Corrupted screen using radeon drivers

Sounds similar to bug 1195509. But that one has intel graphics.

Ok, so new day, new updates, new try, same results, but this time something interesting caught my eye in the unity-system-compositor.log:

"dm_connection_start
set_active_session '0'
set_active_session
Unable to set active session, unknown client name 0
Mesa 9.2.0-devel implementation error: unhandled format!

Please report at https://bugs.freedesktop.org/enter_bug.cgi?product=Mesa
Mesa 9.2.0-devel implementation error: unhandled format!"

Only other thing that changed is that i can no longer change virtual terminals, so had to use the power button to restart.

I captured the actual error message this time :)
===>
[ 52.558241] radeon 0000:01:00.0: r600_cs_track_validate_cb:419 cb height (1080, 0x10, 4) invalid
[ 52.558246] radeon 0000:01:00.0: r600_packet3_check:1720 invalid cmd stream 434
[ 52.558248] [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !

I've also attached excerpts from my dmesg that involved radeon or drm.?field.comment=I captured the actual error message this time :)
===>
[ 52.558241] radeon 0000:01:00.0: r600_cs_track_validate_cb:419 cb height (1080, 0x10, 4) invalid
[ 52.558246] radeon 0000:01:00.0: r600_packet3_check:1720 invalid cmd stream 434
[ 52.558248] [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !

I've also attached excerpts from my dmesg that involved radeon or drm. Once again, this only happens when Mir is on. Xorg is fine, tested out a 3D game (Nexuiz) in Xorg and no corruption either.

Marking as confirmed as it clearly affects other people.

With the latest Mir I get the screen corruption that I show in the photograph below.

@Simos:
Your corruption looks different from mine. Mine looks like a jumbled mess of whatever it's trying to display on the screen (it seems like an alignment issue.. I'll paste a screenshot at my next opportunity. Yours is like noise.

Can you drop into a VT (One of Ctrl+Alt+F[1-6] should work), log in, type dmesg, and see if you get the same error message as I am getting?

@Raymond Wells: Tried again today, and I believe I got a corruption that is similar to your case.

The attached picture shows one type of the corruption. Maybe the Dash is showing or something.
You can see the hardware cursor that is showing correctly and nicely.

This shows the screen corruption when I move the mouse a bit further away. This was what I would see more often.

The hardware mouse still shows well.

These are the errors found in kernel.log.
I would get about 40 such set of errors per second (for 02:17:37 the following four lines were repeated 41 times).
The graphics card is a Radeon HD 6320 (Evergreen, PALM)

Jul 11 02:17:37 laptop kernel: [13151.092735] radeon 0000:00:01.0: evergreen_surface_check_2d:277 cb pitch 1376 invalid must be aligned with 128
Jul 11 02:17:37 laptop kernel: [13151.092757] radeon 0000:00:01.0: evergreen_cs_track_validate_cb:426 cb[0] invalid (0x000000ab 0x0000407f 0x00190890 0x01088468)
Jul 11 02:17:37 laptop kernel: [13151.092768] radeon 0000:00:01.0: evergreen_packet3_check:1958 invalid cmd stream 465
Jul 11 02:17:37 laptop kernel: [13151.092777] [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !

according to Marten there is a fix in http://paste.debian.net/15562/ that should address the alignment issue. Chris, can you pls have a look at that?

@Olli: I tried the patch at http://paste.debian.net/15562/ and it works.
See attached image.

The patch on the pastebin doesn't handle 1D tiling. Attached is an updated patch which adds support for 1D tiling as well.

The attachment "Fix 1D and 2D tiling" seems to be a patch. If it isn't, please remove the "patch" flag from the attachment, remove the "patch" tag, and if you are a member of the ~ubuntu-reviewers, unsubscribe the team.

[This is an automated message performed by a Launchpad user owned by ~brian-murray, for any issues please contact him.]

I tried the patch at #12 (Alex Deucher) and it worked fine for me (Radeon HD 6320).

My testing procedure: https://plus.google.com/107558083186116705038/posts/61v9sEcoFX5

The patch works for me, also.

@Simos: In your install instructions the apt-get line didn't work, so I used dpkg -i xserver-xorg-video*deb to install the packages instead, and that seemed to work, although the dbg packages needed to pull in some additional deps, but it was fine.
$ sudo dpkg -i xserver-xorg-video*deb
$ sudo apt-get install -f

worked for me.

(@Raymond: Indeed, the command should have been 'dpkg -i ...'. I just fixed the G+ post. Thanks)

Any progress on getting this committed? And is there anything I can do to help it move forward? (I'm a C++ developer, but not familiar with Mir)

I'm an end user that is affected by this on my HD 4570 :(

At the moment I'm running in SOFTWARE mode because this open source driver doesn't work for me, and ATI's binary legacy driver is incompatible with the version of Xorg in ubuntu stable.

What can I do to help sort this out?

Seems to be fixed for me. Installed from the ppa and worked out-of-box without the patch.

Works for me too :). Still getting some weird errors for input, but that's another bug ^_^.

@sheridan: There's a PPA here that has the old fglrx drivers... be sure to heed his warning about Unity though as I did have a bit of issues getting unity running on my laptop after installing it: https://launchpad.net/~makson96/+archive/fglrx

... sry for the double post.. should have mentioned above that my card is the RadeonHD 2600XT 512MB model. And it is confirmed working. :)

Thank you for the help. I tried the legacy fglrx driver through the PPA route you posted. It the driver loads but crashes more unity and cinnamon.

I guess I need to wait until either the next release or maybe AMD will update it's legacy drivers soon. It's been 6 months so here's hoping it's soon.

I just filed bug 1210377, but I'm wondering if it might not be a duplicate of this bug. Can anyone take a look and tell me what they think?

Ian,

Bug 1210377 seems quite different (and not Mir-related).

This bug was fixed in the package xserver-xorg-video-ati - 1:7.2.0-0ubuntu3

---------------
xserver-xorg-video-ati (1:7.2.0-0ubuntu3) saucy; urgency=low

  * xmir.patch: Also support 1D tiling. (LP: #1195425)
 -- Christopher James Halse Rogers <email address hidden> Thu, 15 Aug 2013 10:31:16 +1000