My favorites | Sign in
Project Home Downloads Wiki Issues Code Search
New issue   Search
for
  Advanced search   Search tips   Subscriptions
Issue 145387: Content area not redrawing correctly
15 people starred this issue and may be notified of changes. Back to list
 
Project Member Reported by pkasting@chromium.org, Aug 28, 2012
Chrome Version       : 23.0.1246.0
OS Version: 6.0 (Windows Vista, Windows Server 2008)

My Chrome Dev build seems to be having major problems with redrawing the content area, especially when my window is maximized.  Doesn't happen on every URL but it does happen on a lot.  Bug reports seem to trigger it nicely.

Attached is what I occasionally see when I do a navigation.  Note that the black frame is approximately the size of my non-maximized Chrome window.  Other times I'll just see white.  Restoring my window immediately causes the page to draw and I can then re-maximize.

Sometimes I also see images like those described above flash visibly as I switch tabs.  When this happens, it happens consistently and repeatedly as I switch back and forth -- that is, it's not the issue where the tab framebuffer image is stale from the window being resized while the tab was invisible (which corrects itself after one time).  Instead I'll see something like the black frame every single time until I close some tabs and open new ones.

I don't have any weird about:flags enabled.  I'm happy to disable some to test things on request.

Guessing this is GPU/compositing, but also CCing sky in case there's some weird "browser is not properly triggering a render area repaint" issue like he's fixed numerous times.
redraw.png
65.7 KB   View   Download
Aug 28, 2012
#1 pkasting@chromium.org
Note: Still other times I see the blue-grey color here but without a black frame or white resize corner.
Aug 28, 2012
#2 jamesr@chromium.org
Can you paste your about:gpu ?
Aug 28, 2012
#3 pkasting@chromium.org
Graphics Feature Status
Canvas: Hardware accelerated
Compositing: Hardware accelerated on all pages
3D CSS: Hardware accelerated
CSS Animation: Accelerated
WebGL: Hardware accelerated
WebGL multisampling: Hardware accelerated
Flash 3D: Hardware accelerated
Flash Stage3D: Hardware accelerated
Texture Sharing: Hardware accelerated
Video Decode: Hardware accelerated
Version Information
Data exported	Tue Aug 28 2012 16:54:21 GMT-0700 (Pacific Daylight Time)
Chrome version	23.0.1246.0 (Official Build 153452) dev-m
Operating system	Windows NT 6.0 SP2
Software rendering list version	2.9
ANGLE revision	1254
2D graphics backend	Skia
Performance Information
Graphics	0.0
Gaming	0.0
Overall	0.0
Driver Information
Initialization time	388
GPU0	VENDOR = 0x10de, DEVICE= 0x0658
Optimus	false
AMD switchable	false
Driver vendor	NVIDIA
Driver version	7.15.11.8265
Driver date	4-22-2009
Pixel shader version	3.0
Vertex shader version	3.0
GL version	
GL_VENDOR	
GL_RENDERER	
GL_VERSION	
GL_EXTENSIONS	
Diagnostics
0
b3DAccelerationEnabled	true
b3DAccelerationExists	true
bAGPEnabled	true
bAGPExistenceValid	true
bAGPExists	true
bCanRenderWindow	true
bDDAccelerationEnabled	true
bDriverBeta	false
bDriverDebug	false
bDriverSigned	false
bDriverSignedValid	false
bNoHardware	false
dwBpp	32
dwDDIVersion	10
dwHeight	1920
dwRefreshRate	60
dwWHQLLevel	0
dwWidth	1200
iAdapter	0
lDriverSize	8636416
lMiniVddSize	0
szAGPStatusEnglish	Enabled
szAGPStatusLocalized	Enabled
szBGRASupported	true
szChipType	Quadro FX 380
szD3DStatusEnglish	Enabled
szD3DStatusLocalized	Enabled
