Skip to contentSkip to conversation

2.6.27.y: fix NULL ptr deref in cx23885 video_open

PATCH New Jarod Wilson 5 participants 8 messages Feb 24, 2009 – Mar 13, 2009 Media (V4L/DVB)

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 2.6.27.y: fix NULL ptr deref in cx23885 video_open

From: Mark Jenks
https://www.redhat.com/mailman/private/video4linux-list/2009-January/msg00041.html

The Hauppauge WinTV HVR-1800 tv tuner card has both digital and analog abilities, both of which are supported by v4l/dvb under 2.6.27.y. The analog side also features a hardware mpeg2 encoder. The HVR-1250 tv tuner card has both digital and analog abilities, but analog isn't currently supported under any kernel. These cards both utilize the cx23885 driver, but with slightly different usage. When the code paths for each card is executed, they wind up poking a cx23885_devlist, which contains devices from both of the cards, and access attempts are made to portions of 'struct cx23885_dev' that aren't valid for that device. Simply add some extra checks before trying to access these structs.

More gory details:
http://article.gmane.org/gmane.linux.drivers.dvb/46630

This was triggering on my own system at home w/both cards in it, and no longer happens with this patch included.

---

 drivers/media/video/cx23885/cx23885-417.c   |    2 +-
 drivers/media/video/cx23885/cx23885-video.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/media/video/cx23885/cx23885-417.c +1 -1
index 7b0e8c0..19154b6 100644
@@ -1585,7 +1585,7 @@ static int mpeg_open(struct inode *inode, struct file *file)
 
 	list_for_each(list, &cx23885_devlist) {
 		h = list_entry(list, struct cx23885_dev, devlist);
-		if (h->v4l_device->minor == minor) {
+		if (h->v4l_device && h->v4l_device->minor == minor) {
 			dev = h;
 			break;
 		}
drivers/media/video/cx23885/cx23885-video.c +1 -1
index 6047c78..a2b5a0c 100644
@@ -733,7 +733,7 @@ static int video_open(struct inode *inode, struct file *file)
 
 	list_for_each(list, &cx23885_devlist) {
 		h = list_entry(list, struct cx23885_dev, devlist);
-		if (h->video_dev->minor == minor) {
+		if (h->video_dev && h->video_dev->minor == minor) {
 			dev  = h;
 			type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
 		}
Signature · 2 lines
Jarod Wilson
[email protected]
#1 lore raw

PATCH 1/1 2.6.27.y: fix NULL ptr deref in cx23885 video_open

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

---------- Forwarded message ----------
From: Jarod Wilson <[email protected]>
Date: Tue, Feb 24, 2009 at 6:00 PM
Subject: [stable] [PATCH] 2.6.27.y: fix NULL ptr deref in cx23885 video_open
To: [email protected]
Cc: Mike Krufky <[email protected]>

From: Mark Jenks
https://www.redhat.com/mailman/private/video4linux-list/2009-January/msg00041.html

The Hauppauge WinTV HVR-1800 tv tuner card has both digital and analog abilities, both of which are supported by v4l/dvb under 2.6.27.y. The analog side also features a hardware mpeg2 encoder. The HVR-1250 tv tuner card has both digital and analog abilities, but analog isn't currently supported under any kernel. These cards both utilize the cx23885 driver, but with slightly different usage. When the code paths for each card is executed, they wind up poking a cx23885_devlist, which contains devices from both of the cards, and access attempts are made to portions of 'struct cx23885_dev' that aren't valid for that device. Simply add some extra checks before trying to access these structs.

More gory details:
 http://article.gmane.org/gmane.linux.drivers.dvb/46630

This was triggering on my own system at home w/both cards in it, and no longer happens with this patch included.

---

 drivers/media/video/cx23885/cx23885-417.c   |    2 +- drivers/media/video/cx23885/cx23885-video.c |    2 +-

 2 files changed, 2 insertions(+), 2 deletions(-)
(quoted diff) +0 -0

b/drivers/media/video/cx23885/cx23885-417.c
index 7b0e8c0..19154b6 100644

drivers/media/video/cx23885/cx23885-417.c +0 -0
@@ -1585,7 +1585,7 @@ static int mpeg_open(struct inode *inode, struct

file *file)

       list_for_each(list, &cx23885_devlist) {
               h = list_entry(list, struct cx23885_dev, devlist);
-               if (h->v4l_device->minor == minor) {
+               if (h->v4l_device && h->v4l_device->minor == minor) { dev = h;
                       break;
               }

(quoted diff) +0 -0

b/drivers/media/video/cx23885/cx23885-video.c
index 6047c78..a2b5a0c 100644

drivers/media/video/cx23885/cx23885-video.c +0 -0
@@ -733,7 +733,7 @@ static int video_open(struct inode *inode, struct

file *file)

       list_for_each(list, &cx23885_devlist) {
               h = list_entry(list, struct cx23885_dev, devlist);
-               if (h->video_dev->minor == minor) {
+               if (h->video_dev && h->video_dev->minor == minor) { dev  = h;
                       type = V4L2_BUF_TYPE_VIDEO_CAPTURE; }

--
Jarod Wilson
[email protected]

#2 lore raw
Quoted 4 lines · Michael Krufky

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

Mike,

A version of this is already in the v4l-dvb hg development repository:

hg log -vp --limit 1 linux/drivers/media/video/cx23885/cx23885-417.c hg log -vp --limit 2 linux/drivers/media/video/cx23885/cx23885-video.c

I helped Mark work through the solution: I coded some of it, he coded some of it and he also tested it.

