Bug 187111

Summary: nouveau 0000:00:10.0: fifo: CACHE_ERROR - ch 0 [DRM] subc 3 mthd 0db4 data 00000000
Product: Drivers Reporter: Mathieu Malaterre (mathieu.malaterre)
Component: Video(Other)Assignee: drivers_video-other
Status: NEW ---    
Severity: normal CC: g.yudintsev, glex.spb, Martin.vGagern
Priority: P1    
Hardware: PPC-32   
OS: Linux   
URL: https://bugs.debian.org/842971
Kernel Version: 4.7.8 Subsystem:
Regression: No Bisected commit-id:

Description Mathieu Malaterre 2016-11-07 06:30:59 UTC
Every time I load `nouveau` with `modprobe nouveau` on tty console the screeen become unresponsive, i.e. the change of tty with Ctrl+Alt+F[N] has no effect and Ctrl+Alt+Del is the only thing I can do.
I'm on an iMac G4 PowerMac6,1 with an NVIDIA GeForce2 MX/MX 400.
Comment 1 Mathieu Malaterre 2016-11-07 06:31:42 UTC
The full dmesg is available at: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=842971#23
Comment 2 Mathieu Malaterre 2016-11-09 07:55:16 UTC
Update the original dmesg was bogus, the correct one is here:

https://bugs.debian.org/cgi-bin/bugreport.cgi?att=1;bug=842971;filename=syslog;msg=59

[...]
Nov  9 07:55:13 debian systemd[1]: Started System Logging Service.
Nov  9 07:55:13 debian kernel: [   33.630577] [TTM] Initializing pool allocator
Nov  9 07:55:13 debian kernel: [   33.630659] nouveau 0000:00:10.0: DRM: VRAM: 63 MiB
Nov  9 07:55:13 debian kernel: [   33.630675] nouveau 0000:00:10.0: DRM: GART: 128 MiB
Nov  9 07:55:13 debian kernel: [   33.630696] nouveau 0000:00:10.0: DRM: BMP version 5.16
Nov  9 07:55:13 debian kernel: [   33.630714] nouveau 0000:00:10.0: DRM: No DCB data found in VBIOS
Nov  9 07:55:13 debian kernel: [   33.631863] nouveau 0000:00:10.0: DRM: Adaptor not initialised, running VBIOS init tables.
Nov  9 07:55:13 debian kernel: [   33.637612] nouveau 0000:00:10.0: DRM: Saving VGA fonts
Nov  9 07:55:13 debian kernel: [   33.697218] nouveau 0000:00:10.0: DRM: No DCB data found in VBIOS
Nov  9 07:55:13 debian kernel: [   33.699824] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
Nov  9 07:55:13 debian kernel: [   33.699867] [drm] Driver supports precise vblank timestamp query.
Nov  9 07:55:13 debian kernel: [   33.705945] nouveau 0000:00:10.0: DRM: MM: using M2MF for buffer copies
Nov  9 07:55:13 debian kernel: [   33.739537] nouveau 0000:00:10.0: fifo: DMA_PUSHER - ch 0 [DRM] get 80086078 put 0000016c state c000016c (err: MEM_FAULT) push 00000000
Nov  9 07:55:13 debian kernel: [   33.739625] nouveau 0000:00:10.0: fifo: DMA_PUSHER - ch 0 [DRM] get 0000016c put 00000188 state 80000000 (err: INVALID_CMD) push 00000000
Nov  9 07:55:13 debian kernel: [   33.739661] nouveau 0000:00:10.0: DRM: allocated 1024x768 fb: 0x8000, bo cfadcc00
Nov  9 07:55:13 debian kernel: [   33.856422] Console: switching to colour frame buffer device 128x48
Nov  9 07:55:13 debian kernel: [   33.856488] nouveau 0000:00:10.0: fifo: DMA_PUSHER - ch 0 [DRM] get 00000188 put 000001bc state 80000000 (err: INVALID_CMD) push 00000000
Nov  9 07:55:13 debian kernel: [   33.856948] nouveau 0000:00:10.0: fifo: DMA_PUSHER - ch 0 [DRM] get 000001bc put 000009ec state 80000000 (err: INVALID_CMD) push 00000000
[...]
Nov  9 07:55:13 debian kernel: [   35.583472] nouveau 0000:00:10.0: fb0: nouveaufb frame buffer device
Nov  9 07:55:13 debian kernel: [   35.601195] [drm] Initialized nouveau 1.3.1 20120801 for 0000:00:10.0 on minor 0
Comment 3 Mathieu Malaterre 2016-11-09 07:56:55 UTC
Maybe related: https://bugs.freedesktop.org/show_bug.cgi?id=70390
Comment 4 Gleb Mazovetskiy 2022-12-06 00:16:09 UTC
Still happens on kernel 6.0.10 apparently
Comment 5 Gleb Mazovetskiy 2022-12-06 00:18:06 UTC
There is a potential patch in the last message of this thread: https://ubuntuforums.org/archive/index.php/t-2307087.html
Comment 6 dk777 2023-03-01 03:53:39 UTC
Hello.

Similar messages appeared in the system log (i.e.

nouveau 0000:00:0d.0: fifo: CACHE_ERROR - ch 0 [kworker/0:0[7]] subc 2 mthd 0130 data 00000000

) after updating the linux kernel to version 6.2.1 .

At the same time Sway and other applications run and work fine. But the problem is that it spams these messages very often and a lot.

System log:
1:
https://0x0.st/HzcX.txt
https://zerobin.net/?4f2acc4ca3de195e#rEwA6nZ3tNUMPun2bTlwtw7+7M/v178j3IqNnTUzqNE=

2 (larger in size, more similar messages):
https://0x0.st/Hzc8.txt
https://zerobin.net/?cf345e58fc119a14#j+ZWG6/okKeSGDGhjNOFO9OCOM0+SHFH34Fcc6XqUf0=

I don't see this kind of behavior on kernel versions 6.1.12 and 6.1.14.

I wrote about this problem to the arch linux bugtracker:
https://bugs.archlinux.org/task/77670

The first comment did not confirm this behavior.
Maybe it happens only on some hardware.

Distribution: Arch Linux
The kernel version on which this occurs: 6.2.1 (package linux-6.2.1-arch1-1).
CPU architecture: x86_64
GPU: nvidia geforce 7025 / nforce 630a (lspci)
GPU driver: nouveau