[SOLVED] Chromium 35 rendering corruption on intel graphics card

adirat · 2014-05-21 14:35:55

Hi. Just upgraded to Chromium 35 and now get rendering artefacts all over the screen like on this panel: http://i.imgur.com/GUx0Q0C.jpg
Those artefacts appear only when Chromium is running and they were not present in Chromium 34.

From lspci, the graphics card is intel hd 4000

00:00.0 Host bridge: Intel Corporation 3rd Gen Core processor DRAM Controller (rev 09)
00:01.0 PCI bridge: Intel Corporation Xeon E3-1200 v2/3rd Gen Core processor PCI Express Root Port (rev 09)
00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09)

kernel, xserver and ddx

core/linux 3.14.4-1 (base) [installed]
extra/xorg-server 1.15.1-1 (xorg) [installed]
extra/xf86-video-intel 2.99.911-2 (xorg-drivers xorg) [installed]

cat /etc/X11/xorg.conf.d/20-intel.conf

Section "Device"
   Identifier  "Intel Graphics"
   Driver      "intel"
   Option      "AccelMethod"  "sna"
   Option      "TearFree"     "true"
EndSection

Changed AccelMethod from sna to uxa and the artefacts are gone, all rendering is ok except that I get screen tearing.
The only method I found to get rid of tearing is to use sna..

So my solutions are these:
1. Use Chromium 35, AccelMethod "uxa" with screen tearing
2. Downgrade to Chromium 34, AccelMethod "sna" without screen tearing

Does anyone else have this problem? Any help is appreciated

Thanks

Last edited by adirat (2014-05-21 20:13:16)

ooo · 2014-05-21 15:08:10

No rendering issues here, on gnome with intel, sna and tearfree enabled. Other than that, I've disabled vsync if that matters.

You could try glamor accelmethod. iirc there should be no tearing at least with tearfree enabled plus it should be faster than uxa too.

graysky · 2014-05-21 15:24:32

No problems here but I do not have /etc/X11/xorg.conf.d/20-intel.conf at all.

% lspci | grep -i graphic
00:02.0 VGA compatible controller: Intel Corporation Xeon E3-1200 v2/3rd Gen Core processor Graphics Controller (rev 09)

xorg-server 1.15.1-1
xf86-video-intel 2.99.911-2

TrevorBramble · 2014-05-21 16:24:11

Yup, I see this behavior too. Just reverted Chromium to 34 until I have time to troubleshoot.

$ lspci |grep -i graphic
00:02.0 VGA compatible controller: Intel Corporation Haswell-ULT Integrated Graphics Controller (rev 0b)

I don't have any AccelMethod defined in Xorg configs.

Shinto · 2014-05-21 16:41:40

Same problem here, when I try to resize the window it leaves artifacts and I have to restart Chromium.

lspci | grep -i graphic
3:00:02.0 VGA compatible controller: Intel Corporation Xeon E3-1200 v3/4th Gen Core Processor Integrated Graphics Controller (rev 06)

blackout23 · 2014-05-21 17:24:41

I heard someone say that disabling "threaded compositing" for chromium solves some of the redraw problems. Might help here aswell. Try to launch it with --disable-threaded-compositing.

Last edited by blackout23 (2014-05-21 17:26:07)

dfeng · 2014-05-21 19:22:07

same issue
fixed it:
1.chromium-settings-"use hardware acceleration when available" on
2./etc/chromium/default, add flags
CHROMIUM_FLAGS="--ignore-gpu-blacklist"

adirat · 2014-05-21 20:12:11

dfeng wrote:

same issue
fixed it:
1.chromium-settings-"use hardware acceleration when available" on
2./etc/chromium/default, add flags
CHROMIUM_FLAGS="--ignore-gpu-blacklist"

This fixed it! Now Chromium 35 + sna + no tearing work without issue

Thank you very much everyone

TrevorBramble · 2014-05-22 00:16:42

Worked for me too; thanks adirat!

drog · 2014-05-22 01:21:51

I have the same issue, but with google-chrome from AUR
¿where is the configuration file?

Last edited by drog (2014-05-22 02:16:58)

adirat · 2014-05-22 03:05:22

TrevorBramble wrote:

Worked for me too; thanks adirat!

Thank dfeng, because it's his solution

drog wrote:

I have the same issue, but with google-chrome from AUR
¿where is the configuration file?

It should be /etc/default/google-chrome, if it does not exist, create it.
But first, try to start chrome from the cmd line and append --ignore-gpu-blacklist to see if it works

Later edit: Or you can navigate in chrome to chrome://flags/ there you have the flag "Override software rendering list", try enabling it, this is the easiest method

Last edited by adirat (2014-05-22 03:17:10)

Pse · 2014-05-22 03:45:35

Ironic that disabling the blacklist solves a problem

ezzetabi · 2014-05-22 10:00:46

Solved here too. Thanks. It is really ironic, yes.

