[PATCH] btrfs: lzo: reject inline extents without both headers

[PATCH] btrfs: lzo: reject inline extents without both headers

[David Lee]

inline Btrfs LZO payload shorter than two LZO headers.

Fix the validation or lifetime rule at the vulnerable boundary so malformed
or racing input cannot reach the faulting path.

Signed-off-by: David Lee <david.lee@trailofbits.com>
Assisted-by: Codex:gpt-5.5
---
Trail of Bits has a reproducer for this bug demonstrating Kernel Panic which can be shared further if needed.

fs/btrfs/lzo.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c
--- a/fs/btrfs/lzo.c
+++ b/fs/btrfs/lzo.c
@@ -552,7 +552,8 @@ int lzo_decompress(struct list_head *ws, const u8 *data_in,
 	size_t max_segment_len = workspace_buf_length(fs_info);
 	int ret;
 
-	if (unlikely(srclen < LZO_LEN || srclen > max_segment_len + LZO_LEN * 2))
+	if (unlikely(srclen < LZO_LEN * 2 ||
+		     srclen > max_segment_len + LZO_LEN * 2))
 		return -EUCLEAN;
 
 	in_len = get_unaligned_le32(data_in);

Re: [PATCH] btrfs: lzo: reject inline extents without both headers

[Qu Wenruo]

在 2026/7/14 16:40, David Lee 写道:
> inline Btrfs LZO payload shorter than two LZO headers.
> 
> Fix the validation or lifetime rule at the vulnerable boundary so malformed
> or racing input cannot reach the faulting path.
> 
> Signed-off-by: David Lee <david.lee@trailofbits.com>
> Assisted-by: Codex:gpt-5.5
> ---
> Trail of Bits has a reproducer for this bug demonstrating Kernel Panic which can be shared further if needed.
> 
> fs/btrfs/lzo.c |    3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c
> --- a/fs/btrfs/lzo.c
> +++ b/fs/btrfs/lzo.c
> @@ -552,7 +552,8 @@ int lzo_decompress(struct list_head *ws, const u8 *data_in,
>   	size_t max_segment_len = workspace_buf_length(fs_info);
>   	int ret;
>   
> -	if (unlikely(srclen < LZO_LEN || srclen > max_segment_len + LZO_LEN * 2))
> +	if (unlikely(srclen < LZO_LEN * 2 ||

srclen == LZO_LEN * 2 is also invalid, as that means the lzo payload is 
empty.


Also your base is a little too old.

The latest for-next branch has extra error messages, thus your patch 
doesn't apply cleanly.

Please rebase to the latest for-next branch:

  https://github.com/btrfs/linux.git for-next

Otherwise the fix looks good to me.

Thanks,
Qu

> +		     srclen > max_segment_len + LZO_LEN * 2))
>   		return -EUCLEAN;
>   
>   	in_len = get_unaligned_le32(data_in);
> 
> 

Re: [PATCH] btrfs: lzo: reject inline extents without both headers

[Qu Wenruo]

在 2026/7/14 18:08, Qu Wenruo 写道:
> 
> 
> 在 2026/7/14 16:40, David Lee 写道:
>> inline Btrfs LZO payload shorter than two LZO headers.
>>
>> Fix the validation or lifetime rule at the vulnerable boundary so 
>> malformed
>> or racing input cannot reach the faulting path.
>>
>> Signed-off-by: David Lee <david.lee@trailofbits.com>
>> Assisted-by: Codex:gpt-5.5
>> ---
>> Trail of Bits has a reproducer for this bug demonstrating Kernel Panic 
>> which can be shared further if needed.

Also forgot to mention, it's always better to provide the full calltrace.

It will benefit people who hit a similar crash to determine if it's 
already fixed.
>>
>> fs/btrfs/lzo.c |    3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c
>> --- a/fs/btrfs/lzo.c
>> +++ b/fs/btrfs/lzo.c
>> @@ -552,7 +552,8 @@ int lzo_decompress(struct list_head *ws, const u8 
>> *data_in,
>>       size_t max_segment_len = workspace_buf_length(fs_info);
>>       int ret;
>> -    if (unlikely(srclen < LZO_LEN || srclen > max_segment_len + 
>> LZO_LEN * 2))
>> +    if (unlikely(srclen < LZO_LEN * 2 ||
> 
> srclen == LZO_LEN * 2 is also invalid, as that means the lzo payload is 
> empty.
> 
> 
> Also your base is a little too old.
> 
> The latest for-next branch has extra error messages, thus your patch 
> doesn't apply cleanly.
> 
> Please rebase to the latest for-next branch:
> 
>   https://github.com/btrfs/linux.git for-next
> 
> Otherwise the fix looks good to me.
> 
> Thanks,
> Qu
> 
>> +             srclen > max_segment_len + LZO_LEN * 2))
>>           return -EUCLEAN;
>>       in_len = get_unaligned_le32(data_in);
>>
>>
> 
>