2023-10-02 14:32:39

by Juntong Deng

Subject: [PATCH v2] fs/jfs: Add check for negative db_l2nbperpage

l2nbperpage is log2(number of blks per page), and the minimum legal
value should be 0, not negative.

In the case of l2nbperpage being negative, an error will occur
when subsequently used as shift exponent.

Syzbot reported this bug:

UBSAN: shift-out-of-bounds in fs/jfs/jfs_dmap.c:799:12
shift exponent -16777216 is negative

Reported-by: [email protected]
Closes: https://syzkaller.appspot.com/bug?extid=debee9ab7ae2b34b0307
Signed-off-by: Juntong Deng <[email protected]>
---
fs/jfs/jfs_dmap.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/jfs/jfs_dmap.c b/fs/jfs/jfs_dmap.c
index 88afd108c2dd..3a1842348112 100644
--- a/fs/jfs/jfs_dmap.c
+++ b/fs/jfs/jfs_dmap.c
@@ -180,7 +180,8 @@ int dbMount(struct inode *ipbmap)
bmp->db_nfree = le64_to_cpu(dbmp_le->dn_nfree);

bmp->db_l2nbperpage = le32_to_cpu(dbmp_le->dn_l2nbperpage);
- if (bmp->db_l2nbperpage > L2PSIZE - L2MINBLOCKSIZE) {
+ if (bmp->db_l2nbperpage > L2PSIZE - L2MINBLOCKSIZE ||
+ bmp->db_l2nbperpage < 0) {
err = -EINVAL;
goto err_release_metapage;
}
--
2.39.2

2023-10-02 16:09:24

by Ricardo B. Marliere

Subject: Re: [PATCH v2] fs/jfs: Add check for negative db_l2nbperpage

On 23/10/02 05:56PM, Juntong Deng wrote:
> Signed-off-by: Juntong Deng <[email protected]>
> ---
> fs/jfs/jfs_dmap.c | 3 ++-

Hey there, looks like you forgot to add a changelog for v1->v2? I see
you just changed the commit tags ordering but you should have that in
mind for next time :)

- Ricardo

2023-10-02 22:25:58

by Juntong Deng

Subject: Re: [PATCH v2] fs/jfs: Add check for negative db_l2nbperpage

On 2023/10/2 23:56, Ricardo B. Marliere wrote:
> On 23/10/02 05:56PM, Juntong Deng wrote:
>> Signed-off-by: Juntong Deng <[email protected]>
>> ---
>> fs/jfs/jfs_dmap.c | 3 ++-
>
> Hey there, looks like you forgot to add a changelog for v1->v2? I see
> you just changed the commit tags ordering but you should have that in
> mind for next time :)
>
> - Ricardo

Hi, Ricardo.

I just made a mistake in the order of the tags, I should have
put Signed-off-by last, I did not change any code so I did not
add the changelog.

I am not sure if the changelog should be added if the code of
the patch is not changed.

Thanks.

2023-10-03 23:28:35

by Dave Kleikamp

Subject: Re: [PATCH v2] fs/jfs: Add check for negative db_l2nbperpage

On 10/2/23 4:56AM, Juntong Deng wrote:
> l2nbperpage is log2(number of blks per page), and the minimum legal
> value should be 0, not negative.
>
> In the case of l2nbperpage being negative, an error will occur
> when subsequently used as shift exponent.
>
> Syzbot reported this bug:
>
> UBSAN: shift-out-of-bounds in fs/jfs/jfs_dmap.c:799:12
> shift exponent -16777216 is negative

Looks good. (A little part of me wants to go through the code and make a
lot of signed ints unsigned, but not today.)

Applied to jfs-next

Thanks,
Shaggy

>
> Reported-by: [email protected]
> Closes: https://syzkaller.appspot.com/bug?extid=debee9ab7ae2b34b0307
> Signed-off-by: Juntong Deng <[email protected]>
> ---
> fs/jfs/jfs_dmap.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/fs/jfs/jfs_dmap.c b/fs/jfs/jfs_dmap.c
> index 88afd108c2dd..3a1842348112 100644
> --- a/fs/jfs/jfs_dmap.c
> +++ b/fs/jfs/jfs_dmap.c
> @@ -180,7 +180,8 @@ int dbMount(struct inode *ipbmap)
> bmp->db_nfree = le64_to_cpu(dbmp_le->dn_nfree);
>
> bmp->db_l2nbperpage = le32_to_cpu(dbmp_le->dn_l2nbperpage);
> - if (bmp->db_l2nbperpage > L2PSIZE - L2MINBLOCKSIZE) {
> + if (bmp->db_l2nbperpage > L2PSIZE - L2MINBLOCKSIZE ||
> + bmp->db_l2nbperpage < 0) {
> err = -EINVAL;
> goto err_release_metapage;
> }