szDACType	Integrated RAMDAC
szDDIVersionEnglish	10
szDDIVersionLocalized	10
szDDStatusEnglish	Enabled
szDDStatusLocalized	Enabled
szDXVAModes	ModeMPEG2_A ModeMPEG2_C ModeVC1_C ModeWMV9_C
szDescription	NVIDIA Quadro FX 380
szDeviceId	0x0658
szDeviceIdentifier	{D7B71E3E-4518-11CF-2753-302601C2CA35}
szDeviceName	\\.\DISPLAY1
szDisplayMemoryEnglish	2288 MB
szDisplayMemoryLocalized	2288 MB
szDisplayModeEnglish	1200 x 1920 (32 bit) (60Hz)
szDisplayModeLocalized	1200 x 1920 (32 bit) (60Hz)
szDriverAttributes	Final Retail
szDriverDateEnglish	4/22/2009 10:34:00
szDriverDateLocalized	4/22/2009 10:34:00 AM
szDriverLanguageEnglish	English
szDriverLanguageLocalized	English
szDriverName	nvd3dumx.dll,nvwgf2umx.dll,nvd3dum,nvwgf2um
szDriverSignDate	
szDriverVersion	7.15.0011.8265
szKeyDeviceID	Enum\PCI\VEN_10DE&DEV_0658&SUBSYS_063B10DE&REV_A1
szKeyDeviceKey	\Registry\Machine\System\CurrentControlSet\Control\Video\{B05AF945-43EB-4345-B782-05D7CAF588BD}\0000
szManufacturer	NVIDIA
szMiniVdd	n/a
szMiniVddDateEnglish	n/a
szMiniVddDateLocalized	n/a
szMonitorMaxRes	
szMonitorName	Generic PnP Monitor
szNotesEnglish	No problems found.
szNotesLocalized	No problems found.
szRegHelpText	
szRevision	
szRevisionId	0x00A1
szSubSysId	0x063B10DE
szTestResultD3D7English	Not run
szTestResultD3D7Localized	Not run
szTestResultD3D8English	Not run
szTestResultD3D8Localized	Not run
szTestResultD3D9English	Not run
szTestResultD3D9Localized	Not run
szTestResultDDEnglish	Not run
szTestResultDDLocalized	Not run
szVdd	n/a
szVendorId	0x10DE
1
b3DAccelerationEnabled	true
b3DAccelerationExists	true
bAGPEnabled	true
bAGPExistenceValid	true
bAGPExists	true
bCanRenderWindow	true
bDDAccelerationEnabled	true
bDriverBeta	false
bDriverDebug	false
bDriverSigned	false
bDriverSignedValid	false
bNoHardware	false
dwBpp	32
dwDDIVersion	10
dwHeight	1920
dwRefreshRate	60
dwWHQLLevel	0
dwWidth	1200
iAdapter	1
lDriverSize	8636416
lMiniVddSize	0
szAGPStatusEnglish	Enabled
szAGPStatusLocalized	Enabled
szBGRASupported	true
szChipType	Quadro FX 380
szD3DStatusEnglish	Enabled
szD3DStatusLocalized	Enabled
szDACType	Integrated RAMDAC
szDDIVersionEnglish	10
szDDIVersionLocalized	10
szDDStatusEnglish	Enabled
szDDStatusLocalized	Enabled
szDXVAModes	ModeMPEG2_A ModeMPEG2_C ModeVC1_C ModeWMV9_C
szDescription	NVIDIA Quadro FX 380
szDeviceId	0x0658
szDeviceIdentifier	{D7B71E3E-4518-11CF-2753-302601C2CA35}
szDeviceName	\\.\DISPLAY2
szDisplayMemoryEnglish	2288 MB
szDisplayMemoryLocalized	2288 MB
szDisplayModeEnglish	1200 x 1920 (32 bit) (60Hz)
szDisplayModeLocalized	1200 x 1920 (32 bit) (60Hz)
szDriverAttributes	Final Retail
szDriverDateEnglish	4/22/2009 10:34:00
szDriverDateLocalized	4/22/2009 10:34:00 AM
szDriverLanguageEnglish	English
szDriverLanguageLocalized	English
szDriverName	nvd3dumx.dll,nvwgf2umx.dll,nvd3dum,nvwgf2um
szDriverSignDate	
szDriverVersion	7.15.0011.8265
szKeyDeviceID	Enum\PCI\VEN_10DE&DEV_0658&SUBSYS_063B10DE&REV_A1
szKeyDeviceKey	\Registry\Machine\System\CurrentControlSet\Control\Video\{B05AF945-43EB-4345-B782-05D7CAF588BD}\0001
szManufacturer	NVIDIA
szMiniVdd	n/a
szMiniVddDateEnglish	n/a
szMiniVddDateLocalized	n/a
szMonitorMaxRes	
szMonitorName	Generic PnP Monitor
szNotesEnglish	No problems found.
szNotesLocalized	No problems found.
szRegHelpText	
szRevision	
szRevisionId	0x00A1
szSubSysId	0x063B10DE
szTestResultD3D7English	Not run
szTestResultD3D7Localized	Not run
szTestResultD3D8English	Not run
szTestResultD3D8Localized	Not run
szTestResultD3D9English	Not run
