mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH] media: vivid: check for NULL dev in vivid_probe() control setup
@ 2026-09-22 12:54 Rohinthan
  2026-09-23  8:35 ` Hans Verkuil
  0 siblings, 1 reply; 2+ messages in thread
From: Rohinthan @ 2026-09-22 12:54 UTC (permalink / raw)
  To: Hans Verkuil, Mauro Carvalho Chehab
  Cc: linux-media, linux-kernel, syzbot+caaca441bae2799984d3, Rohinthan

In vivid_probe(), after creating instances, the driver iterates
through vivid_devs[0..n_devs-1] to populate HDMI and S-Video
'Connected To' controls. However, it dereferences dev->has_vid_out
without verifying that dev is non-NULL.

If an instance fails during creation (or if devices race during
probe/remove), vivid_devs[i] remains NULL, causing a general
protection fault:

    Oops: general protection fault
    KASAN: probably user-memory-access in range
    [0x0000000000005f38-0x0000000000005f3f]
    RIP: 0010:vivid_probe+0x579/0x10d0

Add NULL pointer checks in both loops in vivid_probe() before accessing
vivid_devs[i]. Also set .suppress_bind_attrs = true on
vivid_pdrv.driver to prevent concurrent sysfs bind/unbind operations
from triggering races.

Fixes: d7c969f37515 ("media: vivid: Add 'Is Connected To' menu controls")
Reported-by: syzbot+caaca441bae2799984d3@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=caaca441bae2799984d3
Signed-off-by: Rohinthan <rokinthanp03@gmail.com>
---
 drivers/media/test-drivers/vivid/vivid-core.c | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/drivers/media/test-drivers/vivid/vivid-core.c b/drivers/media/test-drivers/vivid/vivid-core.c
index 62cfb5f..31a90ec 100644
--- a/drivers/media/test-drivers/vivid/vivid-core.c
+++ b/drivers/media/test-drivers/vivid/vivid-core.c
@@ -2116,17 +2116,19 @@ static int vivid_probe(struct platform_device *pdev)
 	for (int i = 0; i < n_devs; i++) {
 		struct vivid_dev *dev = vivid_devs[i];
 
+		if (!dev)
+			continue;
 		if (!dev->has_vid_out)
 			continue;
 		for (int j = 0; j < dev->num_outputs && hdmi_count < MAX_MENU_ITEMS; ++j) {
 			if (dev->output_type[j] == HDMI) {
-				vivid_ctrl_hdmi_to_output_instance[hdmi_count] = vivid_devs[i];
+				vivid_ctrl_hdmi_to_output_instance[hdmi_count] = dev;
 				vivid_ctrl_hdmi_to_output_index[hdmi_count++] = j;
 			}
 		}
 		for (int j = 0; j < dev->num_outputs && svid_count < MAX_MENU_ITEMS; ++j) {
 			if (dev->output_type[j] == SVID) {
-				vivid_ctrl_svid_to_output_instance[svid_count] = vivid_devs[i];
+				vivid_ctrl_svid_to_output_instance[svid_count] = dev;
 				vivid_ctrl_svid_to_output_index[svid_count++] = j;
 			}
 		}
@@ -2134,13 +2136,17 @@ static int vivid_probe(struct platform_device *pdev)
 	hdmi_count = min(hdmi_count, MAX_MENU_ITEMS);
 	svid_count = min(svid_count, MAX_MENU_ITEMS);
 	for (int i = 0; i < n_devs; i++) {
-		for (int j = 0; j < vivid_devs[i]->num_hdmi_inputs; j++) {
-			struct v4l2_ctrl *c = vivid_devs[i]->ctrl_hdmi_to_output[j];
+		struct vivid_dev *dev = vivid_devs[i];
+
+		if (!dev)
+			continue;
+		for (int j = 0; j < dev->num_hdmi_inputs; j++) {
+			struct v4l2_ctrl *c = dev->ctrl_hdmi_to_output[j];
 
 			v4l2_ctrl_modify_range(c, c->minimum, hdmi_count - 1, 0, c->default_value);
 		}
-		for (int j = 0; j < vivid_devs[i]->num_svid_inputs; j++) {
-			struct v4l2_ctrl *c = vivid_devs[i]->ctrl_svid_to_output[j];
+		for (int j = 0; j < dev->num_svid_inputs; j++) {
+			struct v4l2_ctrl *c = dev->ctrl_svid_to_output[j];
 
 			v4l2_ctrl_modify_range(c, c->minimum, svid_count - 1, 0, c->default_value);
 		}
@@ -2240,6 +2246,7 @@ static struct platform_driver vivid_pdrv = {
 	.remove		= vivid_remove,
 	.driver		= {
 		.name	= "vivid",
+		.suppress_bind_attrs = true,
 	},
 };
 
-- 
2.53.0


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] media: vivid: check for NULL dev in vivid_probe() control setup
  2026-09-22 12:54 [PATCH] media: vivid: check for NULL dev in vivid_probe() control setup Rohinthan
@ 2026-09-23  8:35 ` Hans Verkuil
  0 siblings, 0 replies; 2+ messages in thread
From: Hans Verkuil @ 2026-09-23  8:35 UTC (permalink / raw)
  To: Rohinthan, Mauro Carvalho Chehab
  Cc: linux-media, linux-kernel, syzbot+caaca441bae2799984d3

