Nouveau causes system freeze

kernel
nvidia
#1

Similar topic: Manjaro freezes - nouveau issue?

During normal work it happens that the system freezes, it often happens when I resize an application’s window.
This problem appears only with LTS kernel 4.14, on this machine I do not use non-LTS kernel.
As you can see below I have a Nvidia card and I use nouveau video drivers.

System:

System:    Host: manjaro-pc Kernel: 4.14.10-2-MANJARO x86_64 bits: 64 gcc: 7.2.1 Desktop: i3 4.14.1
           Distro: Manjaro Linux
Machine:   Device: desktop System: Gigabyte product: EP31-DS3L serial: N/A
           Mobo: Gigabyte model: EP31-DS3L v: x.x serial: N/A BIOS: Award v: F3 date: 04/09/2008
CPU:       Dual core Intel Core2 Duo E7300 (-MCP-) arch: Penryn rev.6 cache: 3072 KB
           flags: (lm nx sse sse2 sse3 sse4_1 ssse3) bmips: 10671
           clock speeds: max: 2667 MHz 1: 1600 MHz 2: 1600 MHz
Graphics:  Card: NVIDIA GK208 [GeForce GT 710B] bus-ID: 01:00.0
           Display Server: N/A driver: nvidia tty size: 179x51
Info:      Processes: 133 Uptime: 9 min Memory: 1114.4/1999.5MB Init: systemd Gcc sys: 7.2.1
           Client: Shell (zsh 5.4.2) inxi: 2.3.53

Log:

Jan 01 19:19:42 manjaro-pc kernel: nouveau 0000:01:00.0: gr: TRAP ch 2 [003fc0a000 systemd-logind[359]]
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: gr: GPC0/TPC0/TEX: 80000049
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: read fault at 0005800000 engine 00 [GR] client 01 [GPC0/T1_0] reason 02 [PTE] on channel 2 [003fc0a000 systemd-login
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: channel 2: killed
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: engine 0: scheduled for recovery
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: engine 6: scheduled for recovery
Jan 01 19:19:43 manjaro-pc kernel: nouveau 0000:01:00.0: systemd-logind[359]: channel 2 killed!
[...]
Jan 01 19:21:29 manjaro-pc kernel: nouveau 0000:01:00.0: compton[446]: failed to idle channel 5 [compton[446]]
Jan 01 19:21:29 manjaro-pc kernel: nouveau 0000:01:00.0: compton[446]: failed to idle channel 5 [compton[446]]
Jan 01 19:21:29 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: read fault at 0000013000 engine 07 [HOST0] client 07 [HOST_CPU] reason 02 [PTE] on channel 5 [003fb2d000 compton[446
Jan 01 19:21:29 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: channel 5: killed
Jan 01 19:21:29 manjaro-pc kernel: nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
Jan 01 19:21:29 manjaro-pc kernel: nouveau 0000:01:00.0: compton[446]: channel 5 killed!
Jan 01 19:21:57 manjaro-pc systemd-logind[359]: Failed to start autovt@tty2.service: Transport endpoint is not connected
Jan 01 19:21:59 manjaro-pc systemd-logind[359]: Failed to start autovt@tty3.service: Transport endpoint is not connected
Jan 01 19:22:01 manjaro-pc systemd-logind[359]: Failed to start autovt@tty2.service: Transport endpoint is not connected

The second part happens after I press poweroff button (immediately X disappears and TTY1 appears) then try to change TTY without success.

Anyone with similar experience?
Is this a kernel problem?
What type of information can I add to this report?
Should I fill an issue on the Linux kernel bugtracker?

#2

I am not an expert on xfce but have you switched the window compositor to compton or is this what Manjaro uses as standard?

#3

Is this your post on arch forums or just related issue? Also, post complete logs, snippets aren’t really useful.

#4

Just a coincidence, a useful coincidence, I’ll look into this tomorrow.
EDIT: in that topic the “solution” is with non-free driver.

Here you can find the complete system log.

#5

I wonder why your microcode is so old? I am not sure it is used for anything Nvidia though.
Mine is microcode updated early to revision 0x1c, date = 2015-02-26

#6

The processor is an Intel Core 2 Duo from 2008, I think this is the reason for the old microcode version.

#7

The problem persist on lats kernel 4.14 subrelease, I’m testing kernel 4.15 right now.

#8

I’ve changed every component, the only common component is the graphic card, so it is entirely a problem of Nouveau driver.
I have also changed desktop environment to KDE Plasma, so it is not a problem between Compton and Nouveau driver.
It surely happens with Linux 4.16, it happened once also on Linux 4.17 but it is still an RC version.
I could “git bisect”, but it is really a long work with the Linux kernel, furthermore the freeze is random and that increases the test time.