szTestResultD3D9Localized	Not run
szTestResultDDEnglish	Not run
szTestResultDDLocalized	Not run
szVdd	n/a
szVendorId	0x10DE
Crash List
09256d1108da9c3f: Friday, April 6, 2012 5:56:05 PM
cb1151ae78fac15d: Wednesday, February 29, 2012 3:19:47 PM
4a74a2ba82d27633: Wednesday, February 29, 2012 2:51:42 PM
42c6b19e691887c0: Tuesday, February 28, 2012 12:11:28 PM
cdcc434c7dd63ddc: Monday, November 21, 2011 1:27:57 PM
ba25c4de4a1c88dd: Monday, November 21, 2011 1:27:38 PM
d18b08dcb79ade34: Friday, October 7, 2011 3:53:25 PM
5655cfec2a4c728b: Wednesday, August 31, 2011 2:54:23 PM
222a3256a1145f33: Thursday, August 25, 2011 10:36:27 AM
e43b3074b6d931fe: Wednesday, July 20, 2011 11:05:31 AM
ddca7d52a81c3928: Wednesday, June 29, 2011 1:58:42 PM
afb6779511836408: Thursday, June 23, 2011 3:53:57 PM
1d971dcf45c77267: Tuesday, March 8, 2011 10:56:08 AM
db432a89d2aaa717: Monday, February 28, 2011 10:56:02 AM
d83412843d9734b3: Wednesday, February 23, 2011 9:05:06 AM
b07e965b8d1f3305: Wednesday, February 9, 2011 5:49:48 PM
2b1d231681a1111e: Monday, November 22, 2010 12:43:21 PM
e4f5d27d39591bd8: Monday, November 1, 2010 9:18:04 AM
70e8f974ee0e7085: Tuesday, October 12, 2010 10:59:16 AM
b57760915542ae90: Thursday, September 23, 2010 6:26:16 PM
6969c004a72416d8: Thursday, September 23, 2010 6:26:01 PM
1b6f4200c38100ac: Thursday, September 23, 2010 5:52:25 PM
4931c62aa2cc3291: Thursday, September 23, 2010 5:52:14 PM
348e31c20448fa46: Thursday, September 23, 2010 5:50:53 PM
45f216a3c1998305: Monday, August 23, 2010 5:26:50 PM
23a305b777ace0c3: Thursday, July 29, 2010 11:04:02 AM
5c8a0728227b404e: Friday, July 23, 2010 2:53:42 PM
bf893d075d92c19d: Friday, July 23, 2010 2:53:40 PM
5c8a0728227b4a2c: Friday, July 23, 2010 2:53:36 PM
5c8a0728227b428b: Friday, July 23, 2010 2:52:27 PM
5c8a0728227b4f67: Friday, July 23, 2010 2:52:18 PM
bf893d075d92cead: Friday, July 23, 2010 2:52:17 PM
fd0deacfb3f3dd3c: Friday, July 23, 2010 2:52:17 PM
5c8a0728227b410c: Friday, July 23, 2010 2:52:17 PM
169f6b89a8da3774: Friday, July 2, 2010 5:32:07 PM
28183616a5c1f4c0: Friday, July 2, 2010 4:56:59 PM
48dbaab303957a0a: Friday, June 18, 2010 12:20:12 PM
72b3becfd34399a5: Friday, June 18, 2010 12:10:22 PM
a80c2ad59f03494e: Friday, June 18, 2010 11:05:28 AM
464e05a17b3585cb: Friday, June 18, 2010 11:00:05 AM
21acb1e1e34693e9: Friday, June 18, 2010 10:40:35 AM
493d9a01701d51bb: Thursday, June 17, 2010 4:38:35 PM
64dd66f7027b07d6: Thursday, June 17, 2010 4:21:14 PM
817259bdd938d6e0: Monday, April 12, 2010 5:38:57 PM
6ead034862332c7e: Monday, April 12, 2010 3:31:35 PM
626273d7f6524841: Thursday, April 1, 2010 10:21:07 AM
1afe35581aabfb35: Friday, March 26, 2010 2:47:44 PM
0ae8a815bd7b56b1: Friday, March 26, 2010 2:43:01 PM
e666b5fd74e25542: Wednesday, March 24, 2010 11:58:07 AM
23b26605aef32bfa: Tuesday, March 23, 2010 1:56:28 PM
Log Messages
[3848:5996:20944865:ERROR:gpu_info_collector_win.cc(91)] : Can't retrieve a valid WinSAT assessment.
Aug 28, 2012
#4 pkasting@chromium.org
Also, clicking/dragging on the web page when it's in this state will do things like follow links (if clicked on), but won't cause the page to redraw.
Aug 28, 2012
#5 vange...@google.com
The flash you're seeing on tab switching is most likely  issue 143854  .  There is a temporary fix that was checked in at r153517. Would you mind trying a continuous build to see if it still repros? 
Aug 28, 2012
#6 pkasting@chromium.org
Is the primary issue described here a different one, then?

