
> >> ---
> >> v2:
> >>  - Only checking either i_links_count == 0 or i_mode == 0
> >>  - Not performing le16_to_cpu() anymore
> >>  - Tested with ocfs2-test
> >> ---
> >> v3:
> >>  - Add checking both high and low bits of i_links_count
> >> ---
> >>  fs/ocfs2/inode.c | 7 +++++++
> >>  1 file changed, 7 insertions(+)
> >>
> >> diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
> >> index 14bf440ea4df..c8b129db756e 100644
> >> --- a/fs/ocfs2/inode.c
> >> +++ b/fs/ocfs2/inode.c
> >> @@ -1456,6 +1456,13 @@ int ocfs2_validate_inode_block(struct super_block *sb,
> >>  		goto bail;
> >>  	}
> >>  
> >> +	if (!ocfs2_read_links_count(di) || !di->i_mode) {
> > 
> > the code logic looks good to me, but I prefer the following code to save a few
> > cpu cycles.
> > 
> > ```
> > /* only check if the "link count" or i_mode is ZERO */
> > if (!(di->i_links_count | di->i_links_count_hi) || !di->i_mode)
> > ```
> > 
> > @Joseph
> > which do you like?
> > 
> Sorry to miss this thread discuss.
> The above check looks wried since "(di->i_links_count|di->i_links_count_hi)"
> is meaningless, though it may work well...
> So I'd prefer the code readability first.
> 
> BTW, from the syzbot report, I haven't seen where to show i_mode=0 or
> i_nlink=0. Am I missing something?

You are right, the syzbot report itself doesn't explicitly dump the inode 
values. I reproduced the issue locally and debugged ocfs2_unlink() to 
inspect the inode attributes just before the crash.
The debug output confirmed that the inode had:
   i_ino = 17057
   i_nlink = 0
   i_mode = 0

It appears the inode block read succeeds (physically), but due to corruption, 
validation was insufficient, allowing a zeroed/corrupted inode struct to be 
initialized in the VFS cache.

> 
> Thanks,
> Joseph
>

Regarding the patch logic: We had previously discussed and seemed to agree 
on an approach in the thread linked below, but since I didn't get a 
final response there, I refreshed the patch here.

https://lore.kernel.org/all/20251022222752.46758-2-eraykrdg1@gmail.com/T/

Thanks,
Eray