Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 367795 - compositing over flash crashes X with nvidia drivers
Summary: compositing over flash crashes X with nvidia drivers
Status: RESOLVED FIXED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: [OLD] Unspecified (show other bugs)
Hardware: AMD64 Linux
: Normal normal (vote)
Assignee: Doug Goldstein (RETIRED)
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-05-17 18:39 UTC by Martin von Gagern
Modified: 2012-07-15 02:16 UTC (History)
4 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments
emerge --info (emerge--info,6.26 KB, text/plain)
2011-05-17 18:39 UTC, Martin von Gagern
Details
xorg.0.log (Xorg.0.log.old,28.58 KB, text/plain)
2011-12-11 16:41 UTC, James Broadhead
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Martin von Gagern 2011-05-17 18:39:55 UTC
Created attachment 273697 [details]
emerge --info

OK, this issue appears difficult to reproduce, but here is what happened:

1. Watched full screen video using www-plugins/adobe-flash-10.3.181.14-r1.
   According to the postinst message, there is no 64bit version, so it
   should be using www-plugins/nspluginwrapper-1.3.0 instead.
2. An E-Mail arrived, causing mail-client/thunderbird-3.1.10 to attempt
   displaying a notification window. Before my last system upgrade,
   when this happened the video image would freeze for a moment,
   presumably while compositing was enabled.
   I guess my compisite manager is kde-base/kwin-4.6.3-r1.
3. This time, however, the screen went black.
4. ESC to leave flash full screen mode didn't work.
5. Remote login via ssh worked.
6. Xorg.0.log contained these two lines after the usual startup stuff:
   (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000e32c, 0x0000e32c)
   (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000e670, 0x0000e670)
   The driver behind this is x11-drivers/nvidia-drivers-270.41.06.
   The X server is x11-base/xorg-server-1.9.5 due to bug #364019 comment #9.
7. killall -KILL for flash or firefox failed to restore my screen.
8. remote reboot was the only option I could think of, so that's what I did.

I guess this will be a major problem to solve, due to the large number of packages involved, the difficulty in reproducing the issue, and the fact that large parts of the involved code are closed source. So I don't expect miracles. Nevertheless, perhaps this report can be useful if other related reports pour in, so that you can form a coherent image of all the things involved.
Comment 1 Jeroen Roovers (RETIRED) gentoo-dev 2011-05-19 16:02:16 UTC
Please post the output of `emerge -vpq x11-drivers/nvidia-drivers' too.
Comment 2 Martin von Gagern 2011-05-19 20:12:28 UTC
(In reply to comment #1)
> Please post the output of `emerge -vpq x11-drivers/nvidia-drivers' too.

[ebuild   R   ] x11-drivers/nvidia-drivers-270.41.06
                USE="acpi gtk (multilib) -custom-cflags"
Comment 3 Vincent Le Ligeour 2011-06-02 15:29:14 UTC
I can confirm this using nvidia-drivers-270.41.06 and www-plugins/adobe-flash-10.3.181.14-r1
I can reproduce this quite easily by watching youtube, going to fullscreen and wait for the end of the video.
I use kde with kwin.
Remotely killing firefox/flash player/kwin does not restor the screen.
It seems that Xorg is waiting for something because when restarting kwin, kwin does not succeed in connecting to X (saying so when killing Xorg).
I never aim this bug before upgrading to the lastest adobe-flash yesterday.
Comment 4 James Broadhead 2011-10-26 20:06:02 UTC
I believe that this is the same problem that I have been having, but am not certain. I am finding that on occasion (hard to deliberately reproduce, but once per day or so), some interaction with flash in chromium causes X to freeze at 100% CPU, disregarding keyboard input and sometimes allowing the mouse to move but not registering clicks (sometimes, the mouse does not move). Restarting X via ssh is ineffective, as it does not die properly following killall, so I resort to restarts. 

Reproduction (often causes this problem, not always)
- Open a flash video in chromium 
- Close the tab
(Repeat as necessary).


My Xorg.0.log contains:
[mi] EQ overflowing. The server is probably stuck in an infinite loop.
0: /usr/bin/X (xorg_backtrace+0x38) [0x80a6678]

I will rebuild X with USE="debug" and try to get better output. 

My kernel messages: 
kernel: [32294.449466] NVRM: Xid (0000:01:00): 13, 0001 00000000 00005097 000015e0 00000000 00000080
This message gets about ~50 Google hits, which seems to indicate that it's being experienced elsewhere.