It seems as if even when things do redraw after navigating, they do so slightly later than I expect, almost as if we're dropping the first invalidation on the floor, or not invalidating for a while or something.  If that were true, the flash could be explained as a related case to the doesn't-redraw-at-all case in that in both cases we don't repaint initially like we should, but in the flashing case, we at least then do repaint after a brief period.  But I'm speculating wildly.
Aug 31, 2012
#7 vangelis@chromium.org
(No comment was entered for this change.)
Cc: jbauman@chromium.org
Aug 31, 2012
#8 pkasting@chromium.org
Still happening on a ludicrously high percentage of my page loads.

If there are about:flags switches I should try disabling, let me know.
Aug 31, 2012
#9 jbauman@chromium.org
Could the rwhvw window actually be the size of that smaller window? You can tell by using spy++. If that's the case the size mismatch would cause us to avoid drawing the contents, though I don't know what would cause that in the first place. A recording from about:tracing while you're experiencing the issue might help, though I don't know if it would be feasible to get that. Maybe if you had another window open to go to about:tracing.
Aug 31, 2012
#10 jbauman@chromium.org
Also, if you want a usable computer disabling "GPU compositing on all pages" might help. You could instead try putting in --disable-image-transport-surface on the command line to see if that helps at all. 
Aug 31, 2012
#11 pkasting@chromium.org
Tried --disable-image-transport-surface.  If anything that made things worse -- before, if a page looked weird, switching tabs away and back would force it to redraw.  With this switch, contents tend to lay out at the restored size and not fix themselves until you actually restore and remaximize the window.

