[Bug 899879] New: Black screen during init phase of KDE desktop after installation
http://bugzilla.opensuse.org/show_bug.cgi?id=899879 Bug ID: 899879 Summary: Black screen during init phase of KDE desktop after installation Classification: openSUSE Product: openSUSE Factory Version: 13.2 Beta 1 Hardware: Other OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: Installation Assignee: yast2-maintainers@suse.de Reporter: freek@opensuse.org QA Contact: jsrain@suse.com Found By: --- Blocker: --- I did installations of both openSUSE 13.2 Beta1 and openSUSE Factory snapshot 20141004 on a brand new laptop with a Intel processor N3530 with Intel HD Graphics. The memory is 8 GB. Installation was OK but after first boot I got the log in screen, entered my credentials and got the KDE splash screen with the first icon, the disk, showing. After that the screen turns black. Pressing the power button shortly halts the system in a proper way. A reboot shows the log in screen again, but after log in I don't get the KDE splash again, but it blanks almost immediately. Booting again and pressing the Esc to see the boot log works OK, but I do not get the log in screen anymore. Booting again and pressing the Esc blanks the screen. Still pressing shortly the power button halts the system after a while. Booting again and pressing Esc shows nothing. Will install again but only a system up to text mode. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #1 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Gabriele Mohr
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #3 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #4 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #5 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #7 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #10 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #11 from Freek de Kruijf
I received an answer on the upstream bug report, see Comment 9. There is a request to test the mentioned fixes. However I do not have the knowledge to test this. Can you guide me or indicate that these fixes are implemented in 13.2 GM and/or Tumbleweed?
There is a further answer from Intel, or rather a request to test, from upstream. See https://bugs.freedesktop.org/show_bug.cgi?id=85823 I do not have enough knowledge to test and provide the requested info. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #12 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
--- Comment #13 from Freek de Kruijf
Sorry, building a commplete kernel and provide it as packages (as you would like it) isn't a trivial task. But testing with current kernel-of-the-day RPMs might already help. If required fixes landed upstream already.
I only need to know whether kernel-desktop-3.18.3 or some other related package contains the updated driver from Intel. If so, I can report that it still does not work, because I still need to add nomodeset to start the system and not end up with a black screen. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c23
Freek de Kruijf
(In reply to Freek de Kruijf from comment #21)
kernel-default-4.3.0-1.1
I did an update today to kernel-default-4.5.0-1.1
All fixes should be in there.
You mentioned two things, I only see one thing to test. More details::
1. remove any 'quiet' and 'splash=silent' from the kernel command line to get rid of plymouth - *don't* include 'nomodeset'! - does this freeze?
I did this. It is not so much freeze. I get a black screen and whatever button I press, also <Ctrl><Alt><Backspace><Backspace> , the screen stays black. I am able to log into the system via ssh and everything seems to work, except apparently the X server. Also tried "systemctl restart display-manager.service". Result nothing happens.
2. when you boot with 'nomodeset', (all other options in place), log into the system as root - either remotely or on a text console - and run 'modprobe i915 modeset=1'. Does this freeze up?
Nothing happens.
Does any of the two give you a usable system?
No. Leap 42.1 still does not need the nomodeset. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c24
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c25
Freek de Kruijf
Please try the driver from:
http://download.opensuse.org/repositories/home:/eeich:/Test:/Intel/ openSUSE_Tumbleweed/
I got a huge file /var/log/Xorg.0.log which is still growing fast. AFAIK there are no lines with [EE] in it. Do you need this file? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c26
--- Comment #26 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c27
--- Comment #27 from Egbert Eich
After my previous start of the system with nomodeset removed and also quiet and splash=silent removed. I started the system without changing the boot ^^^^^^^ This means *with* 'nomodeset' on the boot line. Correct? line. Now I got in this file at the end:
[...]
40.256] (EE) No devices detected. [ 40.256] (EE) Fatal server error: [ 40.256] (EE) no screens found(EE) [ 40.256] (EE) Please consult the The X.Org Foundation support at http://wiki.x.org for help. [ 40.256] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information. [ 40.256] (EE)
This is to be expected when you have 'nomodeset' set as the Intel driver only supports KMS. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c28
Egbert Eich
(In reply to Egbert Eich from comment #24)
Please try the driver from:
http://download.opensuse.org/repositories/home:/eeich:/Test:/Intel/ openSUSE_Tumbleweed/
I got a huge file /var/log/Xorg.0.log which is still growing fast. AFAIK there are no lines with [EE] in it. Do you need this file?
Do you still get a black screen with KDE? Yes, I've enabled full debugging output for this test driver. It is just for testing. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c29
Freek de Kruijf
Do you still get a black screen with KDE?
Yes. But it is not so much with KDE. During the boot process - the messages are too fast to see after what message - the screen turns black and nothing appears on the screen any more. I tried all kinds of keys and key combinations, but the screen stays black. I can inspect the log files only by accessing via ssh. I also tried "systemctl restart display-manager.service", but that did not help either. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c30
--- Comment #30 from Egbert Eich
(In reply to Egbert Eich from comment #28)
Do you still get a black screen with KDE?
Yes. But it is not so much with KDE. During the boot process - the messages are too fast to see after what message - the screen turns black and nothing appears on the screen any more. I tried all kinds of keys and key combinations, but the screen stays black.
I can inspect the log files only by accessing via ssh.
I also tried "systemctl restart display-manager.service", but that did not help either.
Ok, please disable the boot splash. Add plymouth.enable=0 to the kernel command line and get rid of the 'splash=silent quiet' options. Also boot init the multi-user mode instead of X by adding a '3' to the command line. See if this gives you a black screen again. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c31
Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c32
Egbert Eich
How can I get a working Tumbleweed again.
This is what we are trying to find out. A first step towards this goal would be for you to answer exactly the questions I have asked. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c33
Freek de Kruijf
Ok, please disable the boot splash. Add plymouth.enable=0 to the kernel command line and get rid of the 'splash=silent quiet' options. Also boot init the multi-user mode instead of X by adding a '3' to the command line. See if this gives you a black screen again.
Both with the old and the new xf86-video-intel, doing the above ends in a black screen. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c34
--- Comment #34 from Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c35
--- Comment #35 from Freek de Kruijf
This means that you cannot get a text console either. Since you can access the system from remote, can you set the 'nomodeset' boot option boot, log in remotey and simply run: 'modprobe i915 modeset=1'? Does this give you a black screen as well?
I have to add, I've got several machines with the same Baytrail GPU here and haven't seen this issue on any of them.
When I start with nomodeset on the boot line I now end with a text console, previously I got the sddm log in screen, so I can log in and give the above command, also via ssh. In both cases nothing visual happens. I still have a text console. This is the output of journalctl -b -u display-manager: -- Logs begin at do 2015-01-22 18:05:59 CET, end at wo 2016-03-23 16:39:42 CET. -- mrt 23 16:34:05 ltfctum.beelaertsict.nl systemd[1]: Starting X Display Manager... mrt 23 16:34:05 ltfctum.beelaertsict.nl display-manager[1227]: /etc/vconsole.conf available mrt 23 16:34:05 ltfctum.beelaertsict.nl display-manager[1227]: KEYMAP: us-acentos mrt 23 16:34:05 ltfctum.beelaertsict.nl display-manager[1227]: Command: localectl set-keymap us-acentos mrt 23 16:34:05 ltfctum.beelaertsict.nl display-manager[1227]: I: Using systemd /usr/share/systemd/kbd-model-map mapping mrt 23 16:34:07 ltfctum.beelaertsict.nl display-manager[1227]: error: unexpectedly disconnected from boot status daemon mrt 23 16:34:08 ltfctum.beelaertsict.nl systemd[1]: Started X Display Manager. mrt 23 16:34:10 ltfctum.beelaertsict.nl systemd[1]: display-manager.service: Main process exited, code=dumped, status=11/SEGV mrt 23 16:34:10 ltfctum.beelaertsict.nl systemd[1]: display-manager.service: Unit entered failed state. mrt 23 16:34:10 ltfctum.beelaertsict.nl systemd[1]: display-manager.service: Failed with result 'core-dump'. mrt 23 16:34:10 ltfctum.beelaertsict.nl sddm-helper[1598]: [PAM] Starting... mrt 23 16:34:10 ltfctum.beelaertsict.nl sddm-helper[1598]: [PAM] Authenticating... mrt 23 16:34:10 ltfctum.beelaertsict.nl sddm-helper[1598]: [PAM] returning. mrt 23 16:34:10 ltfctum.beelaertsict.nl sddm-helper[1598]: Received a wrong opcode instead of AUTHENTICATED: 0 mrt 23 16:34:11 ltfctum.beelaertsict.nl systemd-coredump[1600]: Process 1470 (sddm) of user 0 dumped core. mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: Starting service sddm..done mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: /usr/bin/xauth: (stdin):1: bad "remove" command line mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: /usr/bin/xauth: (stdin):2: bad "add" command line mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: /usr/bin/xauth: (stdin):1: bad "remove" command line mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: /usr/bin/xauth: (stdin):2: bad "add" command line mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: /usr/bin/xauth: (stdin):1: bad "remove" command line mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1227]: /usr/bin/xauth: (stdin):2: bad "add" command line mrt 23 16:34:12 ltfctum.beelaertsict.nl display-manager[1601]: ..done -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c36
--- Comment #36 from Egbert Eich
(In reply to Egbert Eich from comment #34)
This means that you cannot get a text console either. Since you can access the system from remote, can you set the 'nomodeset' boot option boot, log in remotey and simply run: 'modprobe i915 modeset=1'? Does this give you a black screen as well?
I have to add, I've got several machines with the same Baytrail GPU here and haven't seen this issue on any of them.
When I start with nomodeset on the boot line I now end with a text console, previously I got the sddm log in screen, so I can log in and give the above command, also via ssh. In both cases nothing visual happens. I still have a text console.
Again, you didn't answer my question - but for the nomodeset case, could you please provide the output of: 'cat /dev/fb'? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c37
--- Comment #37 from Freek de Kruijf
When I start with nomodeset on the boot line I now end with a text console, previously I got the sddm log in screen, so I can log in and give the above command, also via ssh. In both cases nothing visual happens. I still have a text console.
Again, you didn't answer my question - but for the nomodeset case, could you please provide the output of: 'cat /dev/fb'?
When I wrote I give the above command. this is the command "modprobe i915 modeset=1", so I did answer your question, which is nothing appears to happen after giving that command. So I do not get a black screen. The output of "cat /dev/fb" is: cat: /dev/fb: No such file or directory There is a /dev/fb0, but the output gives question marks on an oval black background. cat /dev/fb0 | wc -c gives: 4227072 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c38
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c39
Freek de Kruijf
Sorry, I meant to say 'cat /proc/fb'. What does this say? Also, please provide the full Xserver log file: /var/log/Xorg.0.log
ltfctum:~ # cat /proc/fb 0 inteldrmfb When using nomodeset it is: 0 EFI VGA The gziped file is attached. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c40
--- Comment #40 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c41
Egbert Eich
Created attachment 670370 [details] /var/log/Xorg.0.log gziped
This was generated for KMS not for the case with 'nomodeset'. Can you get a log file for nomodeset as well, please? This is so we can get the minimal system back to work: nomodeset with fbdev driver. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c42
Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c43
--- Comment #43 from Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c44
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c45
Freek de Kruijf
In addition to what I've suggested in comment #43, you may want to get the openSUSE KOTD: http://download.opensuse.org/repositories/Kernel:/HEAD/standard and install the 'vanilla' version. If this also gives you a black screen, we need to open a ticket on freedesktop.org.
I tried this, however this is on an EFI system and the vanilla kernel apparently has not been properly signed, so I can't start the system with that kernel. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c46
--- Comment #46 from Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c47
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c48
--- Comment #48 from Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c49
Freek de Kruijf
Further question: Originally you reported that the screen turns black only when you log into KDE. XFCE worked for you - even without 'nomodeset'. If I understand your recent reports correctly this has changed and now you get a black screen even in text mode as soon as you omit 'nomodeset'. Is this correct?
Right after installation from the USB with the distribution image I could get a little further, see comment #1. After a few reboots the screen turns black during the boot process. Never got a log in screen again without nomodeset. If you think this is important, I can repeat the installation from scratch, but I will ruin my Tumbleweed system, although I will keep the /home partition. Tumbleweed is not the main system on my laptop. 13.2 still is, and I have Leap 42.1 also. Leap 42.1 does not need the nomodeset on that system, 13.2 still does. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c50
--- Comment #50 from Freek de Kruijf
Is there a way to disable secure boot in your BIOS?
I suppose there is, however I did not succeed in getting to a screen to be able to disable it. I finally gave up. Maybe I should search for the right method. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c51
Egbert Eich
(In reply to Egbert Eich from comment #48)
Further question: Originally you reported that the screen turns black only when you log into KDE. XFCE worked for you - even without 'nomodeset'. If I understand your recent reports correctly this has changed and now you get a black screen even in text mode as soon as you omit 'nomodeset'. Is this correct?
Right after installation from the USB with the distribution image I could get a little further, see comment #1. After a few reboots the screen turns
This is why I've asked. Initially (with a different kernel of course) you were at least able to log in and everything seemed to be fine. Since then things turned worse.
black during the boot process. Never got a log in screen again without nomodeset. If you think this is important, I can repeat the installation from scratch, but I will ruin my Tumbleweed system, although I will keep the /home partition. Tumbleweed is not the main system on my laptop. 13.2 still is, and I have Leap 42.1 also. Leap 42.1 does not need the nomodeset on that system, 13.2 still does.
This reminds me of another ticket that's still open - that one however is not about Intel but about NVIDIA graphics. The patterns are similar, though. I'm starting to wonder if systemd is not playing some tricks on us - systemd is rather intrusive and tries to manage about every aspect of the system. I was thinking about backlight control for instance: Could you add try and add 'systemd.restore_state=0' to the kernel command line? Alternatively, try to move away '/usr/lib/systemd/systemd-backlight'. Another thing you could try is to connect an external display and see if this behaves in the same way. If none of this gets you any further, could you please get me the output of: 'ls /sys/class/drm'. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c52
Freek de Kruijf
(In reply to Freek de Kruijf from comment #49)
(In reply to Egbert Eich from comment #48)
Further question: Originally you reported that the screen turns black only when you log into KDE. XFCE worked for you - even without 'nomodeset'. If I understand your recent reports correctly this has changed and now you get a black screen even in text mode as soon as you omit 'nomodeset'. Is this correct?
Right after installation from the USB with the distribution image I could get a little further, see comment #1. After a few reboots the screen turns
This is why I've asked. Initially (with a different kernel of course) you were at least able to log in and everything seemed to be fine. Since then things turned worse.
black during the boot process. Never got a log in screen again without nomodeset. If you think this is important, I can repeat the installation from scratch, but I will ruin my Tumbleweed system, although I will keep the /home partition. Tumbleweed is not the main system on my laptop. 13.2 still is, and I have Leap 42.1 also. Leap 42.1 does not need the nomodeset on that system, 13.2 still does.
This reminds me of another ticket that's still open - that one however is not about Intel but about NVIDIA graphics. The patterns are similar, though. I'm starting to wonder if systemd is not playing some tricks on us - systemd is rather intrusive and tries to manage about every aspect of the system.
I was thinking about backlight control for instance: Could you add try and add 'systemd.restore_state=0' to the kernel command line?
I removed nomodeset from the command line and added systemd.restore_state=0 . This gives me the the desired log in screen, however on VT8 not VT7. After entering username and pasword the screen becomes black again. I hear the sound played to indicate the KDE session has started. Also <Ctrl>+<Alt>+<F1> does not give me visible VT1. Blindly logging in and shutdown -P now powers down the laptop.
Alternatively, try to move away '/usr/lib/systemd/systemd-backlight'.
Moved away the above file. Removed nomodeset from the command line. After that I got the same behaviour as above, o.a. black screens.
Another thing you could try is to connect an external display and see if this behaves in the same way.
This may take another day.
If none of this gets you any further, could you please get me the output of: 'ls /sys/class/drm'.
# ls -l /sys/class/drm totaal 0 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0-DP-1 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0/card0-DP-1 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0-eDP-1 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0/card0-eDP-1 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0-HDMI-A-1 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0-VGA-1 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0/card0-VGA-1 lrwxrwxrwx 1 root root 0 29 mrt 16:24 controlD64 -> ../../devices/pci0000:00/0000:00:02.0/drm/controlD64 lrwxrwxrwx 1 root root 0 29 mrt 16:24 renderD128 -> ../../devices/pci0000:00/0000:00:02.0/drm/renderD128 -r--r--r-- 1 root root 4096 29 mrt 17:29 version -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c53
Egbert Eich
I removed nomodeset from the command line and added systemd.restore_state=0 . This gives me the the desired log in screen, however on VT8 not VT7. After entering username and pasword the screen becomes black again. I hear the sound played to indicate the KDE session has started. Also <Ctrl>+<Alt>+<F1> does not give me visible VT1. Blindly logging in and shutdown -P now powers down the laptop.
Ok, what was the difference to not having 'systemd.restore_state=0'? IMHO, you were not able to see the login screen, is this correct? What happens if you pick Xfce as desktop, now? Can you still log in there and see something?
This may take another day.
What you could try quickly - create a test user and see if you can log into this newly created test account using KDE.
If none of this gets you any further, could you please get me the output of: 'ls /sys/class/drm'.
# ls -l /sys/class/drm totaal 0 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0-DP-1 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0/card0-DP-1 lrwxrwxrwx 1 root root 0 29 mrt 16:24 card0-eDP-1 -> ../../devices/pci0000:00/0000:00:02.0/drm/card0/card0-eDP-1 [..] Thanks!
-- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c54
Freek de Kruijf
(In reply to Freek de Kruijf from comment #52)
I removed nomodeset from the command line and added systemd.restore_state=0 . This gives me the the desired log in screen, however on VT8 not VT7. After entering username and pasword the screen becomes black again. I hear the sound played to indicate the KDE session has started. Also <Ctrl>+<Alt>+<F1> does not give me visible VT1. Blindly logging in and shutdown -P now powers down the laptop.
Ok, what was the difference to not having 'systemd.restore_state=0'? IMHO, you were not able to see the login screen, is this correct?
Yes.
What happens if you pick Xfce as desktop, now? Can you still log in there and see something?
I tried IceWM and I could start and see an application that I also have in KDE. After logging out I got log in screen again.
This may take another day.
What you could try quickly - create a test user and see if you can log into this newly created test account using KDE.
Did that and I can start a Plasma5 session, end that session and start the session again. I still have file backlight moved away. However when I start a session with my username the screen turns black and also going to VT1 shows a black screen. So something in my user configuration causes the black screen? Do you still need the test with an extra monitor attached? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c55
Egbert Eich
Did that and I can start a Plasma5 session, end that session and start the session again. I still have file backlight moved away. However when I start a session with my username the screen turns black and also going to VT1 shows a black screen. So something in my user configuration causes the black screen?
Yes, Plasma5 sessions store the last known settings and restore them on the next login - even if they were bogus. I'm not sure which settings are wrong: it could be backlight, but Baytrail also has issues with enabling planes. It could be, that the latter is an issue as well. I'd like to know whether the Plasma5 session uses a different mode than the login screen. Please do the following (this assumes you use sddm as login manager): - log in as root from remote. - when the login manager is active (sddm) do from remote: $ xauth=$(p=$(ps -C sddm-greeter -o pid=) && p=${p# } && cat /proc/$p/environ | tr '\0' '\n' | grep XAUTHORITY) $ export $xauth # this is no mistake! leave the '$' in place! $ export DISPLAY=:0 $ xrandr -q > log.file - Now log into the Plasma5 session (with the black screen) - From remote run: $ xrandr -q >> log.file - Get me log.file In log.file you should see 2 outputs of xrandr. Check if the currently active modes listed in each one differ (marked by '*').
Do you still need the test with an extra monitor attached?
We may get by without it. Let's see. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c56
--- Comment #56 from Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c57
Freek de Kruijf
What would also work - and it would be easier to obtain - is an Xserver log file recovered *after* Plasma5 went black. You'd have to make sure, it hasn't been overwritten by the next Xserver, yet, ie. you should recover it while the server that went 'black' is still running (*). I'm looking for any lines containing 'switch to mode'.
(*) If you have rebooted and a new server has started, you should be able to find the previous log in /var/log/Xorg.0.log.old (please make sure that it is the right one by checking the time stamp of the file).
Regarding backlight:
I will attach the requested file. Got it by still having backlight moved away, starting without nomodeset, but with systemd.restore_status=0, as can be seen from the boot line at the beginning of the Xorg.0.log file. After that I started the Plasma5 session and got a black screen. I went to VT1 and entered blind root and the password and a shutdown command. The system started again, but I booted another system and got the file from the Tumbleweed partition. Please let me know what I should do next. Currently I do not have remote access to this system, only tonight. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c58
--- Comment #58 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c59
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c60
Freek de Kruijf
Regarding backlight: (This is best done from remote) - Check: /sys/class/backlight it should contain one or two subdirectories (one should be intel_backlight).
Before starting the system I restored systemd-backlight. # ls -l /sys/devices/pci0000\:00/0000\:00\:02.0/drm/card0/card0-eDP-1/intel_backlight/ totaal 0 -r--r--r-- 1 root root 4096 31 mrt 23:49 actual_brightness -rw-r--r-- 1 root root 4096 31 mrt 23:45 bl_power -rw-r--r-- 1 root root 4096 31 mrt 23:45 brightness lrwxrwxrwx 1 root root 0 31 mrt 23:49 device -> ../../card0-eDP-1 -r--r--r-- 1 root root 4096 31 mrt 23:45 max_brightness drwxr-xr-x 2 root root 0 31 mrt 23:45 power lrwxrwxrwx 1 root root 0 31 mrt 23:45 subsystem -> ../../../../../../../class/backlight -r--r--r-- 1 root root 4096 31 mrt 23:45 type -rw-r--r-- 1 root root 4096 31 mrt 23:45 uevent
The structure of these subdirectories is pretty identical. For you the files brightness and max_brightness are relevant. - Get 'max_brighness':
is 7812
$ cat max_brightness - Set the brightness value: (try 0, the value of max_brightness and values between) $ echo <value> > brightness See if this changes your settings and makes the screen light up again.
echo 0 -n > brightness did not change anything nor did entering 3000 and 7812. (from comment #59)
Also you may want to check DPMS: . When you have a chance for a remote login, could you check the status of /sys/class/drm/card0-eDP-1/dpms when the Plasma5 session is blank, as well? Ie. do 'cat /sys/class/drm/card0-eDP-1/dpms' and check what is read.
