Bug 866992

Summary:	GPU cannot resume from suspend/hibernate/vt-switch
Product:	[Fedora] Fedora	Reporter:	redhat
Component:	xorg-x11-drv-ati	Assignee:	X/OpenGL Maintenance List <xgl-maint>
Status:	CLOSED DUPLICATE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	high	Docs Contact:
Priority:	unspecified
Version:	18	CC:	kevin.macpherson, matteo, xgl-maint
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2013-01-21 12:43:17 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description redhat 2012-10-16 13:52:18 UTC

Description of problem:
When resuming from suspend or hibernate or switching back to x after switching vt. Screen flickers in a loop.

Version-Release number of selected component (if applicable):
7.0.0 - 0.7.20121015gitbd9e2c064.fc18 (but has been there before!)

How reproducible:
nearly everytime (it worked a few times from those many tests i did, but same settings did not work just a minute later, so it was kind of random)


Steps to Reproduce:
1. Suspend (via Gnome or pm-suspend)
2. Resume
  
Actual results:
Screen flickers, unuseable!

Expected results:
Resume...

Additional info:
Mobility Radeon HD 4200
Smolt: https://bugzilla.redhat.com/attachment.cgi?id=628097

Comment 1 redhat 2012-10-16 13:54:01 UTC

Error messages (dmesg):

Sep 26 22:54:10 localhost kernel: [ 1252.770614] radeon 0000:01:05.0: GPU reset 
succeeded, trying to resume
Sep 26 22:54:10 localhost kernel: [ 1252.783900] [drm] PCIE GART of 512M enabled
 (table at 0x00000000C0040000).
Sep 26 22:54:10 localhost kernel: [ 1252.784482] radeon 0000:01:05.0: WB enabled
Sep 26 22:54:10 localhost kernel: [ 1252.784486] radeon 0000:01:05.0: fence driv
er on ring 0 use gpu addr 0x00000000a0000c00 and cpu addr 0xffff88022a200c00
Sep 26 22:54:10 localhost kernel: [ 1252.817338] [drm] ring test on 0 succeeded 
in 0 usecs
Sep 26 22:54:12 localhost kernel: [ 1254.101427] [drm] ib test on ring 0 succeed
ed in 0 usecs
Sep 26 22:54:12 localhost kernel: [ 1254.103015] [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -35!
Sep 26 22:54:12 localhost kernel: [ 1254.105886] radeon 0000:01:05.0: GPU reset succeeded, trying to resume

Comment 2 redhat 2012-10-16 16:49:51 UTC

This seems to be kernel related. I tried kernel from fedora 17 install media and it worked!

Comment 3 Matteo Settenvini 2012-10-18 00:05:32 UTC

I can confirm the same issue on a Thinkpad X121e with a: 

00:01.0 VGA compatible controller: Advanced Micro Devices [AMD] nee ATI Wrestler [Radeon HD 6310] (prog-if 00 [VGA controller])

Kernel driver in use: radeon

Also for me, it's a couple of weeks at least I see this issue. Trying to switch to another console does not work. The only "solution" I have found is holding down the power button and restarting.

The kernel from Fedora 17 is okay for me, too. Oddly enough, trying to open gnome-terminal and using "pm-suspend" does not result in a problem on resume, while closing the laptop lid and waiting for automatic suspension does.

Comment 4 redhat 2012-10-28 01:39:39 UTC

seems fixed in 3.6.3-3.fc18.x86_64?

Comment 5 Matteo Settenvini 2012-10-28 16:40:38 UTC

Looks much better to me, but still not fixed. I can get this issue about one time in 20 with kernel 3.6.3-3. Probably a race condition is still present somewhere?

Comment 6 Kevin MacPherson 2012-12-31 11:43:13 UTC

I've just had this on a resume using 3.6.11-3 on a Samsung RV515-A01 using radeon on a 6310HD.

Comment 7 Matteo Settenvini 2013-01-11 17:37:45 UTC

Had gone away, but resurfaced with the latest 3-4 kernel revisions. Now running 3.7.1-5.fc18.x86_64, I hit this bug around ~50% of the times I resume from suspend. Only solution is an hard reset.

Comment 8 redhat 2013-01-21 12:43:17 UTC


*** This bug has been marked as a duplicate of bug 849347 ***