With this switch in place I also tried checking the window size with spy++.  The window size seems to correctly be the large (near-fullscreen) size.  Interestingly, dragging the window-finder cursor over the content area immediately causes the content area contents to be wrong (near-empty) even if they were drawing correctly until then.

Attaching a trace I took -- still with this flag on -- of me switching to a glitchy-looking tab, waiting, switching away and then back again and waiting another second or two.

Next I'll try disabling "GPU compositing on all pages" (currently set to "Default").
trace
8.8 MB   Download
Aug 31, 2012
#12 pkasting@chromium.org
Removing that flag and disabling "GPU compositing on all pages" seems to be having a somewhat positive effect.  I still have seen some "content initially too small, flashes quickly and resizes to correct size" but that could be the same thing that we've had in Chrome forever if the page is unable to redraw within our timeout on tab switch and we have to show a stale buffer.
Aug 31, 2012
#13 jbauman@chromium.org
Could you get a trace with --disable-image-transport-surface off and  "GPU compositing on all pages" on? It will hopefully be more illuminating.
Aug 31, 2012
#14 pkasting@chromium.org
Done. Attached is a trace of:

* Focused a window
* Ctrl-shift-b to open bookmark bar
* Middle-clicked bookmark to codesearch
* Switched to codesearch tab, observed it draw as a blank white page
trace2
5.4 MB   Download
Sep 1, 2012
#15 jbauman@chromium.org
I see that some of the calls to RenderWidgetHostImpl::GetBackingStore have a size of 732x757, which seems to be approximately the size of that black frame. I don't know how that could happen, especially if Spy++ is correct.
Sep 2, 2012
#16 phistuck
Can you, please, add (more?) testing for this issue?
This comes up every two major version or so...

Thank you!
Sep 4, 2012
#17 vangelis@chromium.org
Peter, are you using a chrome version that was cut after r153517 ? The spy++ weirdness seems to have been an interaction with a DWM call we were making that was removed at that revision. I'm just curious if that would be related to what you're seeing.
Sep 4, 2012
#18 pkasting@chromium.org
In comment 3, I was using "Chrome version 23.0.1246.0 (Official Build 153452) dev-m", i.e. before that change.

My current build is "23.0.1251.2 (Official Build 154299) dev-m", so after.  I'm not sure precisely where in the process of this bug the update occurred, so I don't know whether comment 11 was tested before or after (but since it seems to say that spy++ looks normal, I would guess after; Omaha says Aug 31 is the release date for the current Chrome version).
Sep 4, 2012
#19 vangelis@chromium.org
Thanks. It was actually the window rapidly changing its contents once the spy++ target was on it that I believe was "fixed" in 153517 .  Regardless though, it does sound like you are still experiencing the failing update issue.  You're on Windows Vista, right? 
Sep 4, 2012
#20 pkasting@chromium.org
Yes, Vista x64.
Sep 6, 2012
#21 kar...@google.com
(No comment was entered for this change.)
Status: Assigned
Owner: vangelis@chromium.org
Sep 6, 2012
#22 vangelis@chromium.org
Peter, do you have any extensions installed? 
Sep 6, 2012
#23 pkasting@chromium.org
Nothing meaningful; the extensions page only shows "Google Corporate Extension Reporter".

I'm happy to hand over my seat if you need to debug in person.
Oct 12, 2012
#24 pkasting@chromium.org
This has gotten worse.  Now if I don't force the flag to "Disabled", YouTube pages like http://www.youtube.com/watch?v=VGSZqkUbIZY#t=1m45s render the whole content area black, and don't get fixed by switching tabs away and back; I have to resize the window to force a redraw.