3.0.7-gentoo i686
www-client/chromium-15.0.874.102 gnome kerberos linguas_en_GB
www-client/chromium-16.0.891.0 gnome kerberos linguas_en_GB
www-plugins/adobe-flash-11.0.1.152 kde vdpau
x11-drivers/nvidia-drivers-275.09.07 acpi gtk kernel_linux
x11-base/xorg-server-1.10.4-r1 ipv6 kdrive nptl udev xorg

I switched to Opera for a few days, and did not experience this problem.
Comment 5 James Broadhead 2011-12-11 16:41:18 UTC
Created attachment 295475 [details]
xorg.0.log

This has just happened to me again, having switched back to Chromium-15* for the last few weeks (but with light computer usage). 

x86
www-plugins/adobe-flash-11.1.102.55 kde vdpau
x11-drivers/nvidia-drivers-285.05.09-r1 acpi gtk kernel_linux
x11-base/xorg-server-1.10.4-r1 ipv6 kdrive nptl udev xorg
Comment 6 James Broadhead 2011-12-11 16:44:03 UTC
# dmesg | grep -i nv
[    0.711288] rtc0: alarms up to one month, y3k, 114 bytes nvram, hpet irqs
[    7.511803] nvidia: module license 'NVIDIA' taints kernel.
[    8.526789] nvidia 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[    8.526800] nvidia 0000:01:00.0: setting latency timer to 64
[    8.526927] NVRM: loading NVIDIA UNIX x86 Kernel Module  285.05.09  Fri Sep 23 18:53:08 PDT 2011
[14456.299468] NVRM: Xid (0000:01:00): 13, 0001 00000000 00005097 00000548 3f800000 00000080

This was definitely triggered by having a browser tab open with a flash video - I believe that I was closing the tab at the time (which would be consistent with the earlier report)
Comment 7 James Broadhead 2011-12-11 16:46:29 UTC
Additionally, this laptop has a somewhat unusual graphics card; 

01:00.0 VGA compatible controller: nVidia Corporation G84M [Quadro NVS 140M] (rev a1)
Comment 8 Doug Goldstein (RETIRED) gentoo-dev 2012-07-14 20:24:28 UTC
(In reply to comment #7)
> Additionally, this laptop has a somewhat unusual graphics card; 
> 
> 01:00.0 VGA compatible controller: nVidia Corporation G84M [Quadro NVS 140M]
> (rev a1)

Fairly common in workstation type laptops (e.g. Lenovo's).
Comment 9 Doug Goldstein (RETIRED) gentoo-dev 2012-07-14 20:25:25 UTC
(In reply to comment #5)
> Created attachment 295475 [details]
> xorg.0.log
> 
> This has just happened to me again, having switched back to Chromium-15* for
> the last few weeks (but with light computer usage). 
> 
> x86
> www-plugins/adobe-flash-11.1.102.55 kde vdpau
> x11-drivers/nvidia-drivers-285.05.09-r1 acpi gtk kernel_linux
> x11-base/xorg-server-1.10.4-r1 ipv6 kdrive nptl udev xorg

Doesn't Chrome bundle its own Flash? Or does Chromium not and you have to use the Flash plugin?

Are you still experiencing this issue with the latest drivers, latest X and latest Chromium?
Comment 10 James Broadhead 2012-07-14 23:38:14 UTC
(In reply to comment #9)
> (In reply to comment #5)
> > Created attachment 295475 [details]
> > xorg.0.log
> > 
> > This has just happened to me again, having switched back to Chromium-15* for
> > the last few weeks (but with light computer usage). 
> > 
> > x86
> > www-plugins/adobe-flash-11.1.102.55 kde vdpau
> > x11-drivers/nvidia-drivers-285.05.09-r1 acpi gtk kernel_linux
> > x11-base/xorg-server-1.10.4-r1 ipv6 kdrive nptl udev xorg
> 
> Doesn't Chrome bundle its own Flash? Or does Chromium not and you have to
> use the Flash plugin?
> 
> Are you still experiencing this issue with the latest drivers, latest X and
> latest Chromium?

Good question, I'm not sure. However, I haven't seen this in quite a while, so let's mark RESOLVED.
Comment 11 Doug Goldstein (RETIRED) gentoo-dev 2012-07-15 02:16:39 UTC
Somewhere along the way when something got bumped.