[PATCH] jffs2: add additinal sanity check for jffs2_acl_from_medium()

[PATCH] jffs2: add additinal sanity check for jffs2_acl_from_medium()

[Chengguang Xu]

In the case ACL_USER and ACL_GROUP we check if value has exceeded end,
add same check in the case ACL_OTHER as well.

Signed-off-by: Chengguang Xu <cgxu519@gmx.com>
---
 fs/jffs2/acl.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/jffs2/acl.c b/fs/jffs2/acl.c
index 093ffbd82395..fce32de3a18f 100644
--- a/fs/jffs2/acl.c
+++ b/fs/jffs2/acl.c
@@ -94,6 +94,8 @@ static struct posix_acl *jffs2_acl_from_medium(void *value, size_t size)
 			case ACL_MASK:
 			case ACL_OTHER:
 				value += sizeof(struct jffs2_acl_entry_short);
+				if (value > end)
+					goto fail;
 				break;
 
 			case ACL_USER:
-- 
2.17.1

Re: [PATCH] jffs2: add additinal sanity check for jffs2_acl_from_medium()

[Richard Weinberger]

On Sun, Sep 2, 2018 at 5:45 PM Chengguang Xu <cgxu519@gmx.com> wrote:
>
> In the case ACL_USER and ACL_GROUP we check if value has exceeded end,
> add same check in the case ACL_OTHER as well.

Did you hit a problem in that area or was this found by review?
>From looking at the code I'd say it is fine as is.
In the ACL_MASK/_OTHER case we don't look into the entry object like
ACL_USER/_GROUP
do, we immediately break the switch and run another round in the for loop.
And here we do:
                entry = value;
                if (value + sizeof(struct jffs2_acl_entry_short) > end)
                        goto fail;

Which is what your additional check does. So, we'd check twice.
What do I miss?

> Signed-off-by: Chengguang Xu <cgxu519@gmx.com>
> ---
>  fs/jffs2/acl.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/fs/jffs2/acl.c b/fs/jffs2/acl.c
> index 093ffbd82395..fce32de3a18f 100644
> --- a/fs/jffs2/acl.c
> +++ b/fs/jffs2/acl.c
> @@ -94,6 +94,8 @@ static struct posix_acl *jffs2_acl_from_medium(void *value, size_t size)
>                         case ACL_MASK:
>                         case ACL_OTHER:
>                                 value += sizeof(struct jffs2_acl_entry_short);
> +                               if (value > end)
> +                                       goto fail;
>                                 break;
>
>                         case ACL_USER:
> --
> 2.17.1
>
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/



-- 
Thanks,
//richard

Re: [PATCH] jffs2: add additinal sanity check for jffs2_acl_from_medium()

[cgxu519]

On 09/03/2018 04:45 PM, Richard Weinberger wrote:
> On Sun, Sep 2, 2018 at 5:45 PM Chengguang Xu <cgxu519@gmx.com> wrote:
>> In the case ACL_USER and ACL_GROUP we check if value has exceeded end,
>> add same check in the case ACL_OTHER as well.
> Did you hit a problem in that area or was this found by review?
>  From looking at the code I'd say it is fine as is.
> In the ACL_MASK/_OTHER case we don't look into the entry object like
> ACL_USER/_GROUP
> do, we immediately break the switch and run another round in the for loop.
> And here we do:
>                  entry = value;
>                  if (value + sizeof(struct jffs2_acl_entry_short) > end)
>                          goto fail;
>
> Which is what your additional check does. So, we'd check twice.
> What do I miss?

You are right, it is actually not needed. Sorry, please just drop the patch.

Re: [PATCH] jffs2: add additinal sanity check for jffs2_acl_from_medium()

[Richard Weinberger]

Am Montag, 3. September 2018, 15:28:31 CEST schrieb cgxu519:
> On 09/03/2018 04:45 PM, Richard Weinberger wrote:
> > On Sun, Sep 2, 2018 at 5:45 PM Chengguang Xu <cgxu519@gmx.com> wrote:
> >> In the case ACL_USER and ACL_GROUP we check if value has exceeded end,
> >> add same check in the case ACL_OTHER as well.
> > Did you hit a problem in that area or was this found by review?
> >  From looking at the code I'd say it is fine as is.
> > In the ACL_MASK/_OTHER case we don't look into the entry object like
> > ACL_USER/_GROUP
> > do, we immediately break the switch and run another round in the for loop.
> > And here we do:
> >                  entry = value;
> >                  if (value + sizeof(struct jffs2_acl_entry_short) > end)
> >                          goto fail;
> >
> > Which is what your additional check does. So, we'd check twice.
> > What do I miss?
> 
> You are right, it is actually not needed. Sorry, please just drop the patch.

No problem, that's why we have a review process. :-)

Thanks,
//richard