Regards,
Andy

Quoted 78 lines

---------- Forwarded message ----------
From: Jarod Wilson <[email protected]>
Date: Tue, Feb 24, 2009 at 6:00 PM
Subject: [stable] [PATCH] 2.6.27.y: fix NULL ptr deref in cx23885 video_open
To: [email protected]
Cc: Mike Krufky <[email protected]>

From: Mark Jenks
https://www.redhat.com/mailman/private/video4linux-list/2009-January/msg00041.html

The Hauppauge WinTV HVR-1800 tv tuner card has both digital and analog abilities, both of which are supported by v4l/dvb under 2.6.27.y. The analog side also features a hardware mpeg2 encoder. The HVR-1250 tv tuner card has both digital and analog abilities, but analog isn't currently supported under any kernel. These cards both utilize the cx23885 driver, but with slightly different usage. When the code paths for each card is executed, they wind up poking a cx23885_devlist, which contains devices from both of the cards, and access attempts are made to portions of 'struct cx23885_dev' that aren't valid for that device. Simply add some extra checks before trying to access these structs.

More gory details:
http://article.gmane.org/gmane.linux.drivers.dvb/46630

This was triggering on my own system at home w/both cards in it, and no longer happens with this patch included.

---

 drivers/media/video/cx23885/cx23885-417.c   |    2 +-
 drivers/media/video/cx23885/cx23885-video.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/media/video/cx23885/cx23885-417.c +1 -1
index 7b0e8c0..19154b6 100644
@@ -1585,7 +1585,7 @@ static int mpeg_open(struct inode *inode, struct
 file *file)
 
       list_for_each(list, &cx23885_devlist) {
               h = list_entry(list, struct cx23885_dev, devlist);
-               if (h->v4l_device->minor == minor) {
+               if (h->v4l_device && h->v4l_device->minor == minor) {
                       dev = h;
                       break;
               }
drivers/media/video/cx23885/cx23885-video.c +1 -1
index 6047c78..a2b5a0c 100644
@@ -733,7 +733,7 @@ static int video_open(struct inode *inode, struct
 file *file)
 
       list_for_each(list, &cx23885_devlist) {
               h = list_entry(list, struct cx23885_dev, devlist);
-               if (h->video_dev->minor == minor) {
+               if (h->video_dev && h->video_dev->minor == minor) {
                       dev  = h;
                       type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
               }

--
Jarod Wilson
[email protected]
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw
Quoted 12 lines · Andy Walls
Quoted 5 lines · Michael Krufky

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

Mike,

A version of this is already in the v4l-dvb hg development repository:

Quoted 8 lines

hg log -vp --limit 1 linux/drivers/media/video/cx23885/cx23885-417.c hg log -vp --limit 2 linux/drivers/media/video/cx23885/cx23885-video.c

I helped Mark work through the solution: I coded some of it, he coded some of it and he also tested it.

Regards,
Andy

I'm aware of that, Andy -- That's why I am sending this off to the -stable team for 2.6.27.y

Thanks & regards,

Mike

#4 lore raw
Quoted 16 lines · Michael Krufky
Quoted 15 lines · Andy Walls
Quoted 5 lines · Michael Krufky

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

Mike,

A version of this is already in the v4l-dvb hg development repository:

hg log -vp --limit 1 linux/drivers/media/video/cx23885/cx23885-417.c hg log -vp --limit 2 linux/drivers/media/video/cx23885/cx23885-video.c

Quoted 8 lines
Quoted 5 lines

I helped Mark work through the solution: I coded some of it, he coded some of it and he also tested it.

Regards,
Andy

I'm aware of that, Andy -- That's why I am sending this off to the -stable team for 2.6.27.y

Ooops. Sorry for my cluelessness. :)

Regards,
Andy

Quoted 8 lines

Thanks & regards,

Mike
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#5 lore raw
In reply to Michael Krufky
Quoted 2 lines · Michael Krufky

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

What is the git commit id of the patch in Linus's tree that matches up with this?

thanks,

greg k-h

#6 lore raw
Quoted 10 lines · Greg KH
Quoted 2 lines · Michael Krufky

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

What is the git commit id of the patch in Linus's tree that matches up with this?

thanks,

greg k-h

Now that I look closer, seems there's an even more complete patch already in linus' tree, commit id:

cd8f894eacf13996d920fdd2aef1afc55156b191

Shoulda just forwarded that along, but I was under the impression that area of code had changed significantly in 2.6.28 and the patch was no longer applicable. Maybe that was the v4l/dvb hg tree though... So I'd just grab that, and add:

--jarod

#7 lore raw
Quoted 13 lines · Jarod Wilson
Quoted 7 lines · Greg KH
Quoted 2 lines · Michael Krufky

Can we have this merged into -stable? Jarod Wilson sent this last month, but he left off the cc to [email protected]

What is the git commit id of the patch in Linus's tree that matches up with this?
thanks,
greg k-h

Now that I look closer, seems there's an even more complete patch already in linus' tree, commit id:

Quoted 8 lines

cd8f894eacf13996d920fdd2aef1afc55156b191

Shoulda just forwarded that along, but I was under the impression that area of code had changed significantly in 2.6.28 and the patch was no longer applicable. Maybe that was the v4l/dvb hg tree though... So I'd just grab that, and add:

Ok, will do that.

thanks,

greg k-h