[Bug 1089116] New: Flashing screen on startup - X-Server segmentation fault
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116 Bug ID: 1089116 Summary: Flashing screen on startup - X-Server segmentation fault Classification: openSUSE Product: openSUSE Distribution Version: Leap 15.0 Hardware: x86-64 OS: All Status: NEW Severity: Major Priority: P5 - None Component: X.Org Assignee: xorg-maintainer-bugs@forge.provo.novell.com Reporter: suse.junky@t-online.de QA Contact: xorg-maintainer-bugs@forge.provo.novell.com Found By: --- Blocker: --- Created attachment 766766 --> http://bugzilla.opensuse.org/attachment.cgi?id=766766&action=edit Xorg logfile I installed Leap 15.0 Build191.1 on my Laptop. When i start the system somewhere in the boot process the screen starts flashing and the only way out is to press "Ctrl+Alt+Del" until the system starts rebooting. Booting the Rescue System (from the installation device) and looking at "/var/log/Xorg.0.log" reveals: [ 81.000] (EE) FBDEV(0): FBIOPUTCMAP: Device or resource busy [ 81.000] (EE) FBDEV(0): FBIOPUTCMAP: Device or resource busy [ 81.000] (==) FBDEV(0): DPMS enabled [ 81.000] (--) RandR disabled [ 81.004] (II) AIGLX: Screen 0 is not DRI2 capable [ 81.004] (EE) AIGLX: reverting to software rendering [ 81.018] (II) IGLX: enabled GLX_MESA_copy_sub_buffer [ 81.019] (II) IGLX: Loaded and initialized swrast [ 81.019] (II) GLX: Initialized DRISWRAST GL provider for screen 0 [ 82.582] (II) modeset(G0): Damage tracking initialized [ 82.582] (EE) [ 82.582] (EE) Backtrace: [ 82.582] (EE) 0: /usr/bin/X (xorg_backtrace+0x65) [0x5561c76f9515] [ 82.582] (EE) 1: /usr/bin/X (0x5561c7544000+0x1b92c9) [0x5561c76fd2c9] [ 82.582] (EE) 2: /lib64/libpthread.so.0 (0x7f90df4db000+0x122e0) [0x7f90df4ed2e0] [ 82.582] (EE) 3: /usr/bin/X (RRSetChanged+0x4c) [0x5561c765603c] [ 82.582] (EE) 4: /usr/bin/X (RRScreenSetSizeRange+0x5f) [0x5561c765ab4f] [ 82.582] (EE) 5: /usr/bin/X (xf86RandR12CreateScreenResources+0x2f3) [0x5561c76187f3] [ 82.582] (EE) 6: /usr/bin/X (0x5561c7544000+0xc7490) [0x5561c760b490] [ 82.582] (EE) 7: /usr/bin/X (0x5561c7544000+0x580c1) [0x5561c759c0c1] [ 82.582] (EE) 8: /lib64/libc.so.6 (__libc_start_main+0xea) [0x7f90df141f4a] [ 82.582] (EE) 9: /usr/bin/X (_start+0x2a) [0x5561c7585f1a] [ 82.582] (EE) [ 82.582] (EE) Segmentation fault at address 0xb8 [ 82.582] (EE) Fatal server error: [ 82.582] (EE) Caught signal 11 (Segmentation fault). Server aborting [ 82.582] (EE) [ 82.582] (EE) Please consult the The X.Org Foundation support at http://wiki.x.org for help. [ 82.583] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information. [ 82.583] (EE) [ 83.056] (EE) Server terminated with error (1). Closing log file. (for complete log file see attachement "Xorg.0.log") Hardware: ========= LG 3840x2160 IPS-display NVIDIA GeForce GTX 1060 graphics card Intel HD Graphics 630 In UEFI for the graphic-cards either "DISCRET" or "MSHYBRID" can be selected. How to reproduce: ================= Use Suse studio Imagewriter to put openSUSE-Leap-15.0-DVD-x86_64-Build191.1-Media.iso on an USB-Stick. Set graphic-cards-option in UEFI to "MSHYBRID". Boot from USB-Stick with kernel parameters "acpi_os_name="Linux" nouveau.modeset=0 3" and install with DE=KDE and "/" and "/home" formatted with "ext4". Reboot with same kernel parameters as above and do # nmtui (to get a WLAN connection) # zypper dup (to receive latest updates) # zypper in terminus-bitmap-fonts (to get a font that can be read on 4K display) # setfont ter-132b # echo "FONT=ter-132b" >>/etc/vconsole.conf # zypper in bumblebee VirtualGL-devel dkms # vi /etc/modprobe.d/50-blacklist.conf (to add "blacklist nouveau") # usermod -a -G bumblebee,video username # systemctl enable bumblebeed # systemctl start bumblebeed # systemctl enable dkms # systemctl start dkms # dracut -f # init 6 Result: ======= System starts booting but somewhere the bootprocess stopps, the screen starts flashing (on/off) and the only way out of this situation is to repeatedly press "Ctrl+Alt+Del" until the machine starts rebooting. Expected result: ================ The system boots into graphic mode. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c1
Michal Srb
[ 80.829] (EE) [drm] Failed to open DRM device for pci:0000:01:00.0: -19 ... [ 80.832] (**) FBDEV(0): claimed PCI slot 1@0:0:0
Is nouveau kernel module loaded? Please attach also the output of dmesg. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c2
--- Comment #2 from Stefan Dirsch
Is nouveau kernel module loaded?
Seems it's not. By intention for whatever reasons ... Boot from USB-Stick with kernel parameters "acpi_os_name="Linux" nouveau.modeset=0 3" -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c3
--- Comment #3 from Michal Srb
Boot from USB-Stick with kernel parameters "acpi_os_name="Linux" nouveau.modeset=0 3"
Ok I missed that. And in addition apparently nouveau was blacklisted:
vi /etc/modprobe.d/50-blacklist.conf (to add "blacklist nouveau")
That makes this bug pure duplicate of bug #1068961. I do not really know how is the bumblebee supposed to work, but I would expect you need to have either nouveau or nvidia driver installed. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c4
--- Comment #4 from Bit Juggler
(In reply to Michal Srb from comment #1)
Is nouveau kernel module loaded?
Seems it's not. By intention for whatever reasons ...
Boot from USB-Stick with kernel parameters "acpi_os_name="Linux" nouveau.modeset=0 3"
Those parameters were only used to boot the installation media for installation and to boot the very first time after installation in order to install all software necessary to run bumblebee (both times in runlevel 3). After that no kernel parameters are used. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c5
--- Comment #5 from Bit Juggler
(In reply to Stefan Dirsch from comment #2)
Boot from USB-Stick with kernel parameters "acpi_os_name="Linux" nouveau.modeset=0 3"
Ok I missed that. And in addition apparently nouveau was blacklisted:
vi /etc/modprobe.d/50-blacklist.conf (to add "blacklist nouveau")
That makes this bug pure duplicate of bug #1068961.
I do not really know how is the bumblebee supposed to work, but I would expect you need to have either nouveau or nvidia driver installed.
If you install bumblebee you are requested to blacklist nouveau so that bumblebee can load it under its own control. I'm not expert enough to tell whether bug #1068961 refers to the same problem or not. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c6
--- Comment #6 from Bit Juggler
The crash looks the same as in bug #1068961. ... Is nouveau kernel module loaded? Please attach also the output of dmesg.
Seems that the journal by default is placed in memory. To post the journal i have to find out how to change this. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c7
Bit Juggler
The crash looks the same as in bug #1068961. ... Is nouveau kernel module loaded?
Looking at Xorg.0.log i can see ... [ 80.651] (II) LoadModule: "nouveau" [ 80.651] (II) Loading /usr/lib64/xorg/modules/drivers/nouveau_drv.so [ 80.651] (II) Module nouveau: vendor="X.Org Foundation" [ 80.651] compiled for 1.19.6, module version = 1.0.15 [ 80.651] Module class: X.Org Video Driver [ 80.651] ABI class: X.Org Video Driver, version 23.0 ... so i guess nouveau is loaded (under the control of bumblebee?). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c8
Michal Srb
I'm not expert enough to tell whether bug #1068961 refers to the same problem or not.
Based on the identical backtrace and the combination of fbdev and modesetting it is likely the same. I will try whether the patch linked in the bug fixes it. (In reply to Bit Juggler from comment #7)
(In reply to Michal Srb from comment #1)
Is nouveau kernel module loaded?
Looking at Xorg.0.log i can see
... [ 80.651] (II) LoadModule: "nouveau" [ 80.651] (II) Loading /usr/lib64/xorg/modules/drivers/nouveau_drv.so [ 80.651] (II) Module nouveau: vendor="X.Org Foundation" [ 80.651] compiled for 1.19.6, module version = 1.0.15 [ 80.651] Module class: X.Org Video Driver [ 80.651] ABI class: X.Org Video Driver, version 23.0 ...
so i guess nouveau is loaded (under the control of bumblebee?).
Note that there are two parts to the driver - the nouveau kernel driver and the nouveau Xorg driver (a module to X server). This shows that the nouveau Xorg driver was loaded, but later messages also show that it failed to open the matching DRM device (/dev/dri/card*) to communicate with the kernel driver and it bailed out. Most likely because the nouveau kernel driver was blacklisted. But if it has to be blacklisted for bumblebee, then it is intentional setup. It should not crash, so lets try to fix it. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c9
Michal Srb
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c10
--- Comment #10 from Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c11
Bit Juggler
Can you please test if xorg-x11-server from this project in build service fixes the issue?
It contains patch dd00e5466a0e4ea313d1860824da4123692827ed (xfree86: Remove broken RANDR disabling logic) that should prevent the crash on initialization and an experimental patch to prevent similar crash later when randr is used.
Project: https://build.opensuse.org/project/show/home:michalsrb:branches:bnc1068961: X11:XOrg
Leap 15 repository: https://download.opensuse.org/repositories/home:/michalsrb:/branches:/ bnc1068961:/X11:/XOrg/openSUSE_Leap_15.0/
Added your repository https://download.opensuse.org/repositories/home:/michalsrb:/branches:/bnc106... did zypper dup -r your-repository (Yet I have not applied the 700+ other updates available by "zypper dup" today.) rebooted the system. The system starts up but eventually only gives a black screen. Pressing "Ctrl+Alt+Backspace" twice brings back the console for a second but then there is a black screen again. See attached "Xorg.0.log.15.0.pat" for details. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c12
--- Comment #12 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c13
--- Comment #13 from Bit Juggler
Ok. So the segfault is fixed with these patches. Remains that bumblebee doesn't work. I suggest to add the patches to our Xserver and reassign this ticket to the bugowners/maintainers of Bumblebee.
Now i updated to the latest packages and are still using the X-Server provided by Michal Srb. No fundamental changes. However when the screen becomes black i can switch to a console (Ctrl+Alt+Fx) and do a login. So i gathered some information: systemctl status bumblebeed shows that bumblebeed is running and no errors are reported. journalctl -b 0 -p 3 shows 3 problems: Apr 12 16:18:31 linux-c9lf systemd-vconsole-setup[177]: /usr/bin/loadkeys failed with error code 1. Apr 12 16:18:31 linux-c9lf systemd-vconsole-setup[177]: KD_FONT_OP_GET failed while trying to get the font metadata: Input/output error Apr 12 16:18:31 linux-c9lf systemd-vconsole-setup[177]: Fonts will not be copied to remaining consoles Apr 12 16:18:32 linux-c9lf systemd[1]: Failed to start Setup Virtual Console. Apr 12 16:18:32 linux-c9lf systemd[1]: systemd-vconsole-setup.service: Failed with result 'exit-code'. Apr 12 16:18:33 linux-c9lf systemd-rfkill[957]: Failed to parse state file /var/lib/systemd/rfkill/pci-0000:6e:00.0:wlan: Invalid argument Apr 12 16:18:34 linux-c9lf systemd-rfkill[957]: Failed to parse state file /var/lib/systemd/rfkill/pci-0000:00:14.0-usb-0:8:1.0:bluetooth: Invalid argument For your information I will attache the output of "lsmod" as well. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c14
--- Comment #14 from Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c15
--- Comment #15 from Stefan Dirsch
Ok. So the segfault is fixed with these patches. Remains that bumblebee doesn't work. I suggest to add the patches to our Xserver and reassign this ticket to the bugowners/maintainers of Bumblebee.
done. ;-) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c16
Bit Juggler
(In reply to Stefan Dirsch from comment #12)
Ok. So the segfault is fixed with these patches. Remains that bumblebee doesn't work. I suggest to add the patches to our Xserver and reassign this ticket to the bugowners/maintainers of Bumblebee.
done. ;-)
Well, there are still lots of errors in the Xorg.0.log: one [ 567.532] (EE) [drm] Failed to open DRM device for pci:0000:01:00.0: -19 many [ 567.737] (EE) FBDEV(0): FBIOPUTCMAP: Device or resource busy one [ 567.744] (EE) AIGLX: reverting to software rendering and systemctl status bumblebeed lists NO errors! -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c17
--- Comment #17 from Michal Srb
Well, there are still lots of errors in the Xorg.0.log:
As far as I can tell, these errors are harmless (should be considered as warnings).
[ 567.532] (EE) [drm] Failed to open DRM device for pci:0000:01:00.0: -19
Following the bumblebee tutorial you blacklisted nouveau and did not install nvidia. X server sees nvidia GPU and tries to use xf86-video-nouveau, which tries to open a DRM device for that PCI address, but founds none (there is none because the nouveau kernel module is blacklisted), so it displays this error and X server falls back to fbdev.
[ 567.737] (EE) FBDEV(0): FBIOPUTCMAP: Device or resource busy
Here I am less sure on how big issue is it, but I think it should not be problem either. The fbdev driver failed to set palette for the nvidia's framebuffer. Perhaps the nvidia's fbdev implementation does not support it. It may prevent things like setting gamma correction, but the rendering should still work. As per comment #4, you do not plan to use fbdev after the installation anyway.
[ 567.744] (EE) AIGLX: reverting to software rendering
This is not a problem. The the GLX extension is enabled so X server will provide it, but none of its drivers actually supports acceleration (no nvidia or nouveau), so it has to fallback to software rendering. (In reply to Bit Juggler from comment #4)
Boot from USB-Stick with kernel parameters "acpi_os_name="Linux" nouveau.modeset=0 3"
Those parameters were only used to boot the installation media for installation and to boot the very first time after installation in order to install all software necessary to run bumblebee (both times in runlevel 3).
After that no kernel parameters are used.
So now with the patched X server you were able to boot and get to console without crashing. If you proceed with the bumblebee setup, do things work right? If not, the bug is assigned to bumblebee maintainer. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c18
--- Comment #18 from Bit Juggler
... As far as I can tell, these errors are harmless (should be considered as warnings). ...
Thank you very much for the detailed explanations on all the "(EE)" lines i can see in my Xorg.0.log file. So you are sure that my problem is not related to the X-Server.(?) (In reply to Michal Srb from comment #17)
So now with the patched X server you were able to boot and get to console without crashing. If you proceed with the bumblebee setup, do things work right?
No! I only get a black screen but i can switch to a console (Crtl+Alt+Fx). There i checked for Bumblebee errors but can't find any. (In reply to Michal Srb from comment #17)
If not, the bug is assigned to bumblebee maintainer.
Is there one? Up to now there was no response (although openSUSE Leap 15.0 hits RC phase this week). Are there any other tools to deal with NVIDIA OPTIMUS (instead of using bumblebee)? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c19
--- Comment #19 from Stefan Dirsch
[ 567.737] (EE) FBDEV(0): FBIOPUTCMAP: Device or resource busy
Forget about this one. Everyone using fbdev driver gets this (since years/decades). False positive so to say ... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c20
Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c21
Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
Bit Juggler
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c22
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c23
Bit Juggler
Honestly I meanwhile doubt, that Bumblebee works for anyone on Leap 15/TW.Why is this considered P0/Critical? It's even a manual process to configure it at all.
Well, when i look here (https://news.opensuse.org/) i can read "Based on enterprise code tested millions of times opensuse leap 15 released" and "openSUSE Leap 15 now allows migration to SLE, brings a new partitioner, integrates the Groupware Kopano, moves to Firewalld – and also comes distributed by Linode (for Cloud and infrastructure setups) and on high-end hardware like Tuxedo Laptops (other Cloud and hardware vendors will follow)." When i happen to have such a piece of "high-end hardware like Tuxedo Laptops" and i experience that an average user will not be able to install/use openSUSE Leap 15.0 on on it, then i think this is "Critical" and should be resolved as soon as possible. But may be i'm too much of a sales person. However i found a way out: After installing "xf86-video-intel" from the OSS repository the system starts up in graphic mode. Playing with the KDE compositor options i found that switching the rendering backend from "OpenGL 2.0" to "XRender" will remove the sluggish behavior of the the GUI. In openSUSE Leap 42.3 there was no need to install "xf86-video-intel" or to switch the KDE compositor backend to "XRender". Nor does any of the openSUSE docs (Release notes, SDB:Articles, ...) mention any of this. So i don't know whether i just found a workaround for a bug or whether that's the way it is intended to be. However for me it works so i consider this problem as solved. Thanks to everybody who helped me to solve this. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c24
--- Comment #24 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c25
--- Comment #25 from Bit Juggler
Well, this looks more like a workaround for a bug to me. It's definitely not recommended to still use intel X driver on Kabylake. The issue is, that we never really supported Optimus systems. Bumblebee may have worked on some systems, if you were lucky.
My current guess is that Tuxedo preloads Leap 15 with Intel graphics disabled, proprietary NVIDIA driver preinstalled and possibly with a different desktop than KDE, but I simply don't know ...
If you visit the TUXEDO website and configure the laptop of your choice you may find that TUXEDO does not offer to pre-install openSUSE at all. Only Ubuntu (or Tuxedos special based on Ubuntu) is on offer. So guess why? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116
http://bugzilla.opensuse.org/show_bug.cgi?id=1089116#c26
--- Comment #26 from Stefan Dirsch
participants (1)
-
bugzilla_noreply@novell.com