Bug 1014450

Summary: Graphical issues with Intel chip and Dell Precision 5510
Product: [openSUSE] openSUSE Distribution Reporter: Richard Schärtel <richard.schaertel>
Component: BasesystemAssignee: E-mail List <bnc-team-screening>
Status: RESOLVED FIXED QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None CC: mstaudt, richard.schaertel, sndirsch, tiwai
Version: Leap 42.2   
Target Milestone: ---   
Hardware: Other   
OS: Other   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: dmesg
Xorg.0.log

Description Richard Schärtel 2016-12-08 08:10:45 UTC
I installed Leap 42.2 on Dell Pecision with Intel graphic card:
# lspci -nnk | grep VGA -A2
00:02.0 VGA compatible controller [0300]: Intel Corporation Skylake Integrated Graphics [8086:191b] (rev 06)
	Subsystem: Dell Device [1028:06e5]
	Kernel driver in use: i915

I have to black list i915 in /etc/modprobe.d/50-blacklist.conf in order to be able to use graphical mode (use Mate desktop on top of X Windows, no Gnome/KDE).

My graphical kernel modules look like this:
# lsmod | grep i915
i915                 1314816  3 
i2c_algo_bit           16384  2 i915,nouveau
drm_kms_helper        155648  2 i915,nouveau
drm                   393216  6 ttm,i915,drm_kms_helper,nouveau
video                  40960  4 i915,dell_wmi,nouveau,dell_laptop
button                 16384  2 i915,nouveau
dell5510risc:~ # lsmod | grep nouveau
nouveau              1597440  0 
mxm_wmi                16384  1 nouveau
ttm                   106496  1 nouveau
i2c_algo_bit           16384  2 i915,nouveau
drm_kms_helper        155648  2 i915,nouveau
drm                   393216  6 ttm,i915,drm_kms_helper,nouveau
wmi                    16384  4 dell_led,dell_wmi,mxm_wmi,nouveau
video                  40960  4 i915,dell_wmi,nouveau,dell_laptop
button                 16384  2 i915,nouveau

I cannot blacklist nouveau instead i915, because I will get a blinking cursor instead login screeen (not even tty's will work). 

I cannot add kernel parameters like nomodeset, etc., get blinking cursor or looping login screen.

I did mkinitrd after all testings.


Parts of glxinfo output:
OpenGL renderer string: Mesa DRI Intel(R) HD Graphics 530 (Skylake GT2) 
OpenGL version string: 3.0 Mesa 11.2.2
direct rendering: Yes

I executed:
# grep VGA || lspci | grep VGA | colrm 1 4 ; cat /proc/cpuinfo | \
> egrep "model name|MHz" ; glxinfo | \
> egrep -A2 "direct rendering|OpenGL vendor" ; uname -sr; glxgears & sleep 30 ; \
> killall glxgears 
2.0 VGA compatible controller: Intel Corporation Skylake Integrated Graphics (rev 06)
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 3299.107
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 800.065
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 800.063
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 800.021
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 800.010
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 1100.789
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 3204.970
model name	: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
cpu MHz		: 799.999
direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.4
--
OpenGL vendor string: Intel Open Source Technology Center
OpenGL renderer string: Mesa DRI Intel(R) HD Graphics 530 (Skylake GT2) 
OpenGL core profile version string: 3.3 (Core Profile) Mesa 11.2.2
Linux 4.4.27-2-default
[1] 4853
Running synchronized to the vertical refresh.  The framerate should be
approximately the same as the monitor refresh rate.
302 frames in 5.0 seconds = 60.331 FPS
300 frames in 5.0 seconds = 59.927 FPS
300 frames in 5.0 seconds = 59.934 FPS
300 frames in 5.0 seconds = 59.936 FPS
300 frames in 5.0 seconds = 59.934 FPS
[1]+  Terminated              glxgears

I guess here is a bug in the Intel i915 module, since I shouldn't need to blacklist it, right?
What impact does it have to blacklist i915? Not sure if you can see with the tests if all works fine in my setup, also support with 3D? 

