amdgpu panics

Tue Apr 7 08:27:47 UTC 2020

On 06/04/2020 23:49, Hans Petter Selasky wrote:
> On 2020-04-07 00:07, Grzegorz Junka wrote:
>>
>> Is it possible to at least gather some debug info where this happens? 
>> I don't think there is any core dumped if the system doesn't panic?
>
> Can you SSH to this machine and get dmesg?
>

I sent the dmesg after booting privately as it was quite long. One 
interesting thing I just noticed is that the halt is not a complete 
halt. The system responds to ping and an ssh user session was active, in 
the sense that I could do ls -l and get a response. But it hung as soon 
as I tried su. Same with initiating any new ssh session - the system 
responds with prompt for password but after that nothing happens.

This is the content of the messages log starting at the moment when I 
try to load the modules:

Apr  7 07:54:30 venus kernel: [drm] amdgpu kernel modesetting enabled.
Apr  7 07:54:30 venus kernel: drmn0: <drmn> on vgapci0
Apr  7 07:54:30 venus kernel: vgapci0: child drmn0 requested pci_enable_io
Apr  7 07:54:30 venus syslogd: last message repeated 1 times
Apr  7 07:54:30 venus kernel: [drm] initializing kernel modesetting 
(VEGA10 0x1002:0x687F 0x1002:0x0B36 0xC0).
Apr  7 07:54:30 venus kernel: [drm] register mmio base: 0xFD100000
Apr  7 07:54:30 venus kernel: [drm] register mmio size: 524288
Apr  7 07:54:30 venus kernel: [drm] PCI I/O BAR is not found.
Apr  7 07:54:30 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_gpu_info.bin
Apr  7 07:54:30 venus kernel: [drm] probing gen 2 caps for device 
1022:1471 = 700d03/e
Apr  7 07:54:30 venus kernel: [drm] probing mlw for device 1002:687f = 
400d03
Apr  7 07:54:30 venus kernel: [drm] UVD is enabled in VM mode
Apr  7 07:54:30 venus kernel: [drm] UVD ENC is enabled in VM mode
Apr  7 07:54:30 venus kernel: [drm] VCE enabled in VM mode
Apr  7 07:54:30 venus kernel: ATOM BIOS: 113-D0500500-104
Apr  7 07:54:30 venus kernel: [drm] vm size is 262144 GB, 4 levels, 
block size is 9-bit, fragment size is 9-bit
Apr  7 07:54:30 venus kernel: drmn0: VRAM: 8176M 0x000000F400000000 - 
0x000000F5FEFFFFFF (8176M used)
Apr  7 07:54:30 venus kernel: drmn0: GTT: 256M 0x000000F600000000 - 
0x000000F60FFFFFFF
Apr  7 07:54:30 venus kernel: Successfully added WC MTRR for 
[0xe0000000-0xefffffff]: 0;
Apr  7 07:54:30 venus kernel: [drm] Detected VRAM RAM=8176M, BAR=256M
Apr  7 07:54:30 venus kernel: [drm] RAM width 2048bits HBM
Apr  7 07:54:30 venus kernel: [TTM] Zone  kernel: Available graphics 
memory: 33495488 kiB
Apr  7 07:54:30 venus kernel: [TTM] Zone   dma32: Available graphics 
memory: 2097152 kiB
Apr  7 07:54:30 venus kernel: [TTM] Initializing pool allocator
Apr  7 07:54:30 venus kernel: [drm] amdgpu: 8176M of VRAM memory ready
Apr  7 07:54:30 venus kernel: [drm] amdgpu: 8176M of GTT memory ready.
Apr  7 07:54:30 venus kernel: i_size_write unimplemented
Apr  7 07:54:30 venus kernel: [drm] GART: num cpu pages 65536, num gpu 
pages 65536
Apr  7 07:54:30 venus kernel: [drm] PCIE GART of 256M enabled (table at 
0x000000F400800000).
Apr  7 07:54:31 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_sos.bin
Apr  7 07:54:32 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_asd.bin
Apr  7 07:54:32 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_acg_smc.bin
Apr  7 07:54:33 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_pfp.bin
Apr  7 07:54:33 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_me.bin
Apr  7 07:54:34 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_ce.bin
Apr  7 07:54:34 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_rlc.bin
Apr  7 07:54:35 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_mec.bin
Apr  7 07:54:35 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_mec2.bin
Apr  7 07:54:35 venus kernel: i_size_write unimplemented
Apr  7 07:54:35 venus syslogd: last message repeated 9 times
Apr  7 07:54:36 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_sdma.bin
Apr  7 07:54:36 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_sdma1.bin
Apr  7 07:54:36 venus kernel: [drm] use_doorbell being set to: [true]
Apr  7 07:54:36 venus kernel: i_size_write unimplemented
Apr  7 07:54:36 venus kernel: [drm] use_doorbell being set to: [true]
Apr  7 07:54:36 venus kernel: i_size_write unimplemented
Apr  7 07:54:37 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_uvd.bin
Apr  7 07:54:37 venus kernel: [drm] Found UVD firmware Version: 65.29 
Family ID: 17
Apr  7 07:54:37 venus kernel: [drm] PSP loading UVD firmware
Apr  7 07:54:37 venus kernel: i_size_write unimplemented
Apr  7 07:54:37 venus syslogd: last message repeated 2 times
Apr  7 07:54:37 venus kernel: drmn0: successfully loaded firmware image 
with name: amdgpu/vega10_vce.bin
Apr  7 07:54:37 venus kernel: [drm] Found VCE firmware Version: 57.4 
Binary ID: 4
Apr  7 07:54:37 venus kernel: [drm] PSP loading VCE firmware
Apr  7 07:54:37 venus kernel: i_size_write unimplemented
Apr  7 07:54:37 venus syslogd: last message repeated 2 times
Apr  7 07:54:38 venus kernel: [drm] Display Core initialized with v3.1.27!
Apr  7 07:54:38 venus kernel: [drm] Connector DP-1: get mode from tunables:
Apr  7 07:54:38 venus kernel: [drm]   - kern.vt.fb.modes.DP-1
Apr  7 07:54:38 venus kernel: [drm]   - kern.vt.fb.default_mode
Apr  7 07:54:38 venus kernel: [drm] Connector DP-2: get mode from tunables:
Apr  7 07:54:38 venus kernel: [drm]   - kern.vt.fb.modes.DP-2
Apr  7 07:54:38 venus kernel: [drm]   - kern.vt.fb.default_mode
Apr  7 07:54:38 venus kernel: [drm] Connector DP-3: get mode from tunables:
Apr  7 07:54:38 venus kernel: [drm]   - kern.vt.fb.modes.DP-3
Apr  7 07:54:38 venus kernel: [drm]   - kern.vt.fb.default_mode

GrzegorzJ