Last edited by ezzetabi (2014-05-22 10:01:26)

adventurer · 2014-05-22 12:24:42

Both suggested settings/flags were enabled here but I still had rendering problems. However, after enabling "multithreaded GPU compositing of web content" in chrome://flags (--enable-threaded-compositing) everything seems to be okay. (This flag is only avalable for Linux.)

LeCrayonVert · 2014-05-22 17:27:05

What about CPU usage ? I've noticed that the CPU usage increases dramatically when playing a video (either an HTML5 video or a flash video (using the pepper api from chrome)). It used to be very low even on a 1080p HD video... (I don't have this issue outside the browser with mplayer or vlc, so it's not a driver issue as I've first thought).
Enabling threaded compositing only makes things worse.
Can anyone confirm this ? (I have an intel hd 4600 graphics).

adventurer · 2014-05-22 17:31:33

LeCrayonVert wrote:

What about CPU usage ? I've noticed that the CPU usage increases dramatically when playing a video (either an HTML5 video or a flash video (using the pepper api from chrome)). It used to be very low even on a 1080p HD video...

I cannot observe that on my system (HD 4000).

adirat · 2014-05-22 19:44:55

LeCrayonVert wrote:

What about CPU usage ? I've noticed that the CPU usage increases dramatically when playing a video (either an HTML5 video or a flash video (using the pepper api from chrome)). It used to be very low even on a 1080p HD video... (I don't have this issue outside the browser with mplayer or vlc, so it's not a driver issue as I've first thought).
Enabling threaded compositing only makes things worse.
Can anyone confirm this ? (I have an intel hd 4600 graphics).

Yes, same issue here. Other apps are unaffected.
cpu ussage goes down when video is running but not displayed (like switching to another tab / app) or paused.
cpu ussage also increases/decreases with the size of the player on screen, so this definetely is some chromium rendering issue.

Edit: Using Chromium 35 + pepper flash on intel hd 4000

Later edit: Same issue with chromium-pepper-flash-dev... All similar bug reports are 'pending' (https://code.google.com/p/chromium/issu … ?id=338578).
The adobe plugin is not supported in chromium anymore, so we're stuck

Last edited by adirat (2014-05-22 20:40:34)

LeCrayonVert · 2014-05-23 17:06:26

adirat wrote:

LeCrayonVert wrote:
What about CPU usage ? I've noticed that the CPU usage increases dramatically when playing a video (either an HTML5 video or a flash video (using the pepper api from chrome)). It used to be very low even on a 1080p HD video... (I don't have this issue outside the browser with mplayer or vlc, so it's not a driver issue as I've first thought).
Enabling threaded compositing only makes things worse.
Can anyone confirm this ? (I have an intel hd 4600 graphics).
Yes, same issue here. Other apps are unaffected.
cpu ussage goes down when video is running but not displayed (like switching to another tab / app) or paused.
cpu ussage also increases/decreases with the size of the player on screen, so this definetely is some chromium rendering issue.
Edit: Using Chromium 35 + pepper flash on intel hd 4000
Later edit: Same issue with chromium-pepper-flash-dev... All similar bug reports are 'pending' (https://code.google.com/p/chromium/issu … ?id=338578).
The adobe plugin is not supported in chromium anymore, so we're stuck

Well even HTML5 videos (which do not depend on pepper flash, right ?) have higher cpu usage.
And I've got the same issue with Firefox (using the extra/flashplugin 11.2.202.359 package).
Weird, isn't it ?

According to the page chrome://gpu, it is somehow related to https://code.google.com/p/chromium/issu … ?id=137247 and https://code.google.com/p/chromium/issu … ?id=133828

Last edited by LeCrayonVert (2014-05-23 17:11:10)

adirat · 2014-05-23 23:39:38

I found this article which is relevant for our issues, the blacklisted gpu rendering and the high CPU usage may to be linked

http://www.phoronix.com/scan.php?page=n … px=MTYyMDk

LeCrayonVert wrote:

Well even HTML5 videos (which do not depend on pepper flash, right ?) have higher cpu usage.
And I've got the same issue with Firefox (using the extra/flashplugin 11.2.202.359 package).
Weird, isn't it ?

Afaik only the flash plugin supports gpu acc, so high cpu usage on html5 format was expected...
But anyway, even without gpu acc we should not be seeing so much cpu usage, so you are right. Something is weird here

Later edit:

There are still rendering problems when for ex. using a plugin like password revealer, the entire screen flickers & goes blank https://chrome.google.com/webstore/deta … nenonklpbn

For the time being I'll switch to firefox full-time. Also, the video cpu usage problem is not so severe there.

