Bug 97561

Summary:	Regression: Issue with ATI HD 6570 (possibly others) HDMI output Purple Line
Product:	Drivers	Reporter:	Rich (forums0)
Component:	Video(DRI - non Intel)	Assignee:	drivers_video-dri
Status:	RESOLVED PATCH_ALREADY_AVAILABLE
Severity:	normal	CC:	alexdeucher, susah.yak
Priority:	P1
Hardware:	x86-64
OS:	Linux
Kernel Version:	4.0 and above	Subsystem:
Regression:	No	Bisected commit-id:
Attachments:	dmesg output with drm.debug=0xe dmesg output with drm.debug=0xe grepped with drm Output of xrandr --verbose

Description Rich 2015-04-30 08:26:42 UTC

Hi,

First ever kernel bug report so apologies for possibly getting it wrong. Both Linux version 4.0 and 4.1-rc now display a vertical purple line on the far left hand side of the screen. This only happens when using HDMI output and seems to be related to HDMI audio.

Happens with MESA 10.5 and MESA 10.6 devel, happens on both Arch Linux and Linux Mint (using a mainline kernel). Problem does NOT happen on Linux 3.19.3 and earlier.


Image of problem:

https://lh3.googleusercontent.com/-i_dXYiwJNBE/VUE1pbSid4I/AAAAAAAAEus/uAfOCER4PSE/s912/IMG_20150429_203014.jpg


Very similar bug: 

https://bugzilla.kernel.org/show_bug.cgi?id=43154 (resolved, but appears to be happening again).


lspci:

01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Turks PRO [Radeon HD 6570/7570/8550]


Giving this command resolves the problem - hence I think it is related to the audio output part of the dri driver and is very similar to the bug linked too above:

xrandr --output HDMI-0 --auto --set audio off


I am happy to provide logs of anything - please let me know what you need. I don't want to take a wild guess and provide everything.

Thanks !

Comment 1 Rich 2015-05-01 07:50:45 UTC

Additional information:
My HDMI screen is not audio enabled. So audio should not be enabled by default when auto-detect is being used.

Adding: xrandr.log
Output of: xrandr --verbose

Adding: dmesg_debug.log
Output of dmesg with kernel option drm.debug=0xe

Adding: dmesg_debug_drm.log
Output of dmesg with kernel option drm.debug=0xe grepped with drm

Comment 2 Rich 2015-05-01 07:51:25 UTC

Created attachment 175421 [details]
dmesg output with drm.debug=0xe

Comment 3 Rich 2015-05-01 07:51:46 UTC

Created attachment 175431 [details]
dmesg output with drm.debug=0xe grepped with drm

Comment 4 Rich 2015-05-01 07:52:08 UTC

Created attachment 175441 [details]
Output of xrandr --verbose

Comment 5 Alex Deucher 2015-05-01 13:55:12 UTC

Does it work correctly with Dave's drm-fixes tree:
http://cgit.freedesktop.org/~airlied/linux/log/?h=drm-fixes
There are a number of fixes related to this.

Comment 6 Rich 2015-05-01 16:04:08 UTC

Looks very similar, I am willing to bet that will fix it. I will have a go at compiling the kernel from source with those patches applied (never done it before so might take me a while).

Are those fixes going to make 4.1-stable?

Comment 7 Alex Deucher 2015-05-01 16:26:04 UTC

(In reply to Rich from comment #6)
> Looks very similar, I am willing to bet that will fix it. I will have a go
> at compiling the kernel from source with those patches applied (never done
> it before so might take me a while).
> 
> Are those fixes going to make 4.1-stable?

Yes, 4.1 and previous stable kernels.

Comment 8 Rich 2015-05-06 20:16:15 UTC

Thanks - 

I did try to compile my own kernel a few times but failed. I managed to 'git clone' the kernel source, apply the drm-fixes patches successfully, compile (with various config settings) but could never get the kernel to boot - but thats my problem and need to learn!

On topic:
4.1-rc2 [4.1.0-040100rc2-generic] fixes the problem. I assume its the same bug as you posted.

I am going to resolve this now (if I can!)
Thanks!