All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Brian Daniels" <briandaniels@google.com>
Cc: virtualization@lists.linux.dev, media-ci@linuxtv.org,
	"Michael S. Tsirkin" <mst@redhat.com>,
	"Eugenio Perez" <eperezma@redhat.com>
Subject: Re: [PATCH v12 6/6] media: virtio: clear the size of legacy controls built on the stack
Date: Sat, 21 Nov 2026 03:17:28 +0000	[thread overview]
Message-ID: <sashiko-outbox-226567@kernel.org> (raw)
In-Reply-To: <20261121025007.3445723-7-briandaniels@google.com>

Thank you for your contribution! Sashiko AI review found 3 potential
issues to consider:
- [High] media: virtio: framevecs array leak in
  scatterlist_builder_free_framevecs()
- [Medium] media: virtio: incomplete stack control init in
  virtio_media_legacy_ctrl()
- [Medium] media: virtio: framevec and pinned page leaks in
  virtio_media_send_ext_controls_ioctl()
--

commit 623bf5e3612ed0e69ea0cdb41218ea9e8d7b62ef
Author: Nick Rogers <nick@getfieldwork.ai>

media: virtio: clear the size of legacy controls built on the stack

This commit adds a helper that checks if the extended control array is on
the kernel stack and zeroes each control's size field before forwarding
extended control ioctls to the VirtIO host.

> diff --git a/drivers/media/virtio/virtio_media_ioctls.c b/drivers/media/virtio/virtio_media_ioctls.c
> index aa56efe9a08c2..ad240073aec9f 100644
> --- a/drivers/media/virtio/virtio_media_ioctls.c
> +++ b/drivers/media/virtio/virtio_media_ioctls.c
[ ... ]
> @@ -593,11 +594,31 @@ static int virtio_media_querycap(struct file *file, void *fh,
>   * Extended control ioctls are handled mostly identically.
>   */
>  
> +/*
> + * VIDIOC_G_CTRL and VIDIOC_S_CTRL reach a driver without a control handler
> + * as one extended control the V4L2 core builds on its own stack, with its
> + * size left uninitialized. They carry a 32-bit value and never a payload,
> + * so a size there is stack garbage, which would be sent as a user pointer
> + * to copy and fail the ioctl. An array from userspace is never on the
> + * stack: the core copies it into an allocation.
> + */
> +static void virtio_media_legacy_ctrl(struct v4l2_ext_controls *ctrls)
> +{
> +	u32 i;
> +
> +	if (!ctrls->count || !object_is_on_stack(ctrls->controls))
> +		return;
> +	for (i = 0; i < ctrls->count; i++)
> +		ctrls->controls[i].size = 0;
> +}

[Severity: Medium]
Does clearing only ctrls->controls[i].size still leave uninitialized
kernel stack fields that get sent across the virtqueue to the host?

When v4l_g_ctrl() and v4l_s_ctrl() build struct v4l2_ext_controls and
struct v4l2_ext_control on the stack without zero-initializing them,
ctrls->error_idx, ctrls->request_fd, struct alignment padding, and the
upper 32 bits of the 8-byte value64 union in ctrl remain uninitialized.
scatterlist_builder_add_ext_ctrls() transmits the entire sizeof(*ctrls)
and sizeof(ctrls->controls[0]) * ctrls->count buffers to the host,
leaking guest kernel stack bytes and passing stack garbage in request_fd,
error_idx, and the upper 32 bits of value64.

Also, detecting legacy controls in the driver with object_is_on_stack()
(including in virtio_media_try_ext_ctrls(), which VIDIOC_G_CTRL and
VIDIOC_S_CTRL never call) works around v4l_g_ctrl() and v4l_s_ctrl() in
drivers/media/v4l2-core/v4l2-ioctl.c rather than zero-initializing the
stack structures there.

