Bug 79231

Summary: Corrupted display on HD 7730
Product: xorg Reporter: madcatx
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED DUPLICATE QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium CC: pali.rohar
Version: unspecified   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
xorg log
none
Xorg with NoAccel off
none
Xorg with NoAccel on
none
dmesg
none
dmesg with "Fix v2" applied
none
Xorg.log with "Fix v2" applied none

Description madcatx 2014-05-25 21:06:52 UTC
Created attachment 99804 [details]
xorg log

Ever since I got this card several months ago I've been having an issue where the sceen is completely messed up. The problem I'm having seems to be identical to that discussed here (https://bbs.archlinux.org/viewtopic.php?id=180541), except that I'm on up to date Fedora 20.

kernel 3.14.4
xorg 1.14.4
xorg-x11-drv-ati
   Version     : 7.2.0
   Release     : 3.20131101git3b38701.fc2
xorg-x11-glamor
   Version     : 0.5.1
   Release     : 3.20140115gitfb4d046c.fc20
mesa 10.1.3

The only way to get rid of these artifacts is to use "RenderAccel" "EXA" or "NoAccel" "yes" in xorg config.

I've tried various kernel parameters to debug this but I haven't found a way how to get the card working properly expect for disabling acceleration altogether.

03:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Cape Verde LE [Radeon HD 7730/8730] (prog-if 00 [VGA controller])
	Subsystem: PC Partner Limited / Sapphire Technology Device e228
	Flags: bus master, fast devsel, latency 0, IRQ 24
	Memory at d0000000 (64-bit, prefetchable) [size=256M]
	Memory at fbc80000 (64-bit, non-prefetchable) [size=256K]
	I/O ports at c000 [size=256]
	Expansion ROM at fbcc0000 [disabled] [size=128K]
	Capabilities: [48] Vendor Specific Information: Len=08 <?>
	Capabilities: [50] Power Management version 3
	Capabilities: [58] Express Legacy Endpoint, MSI 00
	Capabilities: [a0] MSI: Enable- Count=1/1 Maskable- 64bit+
	Capabilities: [100] Vendor Specific Information: ID=0001 Rev=1 Len=010 <?>
	Capabilities: [150] Advanced Error Reporting
	Capabilities: [270] #19
	Kernel driver in use: radeon
	Kernel modules: radeon

Is there anything I can do to help fix this?
Comment 1 madcatx 2014-05-26 16:45:09 UTC
I tried to disable color tiling and color tiling 2D as a suggested workaround for a similar issue but it does not seem to help in this case.
Comment 2 Alex Deucher 2014-05-27 14:26:59 UTC
Please attach your dmesg output and your xorg log with acceleration enabled.
Comment 3 madcatx 2014-05-27 14:40:36 UTC
Created attachment 99952 [details]
Xorg with NoAccel off
Comment 4 madcatx 2014-05-27 14:41:08 UTC
Created attachment 99953 [details]
Xorg with NoAccel on
Comment 5 madcatx 2014-05-27 14:42:07 UTC
Created attachment 99955 [details]
dmesg
Comment 6 Michel Dänzer 2014-05-28 02:46:51 UTC
(In reply to comment #6)
> 03:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI]
> Cape Verde LE [Radeon HD 7730/8730] (prog-if 00 [VGA controller])

'LE' sounds like a harvested card; we're probably trying to use hardware blocks which didn't pass validation.

Bug 60879 has some attempts to address this kind of issue.
Comment 7 madcatx 2014-05-28 08:20:07 UTC
Okay, I'll keep an eye on that report. Thanks for letting me know.

On Wed May 28 2014 04:46:51 GMT+0200 (CEST), bugzilla-daemon@freedesktop.org wrote:
> https://bugs.freedesktop.org/show_bug.cgi?id=79231
> 
> --- Comment #6 from Michel Dänzer <michel@daenzer.net> ---
> (In reply to comment #6)
> > 03:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI]
> > Cape Verde LE [Radeon HD 7730/8730] (prog-if 00 [VGA controller])
> 
> 'LE' sounds like a harvested card; we're probably trying to use hardware blocks
> which didn't pass validation.
> 
> Bug 60879 has some attempts to address this kind of issue.
> 
> -- 
> You are receiving this mail because:
> You reported the bug.
>
Comment 8 Alex 2014-06-24 12:12:51 UTC
I had similar issues with the Gigabyte HD7777, everything was pixelated, and after a hint, that it could be a recycled card, i switched to another same card (had crossfire setup), and it worked! 

Also to note, that the catalyst driver works fine on that card.
Comment 9 madcatx 2014-07-10 17:56:44 UTC
Because I'm still having this problem in 3.15.3, is there *anything* I can try to find out which HW blocks are broken and how to disable them? I have no problem messing with the kernel code on my own but I have only a faint idea where to start. GNOME Shell is quite a pain to use over llvmpipe...
Comment 10 Alex Deucher 2014-07-10 18:35:17 UTC
(In reply to comment #9)
> Because I'm still having this problem in 3.15.3, is there *anything* I can
> try to find out which HW blocks are broken and how to disable them? I have
> no problem messing with the kernel code on my own but I have only a faint
> idea where to start. GNOME Shell is quite a pain to use over llvmpipe...

Did the patches in bug 60879 help?
Comment 11 madcatx 2014-07-10 20:30:44 UTC
Unfortunately they didn't. The only patch that appears to contain a fix for Cape Verde is the "Fix v2" patch for mesa. I rebuilt mesa 10.1.5 (F20's latest) with the patch applied and installed "mesa-dri-drivers" package - as I understand it the patch doesn't affect anything outside this package, right?

Attempt to start X with the patched mesa and acceleration enabled leaves me with black screen and the only way to get any debugging info is over SSH. Obtained dmesg dump and Xorg.log is attached.
Comment 12 madcatx 2014-07-10 20:31:24 UTC
Created attachment 102575 [details]
dmesg with "Fix v2" applied
Comment 13 madcatx 2014-07-10 20:31:52 UTC
Created attachment 102576 [details]
Xorg.log with "Fix v2" applied
Comment 14 Pali Rohár 2014-08-11 16:03:38 UTC
my 7730 is working now with modified patch, see: https://bugs.freedesktop.org/show_bug.cgi?id=60879#c84
Comment 15 Alex Deucher 2014-08-11 16:06:24 UTC

*** This bug has been marked as a duplicate of bug 60879 ***

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.