amdgpu: error results in black screen

[Alexander88207]

I would recommend to upgrade to FreeBSD 15.

There you have drm 6.1, 6.6 and 6.9.

You can also install the new kernels with the new DRM alongside to take a peek without having to upgrade.

 

[tingo]

Heh, I was going to write that this error happens so infrequently that I can live with it, and now it happened again already

Mar 10 03:52:20 kg-core2 kernel: [drm ERROR :amdgpu_job_timedout] ring gfx timeout, signaled seq=88004248, emitted seq=88004250
Mar 10 03:52:20 kg-core2 kernel: [drm ERROR :amdgpu_job_timedout] Process information: process  pid 100275 thread  pid 100275
Mar 10 03:52:20 kg-core2 kernel: drmn0: GPU recovery disabled.

(yes, I know, I shouldn't jinx it)
relevant packages

root@kg-core2:~ # pkg info gpu-firmware-kmod\*
gpu-firmware-kmod-20250109,1
root@kg-core2:~ # pkg info drm\*kmod\*
drm-510-kmod-5.10.163.1305000_13

FreeBSD details

root@kg-core2:~ # freebsd-version -ku
13.5-RELEASE-p10
13.5-RELEASE-p10
root@kg-core2:~ # uname -a
FreeBSD kg-core2.kg4.no 13.5-RELEASE-p10 FreeBSD 13.5-RELEASE-p10 GENERIC amd64

 

[lizzias]

if you are running xlibre and amdgpu you need to

sudo pkg remove xlibre-xf86-video-scfb

and then reboot.

 

[tingo]

happened again.

May  3 14:44:28 kg-core2 kernel: [drm ERROR :amdgpu_job_timedout] ring gfx timeout, signaled seq=77674704, emitted seq=77674706
May  3 14:44:28 kg-core2 kernel: [drm ERROR :amdgpu_job_timedout] Process information: process  pid 100278 thread  pid 100278
May  3 14:44:28 kg-core2 kernel: drmn0: GPU recovery disabled.

packages

root@kg-core2:~ # pkg info gpu-firmware-kmod\*
gpu-firmware-kmod-20250109,1
root@kg-core2:~ # pkg info drm\*kmod\*
drm-510-kmod-5.10.163.1305000_13

FreeBSD

root@kg-core2:~ # freebsd-version -ku
13.5-RELEASE-p13
13.5-RELEASE-p14
root@kg-core2:~ # uname -a
FreeBSD kg-core2.kg4.no 13.5-RELEASE-p13 FreeBSD 13.5-RELEASE-p13 GENERIC amd64

 

[tingo]

and it happened again

May 16 03:52:28 kg-core2 kernel: drmn0: GPU fault detected: 147 0x01cc2001
May 16 03:52:28 kg-core2 kernel: drmn0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x0E13DE0E
May 16 03:52:28 kg-core2 kernel: drmn0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0C020001
May 16 03:52:28 kg-core2 kernel: drmn0: VM fault (0x01, vmid 6) at page 236183054, read from '' (0x00000000) (32)
May 16 03:52:28 kg-core2 kernel: [drm ERROR :amdgpu_job_timedout] ring gfx timeout, signaled seq=573291009, emitted seq=573291011
May 16 03:52:28 kg-core2 kernel: [drm ERROR :amdgpu_job_timedout] Process information: process  pid 100283 thread  pid 100283
May 16 03:52:28 kg-core2 kernel: drmn0: GPU recovery disabled.

at the time, these were the relevant packages

root@kg-core2:~ # pkg info gpu-firmware-kmod\*
gpu-firmware-kmod-20250109,1
root@kg-core2:~ # pkg info drm\*kmod\*
drm-510-kmod-5.10.163.1305000_13

and the machine was running

root@kg-core2:~ # freebsd-version -ku
13.5-RELEASE-p13
13.5-RELEASE-p14
root@kg-core2:~ # uname -a
FreeBSD kg-core2.kg4.no 13.5-RELEASE-p13 FreeBSD 13.5-RELEASE-p13 GENERIC amd64

that's all for now.

 

[ArgentoSoma]

amdgpu on FreeSBD has very broken for years. I have a RX 580 and I can't use on FreeBSD for this. graphics/drm-61-kmod or graphics/drm-latest-kmod on FreeBSD15 maybe help you.

 

[freebsdiz4me]

I had no issues with Xlibre and the Xlibre version of the amdgpu driver. I just needed to put in a conf file for each monitor.