Bug 46525

Summary:

Xrandr broken on Evergreen chips

Product:

xorg

Reporter:

Dejan Noveski <dn>

Component:

Driver/Radeon

Assignee:

xf86-video-ati maintainers <xorg-driver-ati>

Status:

RESOLVED NOTABUG

QA Contact:

Xorg Project Team <xorg-team>

Severity:

major

Priority:

medium

CC:

Version:

7.7 (2012.06)

Hardware:

x86-64 (AMD64)

OS:

Linux (All)

Whiteboard:

i915 platform:

i915 features:

Attachments:

Description	Flags
Xorg.0.log	none
dmesg	none
Screenshot	none
Camera snapshot of the desktop	none

Description Dejan Noveski 2012-02-23 07:58:20 UTC

Created attachment 57533 [details] [review]
Xorg.0.log

Xrandr is broken on Evergreen chips. When in dual-head mode, side by side, the desktop is skewed and streched in the middle of both monitors. This doesn't happen if the monitors are positioned one above another. Tested on Juniper and Cedar chips, both act the same. This happens using a git snapshot xf86-video-ati from 12th February(dunno the exact version).

Same thing happens with or w/o compositing, kde, openbox or plain twm.

One little detail tho - the mouse doesn't appear to be affected. It's displayed properly, and the clicks are registered where the dialogs/buttons/desktop should be.

The monitors are 1: 1920x1200 2: 1680x1050.

Comment 1 Dejan Noveski 2012-02-23 07:59:39 UTC

Created attachment 57534 [details] [review]
dmesg

Comment 2 Alex Deucher 2012-02-23 08:03:50 UTC

Can you attach a picture of the problem?

Comment 3 Dejan Noveski 2012-02-23 08:14:18 UTC

Created attachment 57535 [details]
Screenshot

Screenshot shows normal desktop - taken with imagemagic's import -window.

I'll take a snapshot with my phone and post how it looks to me.

Comment 4 Dejan Noveski 2012-02-23 08:20:32 UTC

Created attachment 57536 [details]
Camera snapshot of the desktop

This is taken with my phone (pardon the poor quality) but you can definitely see what's going on. Every 10? pixels, the vertical is pushed few pixels to the right and the whole desktop is skewed. This is taken on the Juniper, but same thing happened on the Cedar with different resolutions and monitors.

Comment 5 Michel Dänzer 2012-02-23 09:16:14 UTC

> This happens using a git snapshot xf86-video-ati from 12th February(dunno the
> exact version).

Does your drm Git snapshot have commit 9b3ad51ae5fd9654df8ef75de845a519015150bb ('radeon: fix pitch alignment for scanout buffer')?

Comment 6 Dejan Noveski 2012-02-23 09:29:33 UTC

Can't say for sure. I got the packages from a semi-official radeon repo on archlinux. The owner does git snapshots once in a while, and all I know is, this build is made on Feb 12th 2012.

I will try with a more recent build and post the results here.

Comment 7 Dejan Noveski 2012-02-23 10:29:05 UTC

(In reply to comment #5)
> > This happens using a git snapshot xf86-video-ati from 12th February(dunno the
> > exact version).
> 
> Does your drm Git snapshot have commit 9b3ad51ae5fd9654df8ef75de845a519015150bb
> ('radeon: fix pitch alignment for scanout buffer')?

Ok. So I checked out xf86-video-ati master branch and I replaced only the driver, it's the same. So it's not the driver. I'll test out with latest mesa from git. Building them both from source.

Comment 8 Dejan Noveski 2012-02-23 13:49:19 UTC

(In reply to comment #7)
> Ok. So I checked out xf86-video-ati master branch and I replaced only the
> driver, it's the same. So it's not the driver.

That came out wrong. Might be the driver. This regression might be introduced during the middle of february.

Comment 9 Michel Dänzer 2012-02-24 02:05:17 UTC

(In reply to comment #5)
> Does your drm Git snapshot have commit 9b3ad51ae5fd9654df8ef75de845a519015150bb
> ('radeon: fix pitch alignment for scanout buffer')?

Note that this fix wasn't in the X or Mesa driver but in libdrm_radeon. So, I guess I should have asked: Which version / Git snapshot of libdrm are you using?

Comment 10 Dejan Noveski 2012-02-24 02:26:52 UTC

(In reply to comment #9)
> (In reply to comment #5)
> > Does your drm Git snapshot have commit 9b3ad51ae5fd9654df8ef75de845a519015150bb
> > ('radeon: fix pitch alignment for scanout buffer')?
> 
> Note that this fix wasn't in the X or Mesa driver but in libdrm_radeon. So, I
> guess I should have asked: Which version / Git snapshot of libdrm are you
> using?

So, today, I tested with yesterday's git snapshots of mesa and xf86-video-ati. The issue is resolved. Sorry for bugging you.

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.