What can I do to help get this debugged?  I can't be the only user having major problems.
Labels: -Pri-2 Pri-1
Oct 12, 2012
#25 vangelis@chromium.org
John, can you please look at this issue? Peter is on our floor. 
Owner: jbauman@chromium.org
Cc: vangelis@chromium.org
Oct 19, 2012
#26 pkasting@chromium.org
I have users reporting this to me as well.
Oct 19, 2012
#27 pkasting@chromium.org
Attached per vangelis
tab_switch_trace.zip
695 KB   Download
Oct 19, 2012
#28 vangelis@chromium.org
We updated to the latest NVIDIA drivers on pkasting's box (the old ones were circa 2009) and on the first run after the install and reboot the problem was still there.  A little while later, it stopped happening!  Tab switching still exhibits some jank but we were no longer getting tabs that wouldn't redraw. 

Not entirely sure how to proceed from here. I'll leave this bug open for a bit longer to see if we can get other folks who are able to repro.


Status: Available
Owner: ---
Labels: -Pri-1 Pri-2
Oct 19, 2012
#29 darthand...@gmail.com
I just want to chime in and say that I have this issue and have had it for a while.

Google Chrome Version 24.0.1297.0 dev-m
Win 7 x64 SP1
nVidia GTX580 driver v. 304.79

On top of the redraw issue, the flicker of stale buffer when switching tabs (most often a tab I recently closed), I'm also having low refresh rate issues. Page scrolling and other page updates (mouse hover effects) will be jerky, between 3-7 fps if I had to guess. Restoring the window from a maximized state to a smaller size fixes the refresh rate issue as well as any redraw issues, at least until I maximize the window again.

I'm going to see if I can repro it in chromium after updating sources and building.
Oct 19, 2012
#30 geed...@gmail.com
Experiencing this:

