[PATCH] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by

[PATCH] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by

[Kees Cook]

Prepare for the coming implementation by GCC and Clang of the __counted_by
attribute. Flexible array members annotated with __counted_by can have
their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS
(for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
functions).

As found with Coccinelle[1], add __counted_by for struct v4l2_subscribed_event.
Additionally, since the element count member must be set before accessing
the annotated flexible array member, move its initialization earlier.

[1] https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci

Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
Cc: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Cc: lijian <lijian@yulong.com>
Cc: linux-media@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
---
 drivers/media/v4l2-core/v4l2-event.c | 2 +-
 include/media/v4l2-event.h           | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/media/v4l2-core/v4l2-event.c b/drivers/media/v4l2-core/v4l2-event.c
index c5ce9f11ad7b..3898ff7edddb 100644
--- a/drivers/media/v4l2-core/v4l2-event.c
+++ b/drivers/media/v4l2-core/v4l2-event.c
@@ -238,6 +238,7 @@ int v4l2_event_subscribe(struct v4l2_fh *fh,
 	sev = kvzalloc(struct_size(sev, events, elems), GFP_KERNEL);
 	if (!sev)
 		return -ENOMEM;
+	sev->elems = elems;
 	for (i = 0; i < elems; i++)
 		sev->events[i].sev = sev;
 	sev->type = sub->type;
@@ -245,7 +246,6 @@ int v4l2_event_subscribe(struct v4l2_fh *fh,
 	sev->flags = sub->flags;
 	sev->fh = fh;
 	sev->ops = ops;
-	sev->elems = elems;
 
 	mutex_lock(&fh->subscribe_lock);
 
diff --git a/include/media/v4l2-event.h b/include/media/v4l2-event.h
index 4ffa914ade3a..3a0e2588361c 100644
--- a/include/media/v4l2-event.h
+++ b/include/media/v4l2-event.h
@@ -78,7 +78,7 @@ struct v4l2_subscribed_event {
 	unsigned int		elems;
 	unsigned int		first;
 	unsigned int		in_use;
-	struct v4l2_kevent	events[];
+	struct v4l2_kevent	events[] __counted_by(elems);
 };
 
 /**
-- 
2.34.1

Re: [PATCH] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by

[Gustavo A. R. Silva]

On 9/22/23 11:49, Kees Cook wrote:
> Prepare for the coming implementation by GCC and Clang of the __counted_by
> attribute. Flexible array members annotated with __counted_by can have
> their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS
> (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
> functions).
> 
> As found with Coccinelle[1], add __counted_by for struct v4l2_subscribed_event.
> Additionally, since the element count member must be set before accessing
> the annotated flexible array member, move its initialization earlier.
> 
> [1] https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci
> 
> Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
> Cc: Hans Verkuil <hverkuil-cisco@xs4all.nl>
> Cc: lijian <lijian@yulong.com>
> Cc: linux-media@vger.kernel.org
> Signed-off-by: Kees Cook <keescook@chromium.org>

Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>

Thanks
-- 
Gustavo

> ---
>   drivers/media/v4l2-core/v4l2-event.c | 2 +-
>   include/media/v4l2-event.h           | 2 +-
>   2 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/media/v4l2-core/v4l2-event.c b/drivers/media/v4l2-core/v4l2-event.c
> index c5ce9f11ad7b..3898ff7edddb 100644
> --- a/drivers/media/v4l2-core/v4l2-event.c
> +++ b/drivers/media/v4l2-core/v4l2-event.c
> @@ -238,6 +238,7 @@ int v4l2_event_subscribe(struct v4l2_fh *fh,
>   	sev = kvzalloc(struct_size(sev, events, elems), GFP_KERNEL);
>   	if (!sev)
>   		return -ENOMEM;
> +	sev->elems = elems;
>   	for (i = 0; i < elems; i++)
>   		sev->events[i].sev = sev;
>   	sev->type = sub->type;
> @@ -245,7 +246,6 @@ int v4l2_event_subscribe(struct v4l2_fh *fh,
>   	sev->flags = sub->flags;
>   	sev->fh = fh;
>   	sev->ops = ops;
> -	sev->elems = elems;
>   
>   	mutex_lock(&fh->subscribe_lock);
>   
> diff --git a/include/media/v4l2-event.h b/include/media/v4l2-event.h
> index 4ffa914ade3a..3a0e2588361c 100644
> --- a/include/media/v4l2-event.h
> +++ b/include/media/v4l2-event.h
> @@ -78,7 +78,7 @@ struct v4l2_subscribed_event {
>   	unsigned int		elems;
>   	unsigned int		first;
>   	unsigned int		in_use;
> -	struct v4l2_kevent	events[];
> +	struct v4l2_kevent	events[] __counted_by(elems);
>   };
>   
>   /**