Why do I need nouveau?
Are there Intel drivers out there already for Leap42.2?

/Richard
Comment 1 Takashi Iwai 2016-12-09 09:44:09 UTC
First off, please try the kernel in OBS Kernel:openSUSE-42.2 repo, and blacklist nouveau instead of i915.  Nvidia chip is only for additional 3D rendering, and the whole graphics is drawn via Intel chip.  If it works without i915, it means you're running EFI framebuffer instead of the native graphics.

Also, you didn't install Nvidia driver beforehand, right?  Just to make sure.

If you still have an issue with the latest kernel, please give the full kernel messages (dmesg output) after boot with drm.debug=0x0e boot option.  Also, give the X log, e.g. in /var/log/Xorg.0.log.
Comment 2 Richard Schärtel 2016-12-10 22:53:38 UTC
Created attachment 705962 [details]
dmesg
Comment 3 Richard Schärtel 2016-12-10 23:03:27 UTC
Created attachment 705965 [details]
Xorg.0.log
Comment 4 Richard Schärtel 2016-12-10 23:14:25 UTC
I installed the latest kernel:
Linux dell5510risc.sto.suse.home 4.4.36-5-default #1 SMP Tue Dec 6 20:57:24 UTC 2016 (bd51250) x86_64 x86_64 x86_64 GNU/Linux

Nouveau is blacklisted and I did mkinitrd.
First time I was able to login without problems. I configured mate desktop, even compiz is working.

After first shutdown and new startup, I got blinking cursor. I change to tty1 and login. I do a init3 and back init5. 
Now it can happen that I end up in a login loop. Then I restart again. 
Then the login might work. But then I have to restart the network, because this is not working properly.
Then all works....

I added the kernel parameter as you suggested. Attached the dmesg and Xorg logs.

I haven't used nvidia in the tests in the beginning.
Then I just tried to install nvidia driver, even I have only an Intel graphic-card. But I got an error and it didn't work. I just removed all the nvidia stuff again.
Comment 5 Takashi Iwai 2016-12-11 08:09:04 UTC
Are the logs from the working case, or the non-working case?  We need the both cases.

Also, when you go to VT1, does the console appear normally?  If yes, stop X via init 3, and start X manually like "X -retro -ac" there.  Try to get the dmesg and X log from that time point, so that we can see the portion where the problem appears.

Last but not least: you wrote that the first try worked.  What could be the difference?  Does the cold boot behave differently from the warm boot?
Comment 9 Takashi Iwai 2017-01-12 07:56:09 UTC
This doesn't look like a kernel problem, but rather an infamous race at boot up.
It explains why the manual X test failed.

I've observed a similar issue on my HSW machine, too, but only occasionally.  And, in my case, it's XFCE desktop.

What display manager are you using for Mate?  Is it lightdm?  If so, it's as same as XFCE, at least, and it can be a suspect.  Or if it's gdm, we should see the issue on GNOME as well.

Adding Stefan to Cc.  We may test it locally as well.

Meanwhile, as a usual suspect, please try to disable plymouth.  The best would be completely uninstalling the stuff,
  # zypper rm -u plymouth

or you can boot with plymouth.enable=0 option to disable in runtime, too.
I guess the issue still persists, but let's make sure.

The lack of network might be a similar cause, but it's not sure.  Is the network setup via NetworkManager or wicked?  Anyways, you'd better split to two bug reports, one for the wrong X startup and another for the network issue.
Comment 10 Stefan Dirsch 2017-03-07 14:59:10 UTC
Richard, could you please provide feedback to commen#9?
Comment 12 Stefan Dirsch 2017-03-11 08:11:42 UTC
Hmm. Good question. We may have added additional kernel fixes meanwhile. Anyway, let's close this one as fixed.