linux-kernel - Re: UVCvideo: Failed to resubmit video URB (-27) with Linux 3.3.3

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Thu, 3 May 2012 16:34:54 +0200
From:	Anisse Astier <anisse@...ier.eu>
To:	Laurent Pinchart <laurent.pinchart@...asonboard.com>,
	Oleksij Rempel <bug-track@...her-privat.net>
Cc:	linux-uvc-devel@...ts.sourceforge.net, linux-media@...r.kernel.org,
	Mauro Carvalho Chehab <mchehab@...hat.com>,
	linux-kernel@...r.kernel.org, linux-usb@...r.kernel.org
Subject: Re: UVCvideo:  Failed to resubmit video URB (-27) with Linux 3.3.3

On Wed, 02 May 2012 14:24:11 +0200, Laurent Pinchart <laurent.pinchart@...asonboard.com> wrote :

> Hi Anisse,
> 
> On Thursday 26 April 2012 20:07:21 Anisse Astier wrote:
> > Hi,
> > 
> > I'm experiencing a problem with uvcvideo with kernel 3.3.3 and today's
> > Linus' tree.
> > 
> > Problem not reproduced in 3.2.15, so this could be labelled as a regression.
> > 
> > See webcam lsusb and (verbose!) dmesg log in attachment, which exhibits
> > the problem.
> > 
> > We see lots of error (-18 = -EXDEV), that indicate that URB was too late
> > and then dropped, and they add up until we reach the "Failed to resubmit
> > video URB" scheduling issue.
> 
> Those are USB controller issues. The uvcvideo driver submits URBs with the 
> URB_ISO_ASAP transfer flag, so the controller should not fail to schedule 
> them.
Yes, it's weird.

So I followed Oleksij's adviced and reverted the following commit groups:
 - 66847ef [media] uvcvideo: Add UVC timestamps support, 3afedb9 [media]
   uvcvideo: Don't skip erroneous payloads and ed0ee0c [media] uvcvideo:
   Fix race-related crash in uvc_video_clock_update()
 - ab86e9e [media] uvcvideo: Allow userptr IO mode and 6998b6f [media]
   uvcvideo: Use videobuf2-vmalloc
 - 3d95e93 [media] uvcvideo: Move fields from uvc_buffer::buf to
   uvc_buffer
 - c4d99f8 [media] uvcvideo: Ignore GET_RES error for XU controls
 - 806e23e [media] uvcvideo: Fix integer overflow in uvc_ioctl_ctrl_map()
 - d0d9748 [media] uvcvideo: Use kcalloc instead of kzalloc to allocate
   array
None of this fixed the issue.

So I just decided to revert the whole uvc driver: git checkout v3.2 --
drivers/media/video/uvc.
But, the problem was still here.

I reverted the usb host code in drivers/usb/host/. Again the problem was
reproduced (both with and without 3.2's uvcvideo driver)

Then I tested the whole kernel v3.2, and indeed it still works very well.

So this problem could have it's root in USB core changes, or even a
combilation of USB and UVC changes.


> 
> > Installed libv4l version is 0.8.6.
> > I'm reproducing this with: gst-launch-0.10 --verbose v4l2src  ! xvimagesink
> > (Skype exhibits the problem too, while it isn't using gstreamer, so it
> > really seems to come from kernel. Also, doesn't happen with 3.2)
> > 
> > This is the first part of the problem. The second part is that if I
> > restart the webcam with gst-launch after the first failure, I have a
> > total freeze, just after these messages in the log (fetched with
> > netconsole, I wasn't able to get a panic trace):
> > 
> > [  191.796217] uvcvideo: Marking buffer as bad (error bit set).
> > [  191.796233] uvcvideo: Marking buffer as bad (error bit set).
> > [  191.796244] uvcvideo: Marking buffer as bad (error bit set).
> > [  191.796252] uvcvideo: Marking buffer as bad (error bit set).
> > [  191.796259] uvcvideo: Frame complete (EOF found).
> > [  191.796265] uvcvideo: EOF in empty payload.
> > [  192.972803] uvcvideo: Marking buffer as bad (error bit set).
> > [  192.972818] uvcvideo: Dropping payload (out of sync).
> > [  194.289463] uvcvideo: Marking buffer as bad (error bit set).
> > [  194.289478] uvcvideo: Frame complete (FID bit toggled).
> > [  194.289486] uvcvideo: Marking buffer as bad (error bit set).
> > [  194.289493] uvcvideo: Frame complete (FID bit toggled).
> > [  194.289499] uvcvideo: Marking buffer as bad (error bit set).
> > [  194.289505] uvcvideo: Frame complete (FID bit toggled).
> > [  194.289511] uvcvideo: Marking buffer as bad (error bit set).
> > [  194.289518] uvcvideo: Frame complete (FID bit toggled).
> > [  194.289524] uvcvideo: Marking buffer as bad (error bit set).
> > [  194.289531] uvcvideo: Frame complete (FID bit toggled).
> >
> > Last but not least, uvcvideo is un-bisectable because there were a few
> > crash-fixes during the 3.3 development cycle. I started bisecting and got
> > kernel panics.
> 
> Are the kernel panics related to uvcvideo ? There's one known bug introduced 
> between v3.2 and v3.3 and fixed (before v3.3) in commit 
> 8e57dec0454d8a3ba987d18b3ab19922c766d4bc.
I don't think that's it. As I've said, problem exists with both 3.3.3 and
Linus' 3.4-rc5.


--
Anisse
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/