> +
>  static int virtio_media_g_ext_ctrls(struct file *file, void *fh,
>  				    struct v4l2_ext_controls *ctrls)
>  {
>  	struct v4l2_fh *vfh = file_to_v4l2_fh(file);
>  
> +	virtio_media_legacy_ctrl(ctrls);
> +
>  	return virtio_media_send_ext_controls_ioctl(vfh, VIDIOC_G_EXT_CTRLS,
>  						    ctrls);
>  }

[Severity: Medium]
In virtio_media_send_ext_controls_ioctl() (introduced earlier in this
series in commit "media: virtio: Add extended controls support"), do the
two post-command validation checks leak builder.framevecs and leave user
pages permanently pinned?

drivers/media/virtio/virtio_media_ioctls.c:virtio_media_send_ext_controls_ioctl() {
    ...
	if (ctrls->count > num_ctrls)
		return -ENOSPC;
	if (ret < 0) {
    ...
		goto free_framevecs;
	}

	resp_len -= sizeof(struct virtio_media_resp_ioctl);

	/* Make sure that the reply's length covers our v4l2_ext_controls */
	if (resp_len < sizeof(*ctrls))
		return -EINVAL;
    ...
free_framevecs:
	scatterlist_builder_free_framevecs(&builder);
	return ret;
}

When a request includes controls with ctrl->size > 0, returning -ENOSPC
or -EINVAL directly after virtio_media_send_command() skips
free_framevecs, leaking builder.framevecs and never calling
vb2_destroy_framevec() on the pinned user buffers.

>  
>  static int virtio_media_s_ext_ctrls(struct file *file, void *fh,
>  				    struct v4l2_ext_controls *ctrls)
>  {
>  	struct v4l2_fh *vfh = file_to_v4l2_fh(file);
>  
> +	virtio_media_legacy_ctrl(ctrls);
> +
>  	return virtio_media_send_ext_controls_ioctl(vfh, VIDIOC_S_EXT_CTRLS,
>  						    ctrls);
>  }

[Severity: High]
Also in the free_framevecs path of virtio_media_send_ext_controls_ioctl()
(from commit "media: virtio: Add extended controls support"), does
scatterlist_builder_free_framevecs() leak builder->framevecs when an
error occurs before the first framevec is created?

drivers/media/virtio/scatterlist_builder.c:scatterlist_builder_free_framevecs() {
	if (builder->cur_framevec && builder->framevecs) {
		struct frame_vector *vec;

		while (builder->cur_framevec--) {
			vec = builder->framevecs[builder->cur_framevec];
			vb2_destroy_framevec(vec, true);
		}

		kfree(builder->framevecs);
	}

	builder->framevecs = NULL;
    ...
}

scatterlist_builder_allocate_framevecs() allocates builder->framevecs
whenever any control has ctrl->size > 0, while builder->cur_framevec is
still 0.  If an unprivileged user passes an invalid ctrl->ptr on the
first payload control, vb2_create_framevec() fails with cur_framevec == 0
and scatterlist_builder_free_framevecs() skips kfree(builder->framevecs)
before clearing the pointer.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261121025007.3445723-1-briandaniels@google.com?part=6

      reply	other threads:[~2026-11-21  3:17 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-21  2:50 [PATCH v12 0/6] media: add virtio-media driver Brian Daniels
2026-11-21  2:50 ` [PATCH v12 1/6] media: virtio: Add core driver and session management Brian Daniels
2026-11-21  3:13   ` sashiko-bot
2026-11-21  2:50 ` [PATCH v12 2/6] media: virtio: Add V4L2 ioctl forwarding infrastructure Brian Daniels
2026-11-21  3:15   ` sashiko-bot
2026-11-21  2:50 ` [PATCH v12 3/6] media: virtio: Add extended controls support Brian Daniels
2026-11-21  3:14   ` sashiko-bot
2026-11-21  2:50 ` [PATCH v12 4/6] media: virtio: Add buffer streaming, event handling, and poll Brian Daniels
2026-11-21  3:20   ` sashiko-bot
2026-11-21  2:50 ` [PATCH v12 5/6] media: virtio: Add MMAP buffer mapping support Brian Daniels
2026-11-21  3:07   ` sashiko-bot
2026-11-21  2:50 ` [PATCH v12 6/6] media: virtio: clear the size of legacy controls built on the stack Brian Daniels
2026-11-21  3:17   ` sashiko-bot [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-226567@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=briandaniels@google.com \
    --cc=eperezma@redhat.com \
    --cc=media-ci@linuxtv.org \
    --cc=mst@redhat.com \
    --cc=sashiko-reviews@lists.linux.dev \
    --cc=virtualization@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.