kern/58994: System panic with amdgpu and startx

[kikadf.01%gmail.com@localhost]

>Number:         58994
>Category:       kern
>Synopsis:       System panic with amdgpu and startx
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Jan 15 09:25:00 +0000 2025
>Originator:     Robert Bagdan
>Release:        10.1
>Organization:
-
>Environment:
NetBSD yamato 10.1 NetBSD 10.1 (GENERIC) #0: Mon Dec 16 13:08:11 UTC 2024  mkrepro%mkrepro.NetBSD.org@localhost:/usr/src/sys/arch/amd64/compile/GENERIC amd64
>Description:
I have a PC with AMD Ryzen 7700X cpu, and AMD RX550 gpu. If I start NetBSD with the loaded amdgpu driver, when I use startx the system is crashing. 

My boot entry from boot.cfg:
menu=Boot with amdgpu:rndseed /var/db/entropy-file;load drmkms_sched;load amdgpu;boot

The gpu relevant parts from dmesg:
[     1.018418] amdgpu0 at pci1 dev 0 function 0: ATI Technologies Radeon 540/540X/550/550X / RX 540X/550/550X (rev. 0xc7)
[     2.732670] [drm] initializing kernel modesetting (POLARIS12 0x1002:0x699F 0x1002:0x0B04 0xC7).
[     2.732670] [drm] register mmio base: 0xF6F00000
[     2.732670] [drm] register mmio size: 262144
[     2.732670] [drm] PCIE atomic ops is not supported
[     2.732670] [drm] add ip block number 0 <vi_common>
[     2.732670] [drm] add ip block number 1 <gmc_v8_0>
[     2.732670] [drm] add ip block number 2 <tonga_ih>
[     2.732670] [drm] add ip block number 3 <gfx_v8_0>
[     2.732670] [drm] add ip block number 4 <sdma_v3_0>
[     2.732670] [drm] add ip block number 5 <powerplay>
[     2.741941] [drm] add ip block number 6 <dm>
[     2.741941] [drm] add ip block number 7 <uvd_v6_0>
[     2.741941] [drm] add ip block number 8 <vce_v3_0>
[     2.741941] ATOM BIOS: xxx-xxx-xxx
[     2.741941] [drm] UVD is enabled in VM mode
[     2.751729] [drm] UVD ENC is enabled in VM mode
[     2.751729] [drm] VCE enabled in VM mode
[     2.751729] [drm] vm size is 256 GB, 2 levels, block size is 10-bit, fragment size is 9-bit
[     2.751729] amdgpu0: VRAM: 2048M 0x000000F400000000 - 0x000000F47FFFFFFF (2048M used)
[     2.751729] amdgpu0: GART: 256M 0x000000FF00000000 - 0x000000FF0FFFFFFF
[     2.751729] [drm] Detected VRAM RAM=2048M, BAR=4096M
[     2.762173] [drm] RAM width 64bits GDDR5
[     2.762173] Zone  kernel: Available graphics memory: 9007199248635332 KiB
[     2.762173] Zone   dma32: Available graphics memory: 2097152 KiB
[     2.762173] [drm] amdgpu: 2048M of VRAM memory ready
[     2.762173] [drm] amdgpu: 3072M of GTT memory ready.
[     2.762173] [drm] GART: num cpu pages 65536, num gpu pages 65536
[     2.762173] [drm] PCIE GART of 256M enabled (table at 0x000000F400300000).
[     2.772190] amdgpu0: interrupting at msi10 vec 0 (amdgpu0)
[     2.772190] [drm] Chained IB support enabled!
[     2.772190] hwmgr_sw_init smu backed is polaris10_smu
[     2.772190] [drm] Found UVD firmware Version: 1.130 Family ID: 16
[     2.781274] [drm] Found VCE firmware Version: 35.1a Binary ID: 3
[     2.871273] [drm] DM_PPLIB: values for Engine clock
[     2.871273] [drm] DM_PPLIB:  214000
[     2.881273] [drm] DM_PPLIB:  551000
[     2.891272] [drm] DM_PPLIB:  734000
[     2.891272] [drm] DM_PPLIB:  980000
[     2.901272] [drm] DM_PPLIB:  1000000
[     2.911272] [drm] DM_PPLIB:  1020000
[     2.911272] [drm] DM_PPLIB:  1046000
[     2.921272] [drm] DM_PPLIB:  1071000
[     2.931272] [drm] DM_PPLIB: Validation clocks:
[     2.931272] [drm] DM_PPLIB:    engine_max_clock: 107100
[     2.941272] [drm] DM_PPLIB:    memory_max_clock: 150000
[     2.951272] [drm] DM_PPLIB:    level           : 8
[     2.951272] [drm] DM_PPLIB: values for Memory clock
[     2.961271] [drm] DM_PPLIB:  300000
[     2.971271] [drm] DM_PPLIB:  625000
[     2.971271] [drm] DM_PPLIB:  1500000
[     2.981271] [drm] DM_PPLIB: Validation clocks:
[     2.991271] [drm] DM_PPLIB:    engine_max_clock: 107100
[     2.991271] [drm] DM_PPLIB:    memory_max_clock: 150000
[     3.001271] [drm] DM_PPLIB:    level           : 8
[     3.011271] [drm] Display Core initialized with v3.2.69!
[     3.061270] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
[     3.061270] [drm] Driver supports precise vblank timestamp query.
[     3.101269] [drm] UVD and UVD ENC initialized successfully.
[     3.201268] [drm] VCE initialized successfully.
[     3.211268] amdgpufb0 at amdgpu0
[     3.221268] [drm] Initialized amdgpu 3.36.0 20150101 for amdgpu0 on minor 0
[     3.231268] amdgpufb0: framebuffer at 0xfb00830000, size 1920x1080, depth 32, stride 7680
[     3.331266] wsdisplay0 at amdgpufb0 kbdmux 1: console (default, vt100 emulation), using wskbd0

All dmesg output: https://pastebin.com/raw/PtCdSVAR

Xorg log when crashing (all): https://pastebin.com/raw/7brJjEh1

When the os is crashing, I got a dumping error, so /var/crash is empty:
dump ld0: I/O error
i/o error

I don't know this dump error are related with the amdgpu crash, but:
#nvmectl devlist
 nvme0: WD Blue SN580 2TB
    nvme0ns1 (1907729MB)
When I use NetBSD I don't experience problems with the disk, dmesg not show any issues I think.

I take a photo from my screen, when amdgpu is crashing: https://imgur.com/vJQkClM

>How-To-Repeat:
Load amdgpu kernel module, run startx.
>Fix: