You are not logged in.

#1 2021-11-26 13:46:29

archilochus
Member
Registered: 2019-02-04
Posts: 4

Newer kernels and amdgpu

Hardware is an iMac20,1; graphics card is Navi 14 [Radeon RX 5500/5500M / Pro 5500M]. This ran pretty smoothly (I never got sound and wifi working, but that's OK for me) until kernel 5.11. I had to use the boot parameter amdgpu.dpm=0, and everything was stable. With kernel 5.12, the graphics driver would have problems; the boot process does not complete. I used the lts kernel, and that was OK. But since this will soon move to 5.15, I tried the mainline kernel and am having mixed success. I have to leave out the amdgpu.dpm=0 parameter, or it won't boot at all. Without it, the boot sequence is successful about 50 % of the time; I have not found a pattern. For an unsuccessful boot here's what I think are the relevant lines of the journal (last lines):

Nov 26 08:18:43 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: RAS: optional ras ta ucode is not available
Nov 26 08:18:43 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: RAP: optional rap ta ucode is not available
Nov 26 08:18:43 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: SECUREDISPLAY: securedisplay ta ucode is not available
Nov 26 08:18:43 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: use vbios provided pptable
Nov 26 08:18:43 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: smc_dpm_info table revision(format.content): 4.5
Nov 26 08:18:47 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: Failed to enable requested dpm features!
Nov 26 08:18:47 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: Failed to setup smc hw!
Nov 26 08:18:47 imac-archnew kernel: [drm:amdgpu_device_init.cold [amdgpu]] *ERROR* hw_init of IP block <smu> failed -62
Nov 26 08:18:47 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: amdgpu_device_ip_init failed
Nov 26 08:18:47 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: Fatal error during GPU init
Nov 26 08:18:47 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: amdgpu: finishing device.
Nov 26 08:18:47 imac-archnew kernel: [drm] free PSP TMR buffer
Nov 26 08:18:47 imac-archnew systemd-udevd[349]: Worker [381] terminated by signal 9 (KILL)
Nov 26 08:18:47 imac-archnew systemd-udevd[349]: 0000:03:00.0: Worker [381] failed
Nov 26 08:18:47 imac-archnew kernel: amdgpu: probe of 0000:03:00.0 failed with error -62
Nov 26 08:18:47 imac-archnew kernel: BUG: unable to handle page fault for address: ffffa8a4c21a1000
Nov 26 08:18:47 imac-archnew kernel: #PF: supervisor write access in kernel mode
Nov 26 08:18:47 imac-archnew kernel: #PF: error_code(0x0002) - not-present page

This is what it looks like at a successful boot:

Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: RAS: optional ras ta ucode is not available
Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: RAP: optional rap ta ucode is not available
Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: SECUREDISPLAY: securedisplay ta ucode is not available
Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: use vbios provided pptable
Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: smc_dpm_info table revision(format.content): 4.5
Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: SMU is initialized successfully!
Nov 26 08:20:09 imac-archnew kernel: [drm] Display Core initialized with v3.2.149!
Nov 26 08:20:09 imac-archnew kernel: [drm] REG_WAIT timeout 1us * 100000 tries - mpc2_assert_idle_mpcc line:479
Nov 26 08:20:09 imac-archnew kernel: snd_hda_intel 0000:03:00.1: bound 0000:03:00.0 (ops amdgpu_dm_audio_component_bind_ops [amdgpu])
Nov 26 08:20:09 imac-archnew kernel: [drm] kiq ring mec 2 pipe 1 q 0
Nov 26 08:20:09 imac-archnew kernel: [drm] VCN decode and encode initialized successfully(under DPG Mode).
Nov 26 08:20:09 imac-archnew kernel: [drm] JPEG decode initialized successfully.
Nov 26 08:20:09 imac-archnew kernel: kfd kfd: amdgpu: Allocated 3969056 bytes on gart
Nov 26 08:20:09 imac-archnew kernel: memmap_init_zone_device initialised 1048576 pages in 7ms
Nov 26 08:20:09 imac-archnew kernel: amdgpu: HMM registered 4080MB device memory
Nov 26 08:20:09 imac-archnew kernel: amdgpu: SRAT table not found
Nov 26 08:20:09 imac-archnew kernel: amdgpu: Virtual CRAT table created for GPU
Nov 26 08:20:09 imac-archnew kernel: amdgpu: Topology: Add dGPU node [0x7340:0x1002]
Nov 26 08:20:09 imac-archnew kernel: kfd kfd: amdgpu: added device 1002:7340
Nov 26 08:20:09 imac-archnew kernel: amdgpu 0000:03:00.0: amdgpu: SE 1, SH per SE 2, CU per SH 12, active_cu_number 20

I'm lost here - my guess would be that a previous shutdown leaves the gpu in a power state that the next boot doesn't recognize or know how to use. Can anyone provide any insights? Thanks!

Offline

#2 2021-11-26 15:22:16

seth
Member
Registered: 2012-09-03
Posts: 50,008

Re: Newer kernels and amdgpu

Try "iommu=soft"
Edit: https://www.phoronix.com/forums/forum/l … inux-5-4-7 - see comment #10 and also try "pci=noats" instead

Last edited by seth (2021-11-26 15:23:16)

Online

#3 2021-11-26 16:06:06

archilochus
Member
Registered: 2019-02-04
Posts: 4

Re: Newer kernels and amdgpu

Thank you! I will try both parameters and report back; I'll need several boot processes to see if it really works...

Offline

Board footer

Powered by FluxBB