# ls -l /sys/class/drm/card0-eDP-1/dpms -r--r--r-- 1 root root 4096 1 apr 00:05 /sys/class/drm/card0-eDP-1/dpms # cat /sys/class/drm/card0-eDP-1/dpms On This looks to be not off, so on. For the above I only removed nomodeset from the command line. Should I add systemd.restore_state=0 and repeat? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c61
--- Comment #61 from Egbert Eich
Before starting the system I restored systemd-backlight.
That's fine.
# ls -l
[..]
The structure of these subdirectories is pretty identical. For you the files brightness and max_brightness are relevant. - Get 'max_brighness':
is 7812
Can you get the value from 'brighness' right after a boot?
$ cat max_brightness - Set the brightness value: (try 0, the value of max_brightness and values between) $ echo <value> > brightness See if this changes your settings and makes the screen light up again.
echo 0 -n > brightness
This is wrong: if you want to specify the '-n' option it needs to be at the beginning of the command line.
did not change anything nor did entering 3000 and 7812.
This looks to be not off, so on.
For the above I only removed nomodeset from the command line. Should I add systemd.restore_state=0 and repeat?
This just prevents the previously saved value to be restored. Can you try the following? If you reboot after your Plasma5 session went black, and don't specify this option on the next reboot, your will get a black login screen after the reboot. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c62
--- Comment #62 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c63
--- Comment #63 from Freek de Kruijf
Can you get the value from 'brighness' right after a boot?
It is 4687, max_brightness is 7812.
This is wrong: if you want to specify the '-n' option it needs to be at the beginning of the command line.
Did not seem to be of any harm.
For the above I only removed nomodeset from the command line. Should I add systemd.restore_state=0 and repeat?
This just prevents the previously saved value to be restored.
Can you try the following? If you reboot after your Plasma5 session went black, and don't specify this option on the next reboot, your will get a black login screen after the reboot.
Yesterday with your newest xf86-video-intel module, version 2.99.917.590_g094924f-168.1, and nomodeset removed from the boot command, I did not get a black screen any more. However it causes a lot of output to /var/log/Xorg.0.log. This file does not contain a failing dbus message any more. At this moment, same as above, I again hear the sound of an active Plasma5 session, but the screen is black; brightness is at its maximum. I tried different values for brightness and found that 0 gives black, maximum gives black, close to maximum 7500 gives a dim screen, 4687 gives a bright screen. Seems we have found culprit. I will try the xf86-video-intel from the main repository. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c64
--- Comment #64 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c65
--- Comment #65 from Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c66
Egbert Eich
brightness as a workaround, although keeping nomodeset is also workable for me.
I order get all the efforts spent to pay off and to let others for benefit from what we have learned so far, I would like to find a fix for this issue. For this I need more stuff: 1. add drm.debug=0x4 to the kernel command line, reboot and run as root: # dmesg > dmesg.out Please get me the content of 'dmesg.out'. 2. run as root: # intel_bios_dumper intel_bios.data get me the output file 'intel_bios.data' -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c67
Freek de Kruijf
If you want to try a newer user space driver without getting all the annoying messages (which make things really slow), you can use a different repo I've created: http://download.opensuse.org/repositories/home:/eeich:/Test:/Intel-NoDebug/ openSUSE_Tumbleweed/
But at least we got somewhere and did make some progress: we did find that the backlight is the culprit. Fist thing I need from you is 1. What you get when you do: systemctl list-units | grep backlight
# systemctl list-units | grep backlight sys-devices-pci0000:00-0000:00:02.0-drm-card0-card0\x2deDP\x2d1-intel_backlight.device loaded active plugged /sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-eDP-1/intel_backlight systemd-backlight@backlight:intel_backlight.service loaded active exited Load/Save Screen Backlight Brightness of backlight:intel_backlight system-systemd\x2dbacklight.slice loaded active active system-systemd\x2dbacklight.slice
2. What you get by doing: ls /var/lib/systemd/backlight/
# ls /var/lib/systemd/backlight/ pci-0000:00:02.0:backlight:intel_backlight
3. Determine if you get a somewhat linear behaviour between 0 and 4687.
No. 0 is black and even 1 is bright. It looks like values above 4687 make a difference in brightness. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c68
--- Comment #68 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c69
--- Comment #69 from Freek de Kruijf
(In reply to Freek de Kruijf from comment #64) [..] I can make a root cron job at boot to set
brightness as a workaround, although keeping nomodeset is also workable for me.
I order get all the efforts spent to pay off and to let others for benefit from what we have learned so far, I would like to find a fix for this issue. For this I need more stuff: 1. add drm.debug=0x4 to the kernel command line, reboot and run as root: # dmesg > dmesg.out Please get me the content of 'dmesg.out'.
See attached file dmesg.out.gz
2. run as root: # intel_bios_dumper intel_bios.data get me the output file 'intel_bios.data'
# intel_bios_dumper intel_bios.data Couldn't read graphics card ROM: Input/output error -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c70
--- Comment #70 from Egbert Eich
1. What you get when you do: systemctl list-units | grep backlight
# systemctl list-units | grep backlight
sys-devices-pci0000:00-0000:00:02.0-drm-card0-card0\x2deDP\x2d1- intel_backlight.device loaded active plugged /sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-eDP-1/intel_backlight systemd-backlight@backlight:intel_backlight.service loaded active exited Load/Save Screen Backlight Brightness of backlight:intel_backlight system-systemd\x2dbacklight.slice loaded active active system-systemd\x2dbacklight.slice
OK, so there is no ACPI backlight. That's what I was trying to find out.
2. What you get by doing: ls /var/lib/systemd/backlight/
# ls /var/lib/systemd/backlight/ pci-0000:00:02.0:backlight:intel_backlight
Ok. This is the file then where systemd stores the backlight value on shut down.
3. Determine if you get a somewhat linear behaviour between 0 and 4687.
No. 0 is black and even 1 is bright. It looks like values above 4687 make a difference in brightness.
This is something we need to report to Intel. The non-linear behavior is of course he root of all your problems. I need one more info, though: (In reply to Freek de Kruijf from comment #69)
# dmesg > dmesg.out Please get me the content of 'dmesg.out'.
See attached file dmesg.out.gz
Ok, this contains a lot of useful information already.
2. run as root: # intel_bios_dumper intel_bios.data get me the output file 'intel_bios.data'
# intel_bios_dumper intel_bios.data Couldn't read graphics card ROM: Input/output error
Ok, not sure why this happens. We should be able to get the information I'm missing from the kernel logs as well if you set drm.debug=0x6 (instead of 0x4) and get me a dmesg again. Do you have an account on bugs.freedesktop.org? I may open a ticket there for Intel to look into this. It would be good if I could add you to this in case Intel needs further information from you. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c71
--- Comment #71 from Freek de Kruijf
# intel_bios_dumper intel_bios.data Couldn't read graphics card ROM: Input/output error
Ok, not sure why this happens. We should be able to get the information I'm missing from the kernel logs as well if you set drm.debug=0x6 (instead of 0x4) and get me a dmesg again.
See atached file.
Do you have an account on bugs.freedesktop.org? I may open a ticket there for Intel to look into this. It would be good if I could add you to this in case Intel needs further information from you.
Yes. My login name is f.de.kruijf AT gmail DOT com -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c72
Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
http://bugzilla.opensuse.org/show_bug.cgi?id=899879#c73
Egbert Eich
http://bugzilla.opensuse.org/show_bug.cgi?id=899879
Egbert Eich
participants (1)
-
bugzilla_noreply@novell.com