Re: [PATCH] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by

[Kees Cook]

On Fri, 22 Sep 2023 10:49:52 -0700, Kees Cook wrote:
> Prepare for the coming implementation by GCC and Clang of the __counted_by
> attribute. Flexible array members annotated with __counted_by can have
> their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS
> (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
> functions).
> 
> As found with Coccinelle[1], add __counted_by for struct v4l2_subscribed_event.
> Additionally, since the element count member must be set before accessing
> the annotated flexible array member, move its initialization earlier.
> 
> [...]

Since this is a trivial change and it's been 2 week without further
discussion, I'll snag this patch.

Applied to for-next/hardening, thanks!

[1/1] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by
      https://git.kernel.org/kees/c/b1a20bbd4687

Take care,

-- 
Kees Cook

Re: [PATCH] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by

[Hans Verkuil]

On 06/10/2023 22:28, Kees Cook wrote:
> On Fri, 22 Sep 2023 10:49:52 -0700, Kees Cook wrote:
>> Prepare for the coming implementation by GCC and Clang of the __counted_by
>> attribute. Flexible array members annotated with __counted_by can have
>> their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS
>> (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
>> functions).
>>
>> As found with Coccinelle[1], add __counted_by for struct v4l2_subscribed_event.
>> Additionally, since the element count member must be set before accessing
>> the annotated flexible array member, move its initialization earlier.
>>
>> [...]
> 
> Since this is a trivial change and it's been 2 week without further
> discussion, I'll snag this patch.

Ah, I picked this up for media as well. Same for the other patch.
You should have gotten an email from patchwork on Oct 3 that it was accepted.

I think I should be more pro-active and reply when I pick up patches from
non-media developers instead of relying on patchwork. It seems that it is
easy to miss emails from patchwork.

Regards,

	Hans

> 
> Applied to for-next/hardening, thanks!
> 
> [1/1] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by
>       https://git.kernel.org/kees/c/b1a20bbd4687
> 
> Take care,
> 

Re: [PATCH] media: v4l2-event: Annotate struct v4l2_subscribed_event with __counted_by

[Kees Cook]

On Sat, Oct 07, 2023 at 11:21:26AM +0200, Hans Verkuil wrote:
> On 06/10/2023 22:28, Kees Cook wrote:
> > On Fri, 22 Sep 2023 10:49:52 -0700, Kees Cook wrote:
> >> Prepare for the coming implementation by GCC and Clang of the __counted_by
> >> attribute. Flexible array members annotated with __counted_by can have
> >> their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS
> >> (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
> >> functions).
> >>
> >> As found with Coccinelle[1], add __counted_by for struct v4l2_subscribed_event.
> >> Additionally, since the element count member must be set before accessing
> >> the annotated flexible array member, move its initialization earlier.
> >>
> >> [...]
> > 
> > Since this is a trivial change and it's been 2 week without further
> > discussion, I'll snag this patch.
> 
> Ah, I picked this up for media as well. Same for the other patch.
> You should have gotten an email from patchwork on Oct 3 that it was accepted.
> 
> I think I should be more pro-active and reply when I pick up patches from
> non-media developers instead of relying on patchwork. It seems that it is
> easy to miss emails from patchwork.

Ah! Thanks for reminding me. I'll drop them from the hardening tree. :)

-Kees

-- 
Kees Cook