Graphics Feature Status
Canvas: Hardware accelerated
Compositing: Hardware accelerated
3D CSS: Hardware accelerated
CSS Animation: Accelerated
WebGL: Hardware accelerated
WebGL multisampling: Hardware accelerated
Flash 3D: Hardware accelerated
Flash Stage3D: Hardware accelerated
Texture Sharing: Hardware accelerated
undefined: Hardware accelerated
Version Information
Data exported	Fri Oct 19 2012 22:28:09 GMT-0400 (Eastern Daylight Time)
Chrome version	22.0.1229.94 (Official Build 161065) m
Operating system	Windows NT 6.0 SP2
Software rendering list version	2.10
ANGLE revision	1275
2D graphics backend	Skia
Performance Information
Graphics	0.0
Gaming	0.0
Overall	0.0
Driver Information
Initialization time	63
GPU0	VENDOR = 0x10de, DEVICE= 0x05e2
Optimus	false
AMD switchable	false
Driver vendor	NVIDIA
Driver version	9.18.13.623
Driver date	8-30-2012
Pixel shader version	3.0
Vertex shader version	3.0
GL version	
GL_VENDOR	
GL_RENDERER	
GL_VERSION	
GL_EXTENSIONS	
Diagnostics
0
b3DAccelerationEnabled	true
b3DAccelerationExists	true
bAGPEnabled	true
bAGPExistenceValid	true
bAGPExists	true
bCanRenderWindow	true
bDDAccelerationEnabled	true
bDriverBeta	false
bDriverDebug	false
bDriverSigned	false
bDriverSignedValid	false
bNoHardware	false
dwBpp	32
dwDDIVersion	10
dwHeight	1080
dwRefreshRate	60
dwWHQLLevel	0
dwWidth	1920
iAdapter	0
lDriverSize	18229096
lMiniVddSize	0
szAGPStatusEnglish	Enabled
szAGPStatusLocalized	Enabled
szBGRASupported	true
szChipType	GeForce GTX 260
szD3DStatusEnglish	Enabled
szD3DStatusLocalized	Enabled
szDACType	Integrated RAMDAC
szDDIVersionEnglish	10
szDDIVersionLocalized	10
szDDStatusEnglish	Enabled
szDDStatusLocalized	Enabled
szDXVAModes	ModeMPEG2_A ModeMPEG2_C ModeVC1_C ModeWMV9_C
szDescription	NVIDIA GeForce GTX 260
szDeviceId	0x05E2
szDeviceIdentifier	{D7B71E3E-46A2-11CF-9D59-58321CC2C435}
szDeviceName	\\.\DISPLAY1
szDisplayMemoryEnglish	3649 MB
szDisplayMemoryLocalized	3649 MB
szDisplayModeEnglish	1920 x 1080 (32 bit) (60Hz)
szDisplayModeLocalized	1920 x 1080 (32 bit) (60Hz)
szDriverAttributes	Final Retail
szDriverDateEnglish	8/30/2012 15:14:00
szDriverDateLocalized	8/30/2012 3:14:00 PM
szDriverLanguageEnglish	English
szDriverLanguageLocalized	English
szDriverName	nvd3dumx.dll,nvwgf2umx.dll,nvwgf2umx.dll,nvd3dum,nvwgf2um,nvwgf2um
szDriverSignDate	
szDriverVersion	9.18.0013.0623
szKeyDeviceID	Enum\PCI\VEN_10DE&DEV_05E2&SUBSYS_12553842&REV_A1
szKeyDeviceKey	\Registry\Machine\System\CurrentControlSet\Control\Video\{D35AA5F6-82ED-4A5C-A798-DFEA1BFF03C2}\0000
szManufacturer	NVIDIA
szMiniVdd	n/a
szMiniVddDateEnglish	n/a
szMiniVddDateLocalized	n/a
szMonitorMaxRes	
szMonitorName	Generic PnP Monitor
szNotesEnglish	No problems found.
szNotesLocalized	No problems found.
szRegHelpText	
szRevision	
szRevisionId	0x00A1
szSubSysId	0x12553842
szTestResultD3D7English	Not run
szTestResultD3D7Localized	Not run
szTestResultD3D8English	Not run
szTestResultD3D8Localized	Not run
szTestResultD3D9English	Not run
szTestResultD3D9Localized	Not run
szTestResultDDEnglish	Not run
szTestResultDDLocalized	Not run
szVdd	n/a
szVendorId	0x10DE
Log Messages
[5736:6312:43571313:ERROR:gpu_info_collector_win.cc(93)] : Can't retrieve a valid WinSAT assessment.

And fixed by: Go into about:flags and change "GPU compositing on all pages" to "Disabled"
Oct 20, 2012
#31 pkasting@chromium.org
Vangelis, are there steps you can give the folks in comments 29/30 to collect more useful data?
Owner: vangelis@chromium.org
Cc: -vangelis@chromium.org
Labels: -Pri-2 Pri-1
Oct 20, 2012
#32 darthand...@gmail.com
Setting "GPU compositing on all pages" to "Disabled" mitigates, but does not fix the issue for me. Before, the page would not redraw at all. Now, elements which the the browser considers "Drawn" do not redraw (say, background, or switching to another tab), but mouse hover effects and other updates will cause affected sections of the page to redraw. Scrolling does not redraw properly, and leaves visual garbage reminiscent of dragging a frozen window around on the screen circa Windows 95.

I feel it important to note that I only seem to have these issues when my GPU is significantly stressed. For example:
 * VLC full-screen playback hardware-accelerated 1080p video on monitor #1
 * Skype video conference doing screen sharing (of the VLC window)
 * World of Warcraft windowed-borderless on monitor #2
 * Chrome maximized on monitor #3

