squashfs: squashfs_read_data need to check if the length is 0

[PATCH] squashfs: squashfs_read_data need to check if the length is 0

Posted by Lizhi Xu 2 years, 2 months ago

when the length passed in is 0, the subsequent process should be exited.

Reported-by: syzbot+32d3767580a1ea339a81@syzkaller.appspotmail.com
Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com>
---
 fs/squashfs/block.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/squashfs/block.c b/fs/squashfs/block.c
index 581ce9519339..2dc730800f44 100644
--- a/fs/squashfs/block.c
+++ b/fs/squashfs/block.c
@@ -321,7 +321,7 @@ int squashfs_read_data(struct super_block *sb, u64 index, int length,
 		TRACE("Block @ 0x%llx, %scompressed size %d\n", index - 2,
 		      compressed ? "" : "un", length);
 	}
-	if (length < 0 || length > output->length ||
+	if (length <= 0 || length > output->length ||
 			(index + length) > msblk->bytes_used) {
 		res = -EIO;
 		goto out;
-- 
2.25.1

Re: [PATCH] squashfs: squashfs_read_data need to check if the length is 0

Posted by Andrew Morton 2 years, 2 months ago

On Thu, 16 Nov 2023 11:13:52 +0800 Lizhi Xu <lizhi.xu@windriver.com> wrote:

> when the length passed in is 0, the subsequent process should be exited.

Thanks, but when fixing a bug, please always describe the runtime
effects of that bug.  Amongst other things, other people need this
information to be able to decide which kernel versions need patching.

> Reported-by: syzbot+32d3767580a1ea339a81@syzkaller.appspotmail.com

Which is a reason why we're now adding the "Closes:" tag after
Reported-by:.

I googled the sysbot email address and so added

Closes: https://lkml.kernel.org/r/0000000000000526f2060a30a085@google.com

to the changelog.

I'll assume that a -stable kernel backport is needed.

Re: [PATCH] squashfs: squashfs_read_data need to check if the length is 0

Posted by Phillip Lougher 2 years, 2 months ago

On 16/11/2023 21:43, Andrew Morton wrote:
> On Thu, 16 Nov 2023 11:13:52 +0800 Lizhi Xu <lizhi.xu@windriver.com> wrote:
> 
>> when the length passed in is 0, the subsequent process should be exited.
> 
> Thanks, but when fixing a bug, please always describe the runtime
> effects of that bug.  Amongst other things, other people need this
> information to be able to decide which kernel versions need patching.
> 
>> Reported-by: syzbot+32d3767580a1ea339a81@syzkaller.appspotmail.com
> 
> Which is a reason why we're now adding the "Closes:" tag after
> Reported-by:.

Which is also one reason why you should always run scripts/checkpatch.pl
on your patch.  This alerted me to the need for a "Closes:" tag
after Reported-by: on the last patch I sent.

> 
> I googled the sysbot email address and so added
> 
> Closes: https://lkml.kernel.org/r/0000000000000526f2060a30a085@google.com
> 
> to the changelog.

Thanks.  That is indeed the sysbot issue that the patch fixes.

> 
> I'll assume that a -stable kernel backport is needed.
> 
> 

Yes.

Phillip

Re: [PATCH] squashfs: squashfs_read_data need to check if the length is 0

Posted by Phillip Lougher 2 years, 2 months ago

> On 16/11/2023 03:13 GMT Lizhi Xu <lizhi.xu@windriver.com> wrote:
> 
>  
> when the length passed in is 0, the subsequent process should be exited.
> 

Reproduced and tested.

Reviewed-by: Phillip Lougher (phillip@squashfs.org.uk)

> Reported-by: syzbot+32d3767580a1ea339a81@syzkaller.appspotmail.com
> Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com>
> ---
>  fs/squashfs/block.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/fs/squashfs/block.c b/fs/squashfs/block.c
> index 581ce9519339..2dc730800f44 100644
> --- a/fs/squashfs/block.c
> +++ b/fs/squashfs/block.c
> @@ -321,7 +321,7 @@ int squashfs_read_data(struct super_block *sb, u64 index, int length,
>  		TRACE("Block @ 0x%llx, %scompressed size %d\n", index - 2,
>  		      compressed ? "" : "un", length);
>  	}
> -	if (length < 0 || length > output->length ||
> +	if (length <= 0 || length > output->length ||
>  			(index + length) > msblk->bytes_used) {
>  		res = -EIO;
>  		goto out;
> -- 
> 2.25.1