On 22/09/2026 14:54, Rohinthan wrote:
> In vivid_probe(), after creating instances, the driver iterates
> through vivid_devs[0..n_devs-1] to populate HDMI and S-Video
> 'Connected To' controls. However, it dereferences dev->has_vid_out
> without verifying that dev is non-NULL.
> 
> If an instance fails during creation (or if devices race during
> probe/remove), vivid_devs[i] remains NULL, causing a general
> protection fault:
> 
>     Oops: general protection fault
>     KASAN: probably user-memory-access in range
>     [0x0000000000005f38-0x0000000000005f3f]
>     RIP: 0010:vivid_probe+0x579/0x10d0
> 
> Add NULL pointer checks in both loops in vivid_probe() before accessing
> vivid_devs[i]. Also set .suppress_bind_attrs = true on
> vivid_pdrv.driver to prevent concurrent sysfs bind/unbind operations
> from triggering races.
> 
> Fixes: d7c969f37515 ("media: vivid: Add 'Is Connected To' menu controls")
> Reported-by: syzbot+caaca441bae2799984d3@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=caaca441bae2799984d3
> Signed-off-by: Rohinthan <rokinthanp03@gmail.com>
> ---
>  drivers/media/test-drivers/vivid/vivid-core.c | 19 +++++++++++++------
>  1 file changed, 13 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/media/test-drivers/vivid/vivid-core.c b/drivers/media/test-drivers/vivid/vivid-core.c
> index 62cfb5f..31a90ec 100644
> --- a/drivers/media/test-drivers/vivid/vivid-core.c
> +++ b/drivers/media/test-drivers/vivid/vivid-core.c
> @@ -2116,17 +2116,19 @@ static int vivid_probe(struct platform_device *pdev)
>  	for (int i = 0; i < n_devs; i++) {
>  		struct vivid_dev *dev = vivid_devs[i];
>  
> +		if (!dev)
> +			continue;

This can never happen, vivid_devsis always not-NULL for indices up to n_devs - 1.

If there is a crash, then it is caused by something else.

>  		if (!dev->has_vid_out)
>  			continue;
>  		for (int j = 0; j < dev->num_outputs && hdmi_count < MAX_MENU_ITEMS; ++j) {
>  			if (dev->output_type[j] == HDMI) {
> -				vivid_ctrl_hdmi_to_output_instance[hdmi_count] = vivid_devs[i];
> +				vivid_ctrl_hdmi_to_output_instance[hdmi_count] = dev;
>  				vivid_ctrl_hdmi_to_output_index[hdmi_count++] = j;
>  			}
>  		}
>  		for (int j = 0; j < dev->num_outputs && svid_count < MAX_MENU_ITEMS; ++j) {
>  			if (dev->output_type[j] == SVID) {
> -				vivid_ctrl_svid_to_output_instance[svid_count] = vivid_devs[i];
> +				vivid_ctrl_svid_to_output_instance[svid_count] = dev;
>  				vivid_ctrl_svid_to_output_index[svid_count++] = j;
>  			}
>  		}
> @@ -2134,13 +2136,17 @@ static int vivid_probe(struct platform_device *pdev)
>  	hdmi_count = min(hdmi_count, MAX_MENU_ITEMS);
>  	svid_count = min(svid_count, MAX_MENU_ITEMS);
>  	for (int i = 0; i < n_devs; i++) {
> -		for (int j = 0; j < vivid_devs[i]->num_hdmi_inputs; j++) {
> -			struct v4l2_ctrl *c = vivid_devs[i]->ctrl_hdmi_to_output[j];
> +		struct vivid_dev *dev = vivid_devs[i];
> +
> +		if (!dev)
> +			continue;
> +		for (int j = 0; j < dev->num_hdmi_inputs; j++) {
> +			struct v4l2_ctrl *c = dev->ctrl_hdmi_to_output[j];
>  
>  			v4l2_ctrl_modify_range(c, c->minimum, hdmi_count - 1, 0, c->default_value);
>  		}
> -		for (int j = 0; j < vivid_devs[i]->num_svid_inputs; j++) {
> -			struct v4l2_ctrl *c = vivid_devs[i]->ctrl_svid_to_output[j];
> +		for (int j = 0; j < dev->num_svid_inputs; j++) {
> +			struct v4l2_ctrl *c = dev->ctrl_svid_to_output[j];
>  
>  			v4l2_ctrl_modify_range(c, c->minimum, svid_count - 1, 0, c->default_value);
>  		}
> @@ -2240,6 +2246,7 @@ static struct platform_driver vivid_pdrv = {
>  	.remove		= vivid_remove,
>  	.driver		= {
>  		.name	= "vivid",
> +		.suppress_bind_attrs = true,

Nack. We WANT vivid to be unbound. It is part of the regression tests.

>  	},
>  };
>  

Rejecting this patch.

Regards,

	Hans

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-09-23  8:35 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-09-22 12:54 [PATCH] media: vivid: check for NULL dev in vivid_probe() control setup Rohinthan
2026-09-23  8:35 ` Hans Verkuil

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®