linux-kernel - Re: [PATCH] ocfs2: Avoid NULL pointer dereference in dx_dir_lookup

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: <d6a11613-9763-4d9a-b4ad-5f451f770b70@suse.com>
Date: Mon, 30 Jun 2025 10:32:35 +0800
From: Heming Zhao <heming.zhao@...e.com>
To: Joseph Qi <joseph.qi@...ux.alibaba.com>,
 Ivan Pravdin <ipravdin.official@...il.com>, mark@...heh.com,
 jlbec@...lplan.org, ocfs2-devel@...ts.linux.dev, linux-kernel@...r.kernel.org
Cc: syzbot+20282c1b2184a857ac4c@...kaller.appspotmail.com
Subject: Re: [PATCH] ocfs2: Avoid NULL pointer dereference in
 dx_dir_lookup_rec()

On 6/30/25 09:26, Joseph Qi wrote:
> Hi,
> 
> 
> On 2025/6/27 10:38, Ivan Pravdin wrote:
>> When a directory entry is not found, ocfs2_dx_dir_lookup_rec() prints an
>> error message that unconditionally dereferences the 'rec' pointer.
>> However, if 'rec' is NULL, this leads to a NULL pointer dereference and
>> a kernel panic.
>>
> 
> This looks possible, but syzbot reports slab-out-of-bounds Read in
> ocfs2_dx_dir_lookup_rec(), not NULL pointer dereference.
> 
> So I think it is because it construct a malicious image and set a wrong
> l_recs, then access this damaged l_recs.
> 
> Thanks,
> Joseph

I think this proposed fix (at least the fix method) is acceptable.
the crash occurs at ocfs2_error(), where the pointer 'rec' must be incorrect.
look back at the previous code lines:
         for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
                 rec = &el->l_recs[i];

                 if (le32_to_cpu(rec->e_cpos) <= major_hash) {
                         found = 1;
                         break;
                 }
         }

either 'el->l_next_free_rec' or 'el->l_recs[i]' has an incorrect value.
we can do nothing about this kind of error, simply returning errno to caller is sufficient.

btw, ocfs2-tools has a similar function "static errcode_t ocfs2_dx_dir_lookup_rec()"@libocfs2/dir_indexed.c, which sets ret with OCFS2_ET_CORRUPT_EXTENT_BLOCK and return.

- Heming

> 
> 
>> Add an explicit check for a NULL 'rec' and use an alternate error
>> message in that case to avoid unsafe access.
>>
>> Reported-by: syzbot+20282c1b2184a857ac4c@...kaller.appspotmail.com
>> Closes: https://lore.kernel.org/all/67483b75.050a0220.253251.007c.GAE@google.com/T/
>> Signed-off-by: Ivan Pravdin <ipravdin.official@...il.com>
>> ---
>>   fs/ocfs2/dir.c | 16 +++++++++++-----
>>   1 file changed, 11 insertions(+), 5 deletions(-)
>>
>> diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c
>> index 7799f4d16ce9..dccf0349e523 100644
>> --- a/fs/ocfs2/dir.c
>> +++ b/fs/ocfs2/dir.c
>> @@ -809,11 +809,17 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
>>   	}
>>   
>>   	if (!found) {
>> -		ret = ocfs2_error(inode->i_sb,
>> -				  "Inode %lu has bad extent record (%u, %u, 0) in btree\n",
>> -				  inode->i_ino,
>> -				  le32_to_cpu(rec->e_cpos),
>> -				  ocfs2_rec_clusters(el, rec));
>> +		if (rec) {
>> +			ret = ocfs2_error(inode->i_sb,
>> +					"Inode %lu has bad extent record (%u, %u, 0) in btree\n",
>> +					inode->i_ino,
>> +					le32_to_cpu(rec->e_cpos),
>> +					ocfs2_rec_clusters(el, rec));
>> +		} else {
>> +			ret = ocfs2_error(inode->i_sb,
>> +					"Inode %lu has no extent records in btree\n",
>> +					inode->i_ino);
>> +		}
>>   		goto out;
>>   	}
>>   
> 
>