Last edited by adirat (2014-05-24 04:15:51)

ooo · 2014-05-24 09:38:13

adirat wrote:

I found this article which is relevant for our issues, the blacklisted gpu rendering and the high CPU usage may to be linked
http://www.phoronix.com/scan.php?page=n … px=MTYyMDk

That is about hardware accelerated video decoding. The support for that has always been missing from pepper-flash and chromium.
Page rendering and video output should still be hardware accelerated, unless blacklisted/disabled.

It seems that I'm getting some rendering issues after all, even with "Override software rendering list" enabled.
It only happens sometimes, but basically chromium window starts filling with black squares, like it fails to render some parts of the screen. Restarting the browser fixes this though.

LeCrayonVert · 2014-05-24 11:19:24

ooo => Yes, I guess you're right... After all, even without hardware acceleartion, an intel core i5/i7 processor should be able to decode a simple video without eating up all the CPU !

renegat · 2014-05-25 12:07:02

I can confirm massive rendering problems as described above on:
Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller (rev 03)
"Override software rendering list" enabled did solve most but not all issues.
In particular the lxpanel is loosing icons and behaves weird on chromium startup.

adventurer · 2014-05-28 17:28:50

renegat wrote:

I can confirm massive rendering problems as described above on:
Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller (rev 03)
"Override software rendering list" enabled did solve most but not all issues.

Have you also tried enabling

--force-compositing-mode

and

--enable-threaded-compositing

in chrome://flags and hardware acceleration in settings?

With those switches I don't have any probelms any more.

WooLoo · 2014-06-11 17:58:27

Similar issue. Chromium 33 worked fine. Full system upgrade after a few months without.

lspci

00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09)
01:00.0 VGA compatible controller: NVIDIA Corporation GK104M [GeForce GTX 675MX] (rev ff)

I've tried 2 chromium builds from [extra], 35.0.1916.114-1 and 35.0.1916.153-1, as well as chromium-nightly and chromium-browser-bin, with and without flash (or pepper-flash for AUR builds).
I've also tried setting the Intel accel method to UXA, as mentioned above and on the wiki - this causes only a small border of the window to appear, or only an outline of the window, before quickly segfaulting.
Setting the various flags as mentioned in this thread and elsewhere either makes no difference or leads to immediate segfaults related to GTK.

From the moment it starts, one chromium process is using 65% CPU, then quickly jumps to 90% to 100%. The X server process also jumps to about 30% CPU.

I'm using subtle wm (version 0.11.r3243-1). Last two Firefox repo builds work fine.

To see if subtle is the culprit, I've tried running chromium on a fresh minimal LXDE install, at first it had similar compositing issues and the same CPU issue, but deleting the entire ~/.config/chromium dir fixed this, and now a fresh default chromium install works fine, CPU usage is normal too - but only under LXDE.

Chromium was reporting something like

ATTENTION: default value of option force_s3tc_enable overridden by environment.

but that's gone with the very latest build from [extra].

Finally, I've tried switching Intel's driver accelerator method to Glamor, which fixes the compositing issues, but causes chromium to use 75%-85% CPU and X about 95%-100% CPU (and subtle about 20%), again messing with flags changes nothing.

Not sure what to try next (aside from waiting for better Optimus and/or Intel drivers, of course).

foutrelis · 2014-06-12 10:41:44

Can you try installing xf86-video-intel 2.99.912-1 from [testing] and see if that solves these glitches?

Arch Linux

#1 2014-05-21 14:35:55

[SOLVED] Chromium 35 rendering corruption on intel graphics card

#2 2014-05-21 15:08:10

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#3 2014-05-21 15:24:32

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#4 2014-05-21 16:24:11

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#5 2014-05-21 16:41:40

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#6 2014-05-21 17:24:41

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#7 2014-05-21 19:22:07

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#8 2014-05-21 20:12:11

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#9 2014-05-22 00:16:42

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#10 2014-05-22 01:21:51

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#11 2014-05-22 03:05:22

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#12 2014-05-22 03:45:35

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#13 2014-05-22 10:00:46

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#14 2014-05-22 12:24:42

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#15 2014-05-22 17:27:05

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#16 2014-05-22 17:31:33

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#17 2014-05-22 19:44:55

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#18 2014-05-23 17:06:26

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#19 2014-05-23 23:39:38

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#20 2014-05-24 09:38:13

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#21 2014-05-24 11:19:24

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#22 2014-05-25 12:07:02

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#23 2014-05-28 17:28:50

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#24 2014-06-11 17:58:27

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

#25 2014-06-12 10:41:44

Re: [SOLVED] Chromium 35 rendering corruption on intel graphics card

Board footer