Also, disabling GPU accelration on all pages has also increased the incident of crashes. With it set to default, chrome was much more likely to recover from the issues after a period of time.
Oct 22, 2012
#33 darthand...@gmail.com
I can repro in debug build of "Version 24.0.1304.0 (163201)". Let me know if I need to change to a specific revision and what information I can provide or dig into.
Oct 22, 2012
#34 vange...@google.com
darthandroid@ I suspect you're running into driver issues, possibly associated with low VRAM situations.  Your drivers seem to be pretty recent so I doubt that a driver update (if there were one available) would be of any help.  This is a pretty beefy GPU with presumably lots of VRAM but then again you seem to be running some GPU-heavy apps on 3 monitors.  Even so, I'm somewhat surprised to hear it's flaking out like that. If this is getting in the way, my advice would be to run chrome with --disable-accelerated-compositing (it's a command like flag, not available in the about:flags page). 

geedeck@ which of the symptoms are you experiencing? 

Cc: apatr...@chromium.org
Oct 23, 2012
#35 darthand...@gmail.com
It was happening with less than 33% of GFX mem utilized (913MiB/3GiB according to ProcessExplorer, just Skype Screen Sharing + VLC this time as things I would list as GPU-intensive; System memory is 21Gib/32Gib), though I agree it does sound a lot like low vmem issues; I've had problems in the past with Windows complaining that Aero had run out of memory, so I'm wondering if there's a separate compositing buffer somewhere with a smaller limit (Both System and GPU memory were reading well under their maximums when this occurs, though DWM.exe was chewing through close to 2GiB of system memory at the time).

I'm going to see if I can correlate the issue with Skype's Screen Sharing. If it's not screen sharing or VLC, then it may just be the fact that I have 49 windows open on my system.

I'll test out that command switch and see if that resolves the issue for me; I'm content with that work around, and am pretty used to running into unique problems due to the way that I stress the system. Here's my about:gpu for reference: http://pastebin.com/TsyCs8fv

Let me know if anything further I can test or do.
Dec 5, 2012
#37 tornin...@gmail.com
Any update on this? Seems to get worse by each version (beta channel). :(

Vista x64, Chrome 24.0.1312.32 beta-m, AMD GFX, two monitors.
Jan 16, 2013
#38 vangelis@chromium.org
pkasting@ do you still have the vista box?  It looks like this is now happening to a lot more folks.  See issue 169848 .
Jan 16, 2013
#39 darthand...@gmail.com
torninger@ Nothing on my end; I disabled all hardware acceleration in Chrome and the issue went away. I've just been living with this unless someone has something I need to test or debug to help find the cause.
Jan 16, 2013
#40 vange...@google.com
@darthandroid : Any chance we could get you to run the latest stable channel chrome (Chrome 24) with a clean profile and follow the steps from:

https://code.google.com/p/chromium/issues/detail?id=169848#c9

to grab a trace? Feel free to follow the action on that other bug. Thanks in advance!
Jan 16, 2013
#41 vange...@google.com
(No comment was entered for this change.)
Status: Duplicate
Mergedinto: 169848
Jan 16, 2013
#42 pkasting@chromium.org
I do have the box still, although I haven't used it in quite a while.
Jan 16, 2013
#43 vangelis@chromium.org
@pkasting: Mind firing it up and seeing if you can still repro the problem?  Thanks!
Jan 16, 2013
#44 wiltz...@chromium.org
@pkasting we're happy to do this for you, too, if you wouldn't mind letting us borrow the machine.
Jan 16, 2013
#45 pkasting@chromium.org
You're welcome to, and probably should, since I likely won't be physically in the office for another 2 weeks.

It's sitting on the floor of my office, out in the open, right next to my Z620.  I think it's plugged into my KVM and such but feel free to just unplug all that and steal the machine.
Mar 10, 2013
#46 bugdro...@chromium.org
(No comment was entered for this change.)
Labels: -Area-WebKit -Feature-GPU Cr-Content Cr-Internals-GPU
Apr 5, 2013
#47 bugdro...@chromium.org
(No comment was entered for this change.)
Labels: -Cr-Content Cr-Blink
Sign in to add a comment

Powered by Google Project Hosting