Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Already on GitHub? Sign in to your account

GUI distortions after resume on 3.1.2 #2453

Closed
Frank-NL opened this Issue Jul 9, 2013 · 31 comments

Comments

Projects
None yet
4 participants
Contributor

Frank-NL commented Jul 9, 2013

Tested using OE 3.2.1 Intel x64.

Using either Aeon Nox or Xperience1080, after resuming from suspend there are heavy distortions in parts of the GUI. Two photo's here:

http://i.imgur.com/4JrRGVt.jpg
http://i.imgur.com/ktxJ9uG.jpg

When I kill xbmc.bin and xbmc restarts the GUI is OK. Also by switching screens or going back to the home screen the distortions can disappear.

Does NOT happen with Confluence.

A debug log of starting, suspending and resuming here: http://sprunge.us/GCdU

According to 'Black, the Xperience1080 dev this could be the Intel driver.

Using 3.1.1 and previous builds, the distortions were never there after resuming at least a dozen times.

Contributor

fritsch commented Jul 9, 2013

Try the reload skin command after resume.

Contributor

Frank-NL commented Jul 9, 2013

You mean once manually or is there a setting for that?
Do you think the driver is to blame here?

Contributor

fritsch commented Jul 9, 2013

Driver / Kernel / whatever, yes.

http://wiki.xbmc.org/index.php?title=Skinning <- 12 does this "workaround it"?

Contributor

Frank-NL commented Jul 9, 2013

I could not reproduce it on my system after suspending many times. Weird.

After a cold boot and the second (first one was OK) resume the distortion is back. But it's only visible after selecting another item in the menu (so movement triggers it).
First skin reload: still artifacts. Even more!

Second reload: everything looks OK.

So it's not very consistent (which sucks when it comes to debugging).

Contributor

fritsch commented Jul 9, 2013

hehe :-)

Thanks for trying!

Contributor

Frank-NL commented Jul 9, 2013

So... Any idea how to fix this? Or how I can help?
For now I have added a custom sleep script in /storage/.config/sleep.d that restarts XBMC on resume.

Contributor

fritsch commented Jul 10, 2013

You can try to enable SNA acceleration with a simple xorg.conf and see if this workarounds it. In any case make sure, that you have "Vertical Blank Sync" on "Let Driver choose" which is the recommended setting for intel.

simple config could look like this:

Section "Device"
  Identifier  "Device0"
  Driver      "intel"
  VendorName  "INTEL Corporation"
  # Always use Let driver Choose in xbmc vertical blank sync on intel hardware
  Option "AccelMethod" "sna"
EndSection

Section "Extensions"
  # fixes tearing
  Option      "Composite"           "Disable"
EndSection
Contributor

Frank-NL commented Jul 10, 2013

Yeah the Let driver choose settings I remember from the thread at the XBMC forum :)
SNA does not help unfortunately... I still see the distortions.

Contributor

fritsch commented Jul 10, 2013

We will upgrade Intel driver next time, a new version comes out. Sadly we cannot upgrade xorg completely yet - cause of "other binary blobs"

Next idea, after resume:

chvt 1
chvt 7

e.g. switching terminals, does that help? I don't know the numbers on OE exactly, but try some :-)

Ps: Sorry for not remembering your name - i have a problem remembering names, so mostly fixate myself on the content.

Contributor

Frank-NL commented Jul 10, 2013

No, that doesn't help... Let's wait for the update in that case, the workaround is fine for me at the moment.
Could xorg that's used now be too old?

PS: No need to apologize, to refresh your memory this is the post I meant: http://forum.xbmc.org/showthread.php?tid=165707&pid=1433777#pid1433777

Contributor

fritsch commented Jul 10, 2013

Nope, I think this is just an issue in the latest intel driver - if you follow here: http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/log/ you see that a whole lot gets changed over the day. We also upgraded Mesa and stuff :-) Let's see what happens after next driver bump.

Contributor

fritsch commented Jul 11, 2013

Hi, i have talked with the intel guys in #intel-gfx they gave me the following bug report: https://bugzilla.kernel.org/show_bug.cgi?id=60530

Read carefully if it matches your issue and subscribe yourself to it.

Contributor

Frank-NL commented Jul 11, 2013

I subscribed. The distortions have similarities although it's hard to compare since XBMC is not a DE. The fact that restarting X solves it too is an indication this is the same issue.

Contributor

Frank-NL commented Jul 15, 2013

Not fixed in OE 3.1.3 so it looks to be the kernel issue as reported. In that ticket there's also a report that linux 3.10.1 did not solve this. So hopefully it will be fixed in the next minor linux version.

For now I will keep the restart on resume workaround.

Contributor

Frank-NL commented Jul 17, 2013

@fritsch Any chance you can make a quick build with this patch? https://bugzilla.kernel.org/show_bug.cgi?id=60530#c41

Contributor

fritsch commented Jul 18, 2013

Sorry, i am currently quite busy. Won't have time before sunday evening I
think.

2013/7/17 Frank van Klaveren notifications@github.com

@fritsch https://github.com/fritsch Any chance you can make a quick
build with this patch?
https://bugzilla.kernel.org/show_bug.cgi?id=60530#c41


Reply to this email directly or view it on GitHubhttps://github.com/OpenELEC/OpenELEC.tv/issues/2453#issuecomment-21115732
.

               Key-ID:     0x1A995A9B
               keyserver: pgp.mit.edu

Fingerprint: 4606 DA19 EC2E 9A0B 0157 C81B DA07 CF63 1A99 5A9B

Contributor

Frank-NL commented Jul 18, 2013

No problem, it can wait.

Contributor

fritsch commented Jul 20, 2013

http://cgit.freedesktop.org/~danvet/drm-intel/commit/?id=94a335dba34ff47cad3d6d0c29b452d43a1be3c8 It has now landed in drm-fixes, I am confident, that we get it perhaps with 3.10.2

Contributor

Frank-NL commented Jul 20, 2013

Good news. No need for a testing build at the moment!

Contributor

stefansaraev commented Jul 22, 2013

@fritsch should we try to add the patch at next update ? applies and compiles cleanly here

EDIT: patch included in 7c8e1bf
@Frank-NL I'd appreciate if you can test and report back. thanks in advance

Contributor

fritsch commented Jul 22, 2013

@stefansaraev: That would be really cool. I had hoped it comes in with 3.10.2 but it did not.

Contributor

Frank-NL commented Jul 22, 2013

Great! When 3.1.4 arrives I'll let you know. Thanks.

Contributor

fritsch commented Jul 22, 2013

@stefansaraev: Best is to let @Frank-NL test it prior to inclusion, so he needs an intel x64 build.

Contributor

stefansaraev commented Jul 22, 2013

@sraue can you make one from latest master? my buildserver is busy ATM.
EDIT: build started, but will take some time. maybe tomorrow..

Contributor

sraue commented Jul 22, 2013

will do over night, if its ok for @Frank-NL ?

Contributor

Frank-NL commented Jul 22, 2013

Tomorrow is fine :)

Contributor

Frank-NL commented Jul 23, 2013

I have suspended and resumed at least 6 times in a row without restarting XBMC. No distortions or artifacts visible on the GUI. The patch works!

Contributor

fritsch commented Jul 23, 2013

+1 for inclusion. Thanks very much for testing.

Contributor

stefansaraev commented Jul 23, 2013

thanks for testing @Frank-NL and for for the patch @fritsch

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment