[Bug 964429] New: Only 800x600 resolution for Nvidia graphics after kernel update
http://bugzilla.opensuse.org/show_bug.cgi?id=964429 Bug ID: 964429 Summary: Only 800x600 resolution for Nvidia graphics after kernel update Classification: openSUSE Product: openSUSE Distribution Version: Leap 42.1 Hardware: x86-64 OS: openSUSE 42.1 Status: NEW Severity: Major Priority: P5 - None Component: Kernel Assignee: kernel-maintainers@forge.provo.novell.com Reporter: Ulrich.Windl@rz.uni-regensburg.de QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- After kernel update and reboot, the display only shows a resolution of 800x600, while the vesafb detects the full resolution. Some syslog messages: kernel: vesafb: mode is 1680x1050x32, linelength=6720, pages=0 kernel: NVRM: loading NVIDIA UNIX x86_64 Kernel Module 340.96 Sun Nov 8 22:33:28 PST 2015 kernel: NVRM: Your system is not currently configured to drive a VGA console kernel: NVRM: on the primary VGA device. The NVIDIA Linux graphics driver NVRM: requires the use of a text-mode VGA console. Use of other console NVRM: drivers including, but not limited to, vesafb, may result in NVRM: corruption and stability problems, and is not supported.
From Xorg.0.log: (WW) NVIDIA(GPU-0): The EDID read for display device DFP-0 is invalid: the (WW) NVIDIA(GPU-0): checksum for EDID version 1 extension is invalid. (II) NVIDIA(0): NVIDIA GPU GeForce 9500 GT (G96) at PCI:1:0:0 (GPU-0)
List of packets installed (before everything worked fine): x11-video-nvidiaG03-340.96-40.1.x86_64 Mon 01 Feb 2016 11:24:20 AM CET nvidia-computeG03-340.96-40.1.x86_64 Mon 01 Feb 2016 11:24:15 AM CET nvidia-glG03-340.96-40.1.x86_64 Mon 01 Feb 2016 11:24:14 AM CET nvidia-uvm-gfxG03-kmp-default-340.96_k4.1.12_1-40.1.x86_64 Mon 01 Feb 2016 11:23:30 AM CET nvidia-gfxG03-kmp-default-340.96_k4.1.12_1-40.1.x86_64 Mon 01 Feb 2016 11:22:21 AM CET libXext6-32bit-1.3.3-3.1.x86_64 Mon 01 Feb 2016 11:22:21 AM CET kernel-default-devel-4.1.15-8.1.x86_64 Mon 01 Feb 2016 11:22:20 AM CET java-1_7_0-openjdk-1.7.0.95-25.1.x86_64 Mon 01 Feb 2016 11:22:19 AM CET aaa_base-extras-13.2+git20140911.61c1681-10.1.x86_64 Mon 01 Feb 2016 11:22:19 AM CET Mesa-libEGL1-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:22:18 AM CET kernel-devel-4.1.15-8.1.noarch Mon 01 Feb 2016 11:22:18 AM CET Mesa-libGL1-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:22:15 AM CET Mesa-libGLESv2-2-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:22:14 AM CET libOSMesa9-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:22:14 AM CET libX11-6-32bit-1.6.3-3.1.x86_64 Mon 01 Feb 2016 11:22:13 AM CET aaa_base-13.2+git20140911.61c1681-10.1.x86_64 Mon 01 Feb 2016 11:22:12 AM CET libvdpau_nouveau-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:22:11 AM CET libgif6-5.0.5-7.1.x86_64 Mon 01 Feb 2016 11:22:10 AM CET libgbm1-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:22:10 AM CET kernel-macros-4.1.15-8.1.noarch Mon 01 Feb 2016 11:22:10 AM CET kernel-default-4.1.15-8.1.x86_64 Mon 01 Feb 2016 11:21:20 AM CET java-1_7_0-openjdk-headless-1.7.0.95-25.1.x86_64 Mon 01 Feb 2016 11:21:13 AM CET Mesa-libglapi0-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:21:07 AM CET Mesa-11.0.8-137.1.x86_64 Mon 01 Feb 2016 11:21:06 AM CET libxcb1-32bit-1.11.1-1.1.x86_64 Mon 01 Feb 2016 11:21:05 AM CET libXau6-32bit-1.0.8-6.1.x86_64 Mon 01 Feb 2016 11:21:04 AM CET -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c1
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c2
--- Comment #2 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c3
--- Comment #3 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c4
--- Comment #4 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c5
Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c6
Takashi Iwai
I tried booting the previous kernel (via GRUB menu), but that didn't change a thing.
So this is no kernel regression, as it seems.
Next thing I tried was to remove the Nvidia "G03" packages via Yast software management, and surprisingly those modules were not removed, but replaced with Nvidia "G04" modules: nvidia-uvm-gfxG04-kmp-default-352.79_k4.1.12_1-19.1.x86_64 Mon Feb 1 13:12:37 2016 x11-video-nvidiaG04-352.79-19.1.x86_64 Mon Feb 1 13:09:20 2016 nvidia-computeG04-352.79-19.1.x86_64 Mon Feb 1 13:09:15 2016 nvidia-glG04-352.79-19.1.x86_64 Mon Feb 1 13:09:01 2016 nvidia-gfxG04-kmp-default-352.79_k4.1.12_1-19.1.x86_64 Mon Feb 1 13:08:06 2016
It's because of the dependency, I suppose.
These modules cauise a complete failure to star tthe X server (it seems). I had to switch to the text console. There I removed those "G04" packets (which worked). Then I could start an X session with the usual resolution again.
So, this looks more like an issue of Nvidia stuff. Changed the component. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c7
--- Comment #7 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c8
Stefan Dirsch
Please check your monitor/and or cable.
--> NEEDINFO -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c9
--- Comment #9 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c10
--- Comment #10 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c11
Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c12
--- Comment #12 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c13
--- Comment #13 from Ulrich Windl
This time nouveau driver - for whatever reasons. Is this with the same monitor and monitor cable?
The reason is explained in comment #2 (I removed the NVIDIA modules). The cables and monitor are the same. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c14
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c15
Egbert Eich
00384b1e5310000a2020202020200083
Did you check if it was always the same bit that had changed? Ulirch, when you got this error with the NVIDIA driver, did you power cycle each component (ie display, computer and KVM switch)? This could have been a temporary glitch unrelated to the software. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c16
--- Comment #16 from Egbert Eich
These modules cauise a complete failure to star tthe X server (it seems). I had to switch to the text console. There I removed those "G04" packets (which worked). Then I could start an X session with the usual resolution again.
Please define: 'a complete failure to star the X server'. In our nomenclature 'a complete failure' refers to the Xserver not starting at all - in which case you *are* in text console. (In reply to Ulrich Windl from comment #11)
Created attachment 664018 [details] Previous Xorg.0.log.old that did not send a valid video signal to the monitor.
Looking at this file, the time stamps reveal that the Xserver had been active for 1.5 hours. I don't think there were any issues with the video signal. (In reply to Egbert Eich from comment #15)
A single bit has tipped over in the 2nd last byte of the EDID block: < 00384b1e5310000a2020202020200182 ---
00384b1e5310000a2020202020200083
Actually, both EDID blocks are right: one has the extension flag set (the checksum is changed accordingly). The checksum of the extension block seems to be wrong (in which case the NVIDIA driver should probably ignore it). Ulrich, the information you've supplied is still to piecemeal to get NVIDIA involved: It is not clear at all if the log file you supplied in comment #3 is for the G03 or the G04 driver. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c17
Ulrich Windl
Please define: 'a complete failure to star the X server'.
As I wrote: There was no graphical login, but some message saying that the X server wouldn't start. I had to switch to another (text) console to log in.
In our nomenclature 'a complete failure' refers to the Xserver not starting at all - in which case you *are* in text console.
See above.
(In reply to Ulrich Windl from comment #11)
Created attachment 664018 [details] Previous Xorg.0.log.old that did not send a valid video signal to the monitor.
Looking at this file, the time stamps reveal that the Xserver had been active for 1.5 hours. I don't think there were any issues with the video signal.
I deduced from the fact that the monitor went into power save mode that there was no valid video signal. I wouldn't deduce that there was a valid video signal (on the correct output) from the fact that the server process was running for a while.
Ulrich, the information you've supplied is still to piecemeal to get NVIDIA involved: It is not clear at all if the log file you supplied in comment #3 is for the G03 or the G04 driver.
OK: Suggest instructions on how to get the NVIDIA driver working. And if it doesn't, suggest what log files you really need. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c18
Egbert Eich
(In reply to Egbert Eich from comment #16)
Please define: 'a complete failure to star the X server'.
As I wrote: There was no graphical login, but some message saying that the X server wouldn't start. I had to switch to another (text) console to log in.
Ok, now I understand. I seem to be missing a log file for this. The log file from comment #3 shows the Xserver starting and terminating normally. it didn't live ver long (all in all ~6 sec) which indicates that your DM (which one are you using?) terminated it. This might indicated that the DM had issues loading the required libs. Without knowledge of your DM we cannot find out where it logs to, though.
(In reply to Ulrich Windl from comment #11)
Created attachment 664018 [details] Previous Xorg.0.log.old that did not send a valid video signal to the monitor.
Looking at this file, the time stamps reveal that the Xserver had been active for 1.5 hours. I don't think there were any issues with the video signal.
I deduced from the fact that the monitor went into power save mode that there was no valid video signal. I wouldn't deduce that there was a valid video signal (on the correct output) from the fact that the server process was running for a while.
The screen usually goes into power save mode when it's idle for so long. Moreover, this log file is for the fbdev driver which is unlikely to show a black screen if your console was working at all. Also it is even more unlikely that the screen is loosing the video signal when using this driver.
Ulrich, the information you've supplied is still to piecemeal to get NVIDIA involved: It is not clear at all if the log file you supplied in comment #3 is for the G03 or the G04 driver.
OK: Suggest instructions on how to get the NVIDIA driver working. And if it doesn't, suggest what log files you really need.
This is not the prime purpose if this bugzilla. We are trying to determine if there is a bug in the software or the installation procedure. Once these things are fixed it should start working for you again if it hasn't before by the things we asked you to try. Currently, I see two issues: 1. the G03 driver you've used reads a defective EDID extension block which is supplied by your monitor. So either the read procedure is wrong or the block sent by the monitor - in which case the NVIDIA driver should simply ignore it. This is something for NVIDIA to fix. 2. when you deinstall the G03 driver you will get the G04 driver (according to your comment #2). Not sure why this happened. This is something for us to determine. If the G04 packages support your hardware (PCI ID: 10de:0640) as well, it should have worked. Have you rebooted in between? This ticket has seen attempts with too many drivers: NVIDIA G03, G04, fbdev and nouveau. All this is confusing and obfuscating the goal. Please 1. disable the NVIDIA repo, 2. deinstall whatever NVIDIA driver you've installed, 3. update the system to the current maintenance level doing 'zypper dup', 4. re-enable the NVIDIA repo and run 'zypper inr; zypper dup' This should give you the NVIDIA driver back - please let us know which ones... If it works now, 1. is fixed for this driver. BTW: The surest thing to determine if X is basically working and lighting up your screen, is by logging in to the text console as root and do: '/usr/bin/Xorg -retro & sleep 3; xterm -display :0'. You should get an X screen with the old fashioned root weave and an xterm. To determine if there are issues with the openGL libraries (ie you have installed unsuitable ones) you do 'gxlinfo' in this xterm and observe the output and any messages. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c19
--- Comment #19 from Ulrich Windl
Without knowledge of your DM we cannot find out where it logs to, though.
GNOME, so I guess gdm.
The screen usually goes into power save mode when it's idle for so long.
I know.
Moreover, this log file is for the fbdev driver which is unlikely to show a black screen if your console was working at all. Also it is even more unlikely that the screen is loosing the video signal when using this driver.
I was worried about the message: "Use of other console drivers including, but not limited to, vesafb, may result in corruption and stability problems, and is not supported."
This is not the prime purpose if this bugzilla. We are trying to determine if there is a bug in the software or the installation procedure.
YOU complained about the correct log files being missing, so I asked how to test the situation correctly.
Currently, I see two issues: 1. the G03 driver you've used reads a defective EDID extension block which is supplied by your monitor. So either the read procedure is wrong or the block sent by the monitor - in which case the NVIDIA driver should simply ignore it.
I suspected my KVM switch to need a reset (which I did in the meantime).
This is something for NVIDIA to fix. 2. when you deinstall the G03 driver you will get the G04 driver (according to your comment #2). Not sure why this happened. This is something for us to determine. If the G04 packages support your hardware (PCI ID: 10de:0640) as well, it should have worked. Have you rebooted in between?
AFAIR: No (it takes too much time)
This ticket has seen attempts with too many drivers: NVIDIA G03, G04, fbdev and nouveau. All this is confusing and obfuscating the goal.
I had simply installed the kernel update, and I had no video anymore. The rest are just attempts to be able to use the computer again (to be able to report the problem at all, for example).
Please 1. disable the NVIDIA repo, 2. deinstall whatever NVIDIA driver you've installed, 3. update the system to the current maintenance level doing 'zypper dup', 4. re-enable the NVIDIA repo and run 'zypper inr; zypper dup' This should give you the NVIDIA driver back - please let us know which ones... If it works now, 1. is fixed for this driver.
BTW: The surest thing to determine if X is basically working and lighting up your screen, is by logging in to the text console as root and do: '/usr/bin/Xorg -retro & sleep 3; xterm -display :0'. You should get an X screen with the old fashioned root weave and an xterm. To determine if there are issues with the openGL libraries (ie you have installed unsuitable ones) you do 'gxlinfo' in this xterm and observe the output and any messages.
OK, eventually I've got instructions (thanks!); I'll print them in case I have no video while testing... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c20
Egbert Eich
(In reply to Egbert Eich from comment #18)
Without knowledge of your DM we cannot find out where it logs to, though.
GNOME, so I guess gdm.
The screen usually goes into power save mode when it's idle for so long.
I know.
Moreover, this log file is for the fbdev driver which is unlikely to show a black screen if your console was working at all. Also it is even more unlikely that the screen is loosing the video signal when using this driver.
I was worried about the message: "Use of other console drivers including, but not limited to, vesafb, may result in corruption and stability problems, and is not supported."
Right. I see this as a disclaimer so that NVIDIA does not have to support and debug this if anything fails due to this.
Currently, I see two issues: 1. the G03 driver you've used reads a defective EDID extension block which is supplied by your monitor. So either the read procedure is wrong or the block sent by the monitor - in which case the NVIDIA driver should simply ignore it.
I suspected my KVM switch to need a reset (which I did in the meantime).
You should have done this right away.
This is something for NVIDIA to fix. 2. when you deinstall the G03 driver you will get the G04 driver (according to your comment #2). Not sure why this happened. This is something for us to determine. If the G04 packages support your hardware (PCI ID: 10de:0640) as well, it should have worked. Have you rebooted in between?
AFAIR: No (it takes too much time)
Now I'm getting slightly upset. You do not want to reboot the system because it takes too much time although the kernel driver has changed - but you expect me to either spend my company's or my spare time in supporting you? Sorry, I'm ending this here and now. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c21
Ulrich Windl
Now I'm getting slightly upset. You do not want to reboot the system because it takes too much time although the kernel driver has changed - but you expect me to either spend my company's or my spare time in supporting you?
Excuse, but why should I have rebooted the system when Yast did not tell me to do so? I suspected that things might change after a reboot, only that... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c31
--- Comment #31 from Ulrich Windl
This should give you the NVIDIA driver back - please let us know which ones...
These packages were in use: x11-video-nvidiaG03-340.96-40.1.x86_64 Wed 10 Feb 2016 02:07:13 PM CET nvidia-computeG03-340.96-40.1.x86_64 Wed 10 Feb 2016 02:07:07 PM CET nvidia-glG03-340.96-40.1.x86_64 Wed 10 Feb 2016 02:07:05 PM CET nvidia-uvm-gfxG03-kmp-default-340.96_k4.1.12_1-40.1.x86_64 Wed 10 Feb 2016 02:06:30 PM CET nvidia-gfxG03-kmp-default-340.96_k4.1.12_1-40.1.x86_64 Wed 10 Feb 2016 02:05:05 PM CET
If it works now, 1. is fixed for this driver.
I still have 800x600 resolution and an invalid EDID. It's hard to believe that the KVM reports an invalid EDID now; more likely the NVIDIA driver cannot read it correctly. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c32
--- Comment #32 from Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c33
--- Comment #33 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c34
--- Comment #34 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c35
--- Comment #35 from Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c36
Ulrich Windl
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c37
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=964429
http://bugzilla.opensuse.org/show_bug.cgi?id=964429#c38
--- Comment #38 from Egbert Eich
When switching back to nouveau driver (steps 1 and 2 of comment #18, plus reboot), I have full resolution. So I doubt the problem is the KVM switch.
Would you please remove the KVM switch and verify? -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com