You are not logged in.

#1 2016-05-27 17:52:33

GuloGuloDesu
Member
Registered: 2013-07-18
Posts: 30

ATI / NVIDIA video card crash

I just built a new system that contains 2 video cards (NVIDIA 980TI and Radeon R9 270). My end goal is to setup a video passthrough with qemu, but that is another issue altogether.

Unless I boot the system with nomodeset nouveau.modeset=0 then I have the screen go black and the system becomes unresponsive to the keyboard, and the screen switches to "No Signal"
The problem with booting the computer with the nomodeset is that I am unable to startx (fluxbox), also the resolution is terrible. I receive an error in /var/log/Xorg.0.log

(EE) open /dev/dri/card0: No such file or directory
(EE) Screen 0 deleted because of no matching config section.

Also with the nomodeset I receive the following error on boot:

error no ums support in radeon module

If I remove the nomodeset, and add radeon to Modules in mkinitcpio then I am able to boot the system, and I can startx. The problem that I have is that the video eventually crashes. When it crashes the screen goes black and the monitor says "No Signal". This will happen whether I am in X or at the console. Sometimes it even happens during boot.
I assumed that the crash was just video, so I attempted to type sudo reboot and try and reboot the computer. But the computer doesn't reboot until I press the power button or the reboot button. So I think it is a hard crash, but I haven't checked to see if the network stack is still working.
Also, I have the following error on boot:

radon 0000:02:00.0 invalid pci rom header signature: expecting 0xaa55, got 0xfff
nouveau 0000:03:00.0 unknown chipset (120080a1)

I have installed the xf86-video-ati drivers (had to before I could add the radeon to mkinitcpio). But I still have the same problem.

I have tried blacklisting nouveau:

vi /etc/modprobe.d/nouveau_blacklist.conf
blacklist nouveau

But still the same crashing problem, unless I use the nomodeset as a boot parameter.

I checked journalctl --since=today, but it doesn't have any crashes that I can see.

At this point, I am unsure as to where to go form here. I know that my R9 270 works with Arch, as I used this card in my last computer, but it was a standalone card. This time, I am using the NVIDIA which I will assign to a Windows VM when I get the system running and stable.
This is a fresh install with the following packages:

xorg-server 
xorg-xinit 
xorg-xev 
xorg-server-utils 
fluxbox 
wbar 
xterm 
wget
xf86-video-ati

Any help would be greatly appreciated. Even just pointing me to a log file that may have the crash info so that I can hunt down the problem.

I will continue to troubleshoot, hopefully locating the issue.

Offline

#2 2016-05-27 18:35:28

GuloGuloDesu
Member
Registered: 2013-07-18
Posts: 30

Re: ATI / NVIDIA video card crash

Alright, I am a dunce, I found the errors. I should have been running sudo journalctl --since=today:

pcieport 0000:00:02.0: AER: Uncorrected (Fatal) error received: id=0010
pcieport 0000:00:02.0: PCIe BUS Error: severity=Uncorrected (Fatal), type=transaction Layer,
pcieport 0000:00:02.0: device [8086:2f04] error status/mask=00000020/00000000
pcieport 0000:00:02.0: [ 5] Surprise Down Error (First)
pcieport 0000:00:02:0: [14] Completion Timeout
pcieport 0000:00:02.0: broadcast error_detected message
radeon 0000:02:00.0: device has no AER-aware driver
snd_hda_intel 0000:02:00.1: device has no AER aware driver

This only happens when I don't have nomodeset in the mkinitcpio or on boot options.

I'll start researching these errors, but any help is still appreciated.

Offline

#3 2016-05-27 19:17:19

GuloGuloDesu
Member
Registered: 2013-07-18
Posts: 30

Re: ATI / NVIDIA video card crash

Alright, so I tried pci=nomsi as a boot option, and it appears to be working.

I need to do some more research into nomsi to make sure that it doesn't break anything else.

I will wait until next week to mark this as solved in case someone else knows a better way of fixing this problem. I really don't like using fixes that I don't understand.

Offline

#4 2016-05-27 19:54:10

Ropid
Member
Registered: 2015-03-09
Posts: 501

Re: ATI / NVIDIA video card crash

You could try to disable MSI just for certain drivers instead of for everything. You can research a kernel module's parameters by doing for example "modinfo nvidia" or "modinfo snd_hda_intel". When I do that here, I see a parameter named "NVreg_EnableMSI" which probably would disable MSI for that 'nvidia' module, and for 'snd_hda_intel' I see something named "enable_msi". I'd then try lines like this in a file in /etc/modprobe.d/:

options nvidia NVreg_EnableMSI=0
options snd_hda_intel enable_msi=0

Then after restarting, what the drivers are doing can be seen in "cat /proc/interrupts".

Offline

#5 2016-05-31 20:32:42

GuloGuloDesu
Member
Registered: 2013-07-18
Posts: 30

Re: ATI / NVIDIA video card crash

Okay, so I am still having stability issues.
The problem now is that X keeps locking up and quits responding. I am able to move the mouse around, but I can't click on anything, or pull up any menus. It also appears that the keyboard is non-functional. As I am not able to change to a different TTY, nor am I able to CTRL-ALT-DEL to force a reboot. My only recourse has been to hit the reset button on the front of the computer. NUM lock and CAPS lock no longer work on the keyboard either.
This hahppens almost every time that I open Firefox, but it doesn't happen right away. Sometimes I can open a couple of web pages before it locks up.

I installed all updates, and still have the same problem.
I checked journalctl (sudo) and I didn't see anything in there about the lock ups.

I also noticed 2 other issues:
1. The screen resolution (at boot and console) somes times varies. I haven't made any settings changes.
2. Sometimes at the console or in XTerm not all letters show up. Sometimes they are replaced with a black box.

I checked the journal

I installed a bunch more packages:

Install
	grub-bios
	os-prober
	ifplugd
	net-tools
Video
	xorg-server
	xorg-xinit
	xorg-xev
	xorg-server-utils
	fluxbox
	wbar
	xf86-video-ati
	recordmydesktop
	gtk-recordmydesktop
	scrot
OS Tools
	sudo
	wget
	xterm
	unrar
	unzip
	p7zip
	gparted
	lm_sensors
	ethtool
	dosfstools
	nfs-utils
	fuseiso
	dnsutils
	usbip
	pcmanfm
Internet
	chromium
		ttf-dejavu
	firefox
	opera
	flashplugin
	transmission-cli
	rdesktop
	openconnect
Editors
	geany
	vim
	openscad
	kicad
		kicad-library
		kicad-library-3d
	eclipse-cpp
	ghex
Security
	openssh
	xscreensaver
	tor
	gnupg
	john
	steghide
Network
	wireshark-gtk
	nmap
	putty
	synergy
	sshfs
	traceroute
	gnu-netcat
	gpsd
	rtl-sdr

Offline

#6 2016-05-31 20:33:59

GuloGuloDesu
Member
Registered: 2013-07-18
Posts: 30

Re: ATI / NVIDIA video card crash

Ropid wrote:

You could try to disable MSI just for certain drivers instead of for everything. You can research a kernel module's parameters by doing for example "modinfo nvidia" or "modinfo snd_hda_intel". When I do that here, I see a parameter named "NVreg_EnableMSI" which probably would disable MSI for that 'nvidia' module, and for 'snd_hda_intel' I see something named "enable_msi". I'd then try lines like this in a file in /etc/modprobe.d/:

options nvidia NVreg_EnableMSI=0
options snd_hda_intel enable_msi=0

Then after restarting, what the drivers are doing can be seen in "cat /proc/interrupts".

Thanks for the response, I haven't tried this yet, as I am still having stability issues. Once I get the stability issues resolved, then I will play with this.

Offline

#7 2018-08-02 23:48:42

Cadeyrn
Member
Registered: 2013-04-06
Posts: 147

Re: ATI / NVIDIA video card crash

I'm having the exact same problems. It's not nouveau. I spent a while doing this VM setup flawlessly with an NVidia GT 740 as my basic desktop GPU and an AMD R9 Fury as my gaming GPU that swapped between host and vm, using nouveau and amdgpu respectively. 6 months ago, I stopped being able to game on Linux because nouveau suddenly stopped being willing to turn on DRI3, for no reason. One other person confirmed: it affects at least the GT 740 and 730. The bug report is already made and nouveau hasn't figured it out yet. None of us have even figured out the cause. It wasn't caused by any package upgrade (not even the kernel) or any change to system files. It was either magic, or something within the cards, I guess.

I gave up recently and swapped the GT 740 for an R7 240 running the radeon driver. That's when these exact same problems began for me, on a radeon/amdgpu system. And I've only gotten as far as you. Did you ever solve it? Or did you have to switch cards? I thought maybe it was synergy, because synergys dumped a bunch of errors whenever this 2nd type of post-nomsi freeze happened, but now the freeze just happened without synergy running. While I was typing this post, actually (had to retype it).

I think this was 2 problems in one, and we've both only solved the first one. journalctl can't find any errors anymore now, but before I disabled MSI interrupts, it always gave PCI errors on this freeze, and of course it changed to a different kind of freeze after too.

EDIT: Actually, for some reason, I have not experienced these:

GuloGuloDesu wrote:

I also noticed 2 other issues:
1. The screen resolution (at boot and console) somes times varies. I haven't made any settings changes.
2. Sometimes at the console or in XTerm not all letters show up. Sometimes they are replaced with a black box.

If it helps, here's the list of packages in your list that I *don't* have installed:

grub-bios
ifplugd
fluxbox
wbar
(gtk-)recordmydesktop
scrot
nfs-utils
fuseiso
dnsutils
usbip
pcmanfm
chromium
opera
transmission-cli
openconnect
openscad
kicad
eclipse-cpp
ghex
xscreensaver
tor
john
steghide
wireshark-gtk
putty
gnu-netcat
gpsd
rtl-sdr

Last edited by Cadeyrn (2018-08-03 00:03:33)

Offline

#8 2018-08-03 00:23:05

Ziusudra
Member
Registered: 2014-04-19
Posts: 119

Re: ATI / NVIDIA video card crash

Read this while you wait for the mods to arrive: https://wiki.archlinux.org/index.php/Co … bumping.22

Offline

#9 2018-08-03 00:31:31

jasonwryan
Anarchist
From: .nz
Registered: 2009-05-09
Posts: 26,860
Website

Re: ATI / NVIDIA video card crash

Closing


Arch + dwm   •   Mercurial repos  •   Github

Registered Linux User #482438

Online

Board footer

Powered by FluxBB