You are not logged in.
Pages: 1
Hi everybody!
After updating to nvidia-dkms-358.16-2.1 I've found a really strange and unwanted behaviour when switching between TTYs and/or resuming after suspend.
Though I am able to log in with sddm and everything seems to work fine, if I switch to text tty, then switch back, my display becomes black and I am unable to switch back to the text tty. I am not even able to change a numlock state of a keyboard, so i treat is as "Xorg hangs so it doesn't receive any key presses". I wonder how can I check it to be sure..
Here is what journalctl -f tells me:
Nov 28 20:39:17 desktop systemd[1]: Started Getty on tty2.
Nov 28 20:39:34 desktop kernel: nvidia-modeset: ERROR: GPU:0: Idling EVO timed out: 0x0000857d:0:0:0x00000040
And after that "top" shows me "kwin_x11" consuming 100% of a CPU core. Or "Xorg" doing the same, if I kill kwin_x11. But actually I'm not able to kill neither kwin_x11, nor Xorg, even with kill -9.
So "/home" partition can not be safely unmounted, and systemd ends up with skipping this part after waiting for ~3 mins. The same goes for "/tmp".
Here is a journal after stopping sddm:
Nov 28 20:59:48 desktop systemd[1]: Stopping Simple Desktop Display Manager...
Nov 28 20:59:48 desktop sddm[533]: Signal received: SIGTERM
Nov 28 20:59:48 desktop sddm[533]: Socket server stopping...
Nov 28 20:59:48 desktop sddm[533]: Socket server stopped.
Nov 28 20:59:48 desktop sddm[533]: Display server stopping...
Nov 28 20:59:53 desktop sddm[533]: QProcess: Destroyed while process ("/usr/lib/sddm/sddm-helper") is still running.
Nov 28 20:59:53 desktop sddm[533]: Display server stopping...
Nov 28 20:59:58 desktop sddm[533]: QProcess: Destroyed while process ("/usr/bin/X") is still running.
Normally I use kde (plasma5) + sddm, but the same happens with lightdm+xfce.
Passing "nomodeset" to a kernel changes nothing.
Kernel:
Linux desktop 4.2.5-1-ARCH #1 SMP PREEMPT Tue Oct 27 08:13:28 CET 2015 x86_64 GNU/Linux
Nvidia stuff:
multilib/lib32-libvdpau 1.1.1-1
Nvidia VDPAU library
multilib/lib32-nvidia-libgl 358.16-1
NVIDIA drivers libraries symlinks (32-bit)
multilib/lib32-nvidia-utils 358.16-1
NVIDIA drivers utilities (32-bit)
extra/libvdpau 1.1.1-1
Nvidia VDPAU library
community/nvidia-cg-toolkit 3.1-3
NVIDIA Cg libraries
extra/nvidia-dkms 358.16-2.1
NVIDIA driver sources for linux
local/nvidia-hook 2.3-1
mkinitcpio hook to compile the NVIDIA modules
extra/nvidia-libgl 358.16-1
NVIDIA drivers libraries symlinks
extra/nvidia-utils 358.16-1
NVIDIA drivers utilities
Xorg is of version 1.18.0-3
Now I'll try to downgrage Xorg and nvidia drivers and see what happens.
I've found a problem which can be related to mine: https://bbs.archlinux.org/viewtopic.php?id=205460, but I wonder if the root of the problem is the same.
Last edited by mexus (2015-11-28 18:40:01)
Offline
I've downgraded xorg to a previous version: xorg-server-1.17.4-2 and xorg-server-common-1.17.4-2. Although I've been forced to downgrade evdev as well: xf86-input-evdev-2.9.1-1.
Well, for some reason, I wasn't able to test it, because with this configuration xorg can't find any drivers so i'm unable to use mouse/keyboard. It's a shame, but I don't have any time to investigate it further at least at this point.
Offline
BTW I'll really appreciate any info regarding how can I find the real issue and where can I post an "official" bug report!
Thank you!
Offline
I had a very similar issue to what you described. When I would put the computer to sleep with "systemctl hybrid-sleep", > 50% of the time it would just show the same error that you mentioned upon attempting to awake the computer.
nvidia-modeset: ERROR: GPU:0: Idling EVO timed out: 0x0000857d:0:0:0x00000040
Aside from displaying the error, the system was completely non-responsive.
Are you getting errors in your log similar to the ones listed here? https://wiki.archlinux.org/index.php/NV … CPI_events
I was getting these errors, and although the wiki page claims that they are benign, I was suspicious that perhaps the Nvidia driver was not restarting properly after a sleep due not handling ACPI events properly. I installed the "acpid" package since then I haven't had any issues.
Disclaimer: The issue was intermittent, albeit frequent, before I installed "acpid", so it could still be present and I perhaps just haven't had the bad luck to trigger it.
Offline
Nevermind. It just happened to me again. I guess "acpid" wasn't the issue. I am going to file a bug now.
Offline
@isomarcte,
I've installed nvidia-352 from AUR (and all related stuff), and it works like a charm. Maybe i've already forgotten something, but i can't recall any ACPI related errors, and this shit happend to me each and every time, without any exceptions.
Though i didn't try to "hybrid-sleep", only "suspend".
But the main question is, do you have any issues with switching around text and graphical TTYs?
Offline
Pages: 1