GPU not working due to amdgpu: failed to read ip discovery binary from file

Support
1

A terminal application running on the dedicated notebook GPU crashed and now, the GPU cannot be started during during the book process:

This is the relevant boot information. How can this be fixed?

Jan 16 23:33:32 localhost kernel: amdgpu 0000:03:00.0: amdgpu: get invalid ip discovery binary signature from vram
Jan 16 23:33:32 localhost kernel: amdgpu 0000:03:00.0: amdgpu: amdgpu_discovery is not set properly
Jan 16 23:33:32 localhost kernel: amdgpu 0000:03:00.0: amdgpu: failed to read ip discovery binary from file
Jan 16 23:33:32 localhost kernel: [drm:amdgpu_discovery_set_ip_blocks [amdgpu]] ERROR amdgpu_discovery_init failed
Jan 16 23:33:32 localhost kernel: amdgpu 0000:03:00.0: amdgpu: Fatal error during GPU init
Jan 16 23:33:32 localhost kernel: amdgpu 0000:03:00.0: amdgpu: amdgpu: finishing device.
Jan 16 23:33:32 localhost kernel: amdgpu: probe of 0000:03:00.0 failed with error -22
Jan 16 23:33:32 localhost kernel: checking generic (fc20000000 300000) vs hw (fc20000000 10000000)
Jan 16 23:33:32 localhost kernel: checking generic (fc20000000 300000) vs hw (fc20000000 10000000)
Jan 16 23:33:32 localhost kernel: fb0: switching to amdgpu from EFI VGA
Jan 16 23:33:32 localhost kernel: Console: switching to colour dummy device 80x25
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: vgaarb: deactivate vga console
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: enabling device (0006 → 0007)
Jan 16 23:33:32 localhost kernel: [drm] initializing kernel modesetting (RENOIR 0x1002:0x1638 0x1462:0x1316 0xC4).
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: Trusted Memory Zone (TMZ) feature enabled
Jan 16 23:33:32 localhost kernel: [drm] register mmio base: 0xFC900000
Jan 16 23:33:32 localhost kernel: [drm] register mmio size: 524288
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 0 <soc15_common>
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 1 <gmc_v9_0>
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 2 <vega10_ih>
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 3
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 4
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 5
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 6 <gfx_v9_0>
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 7 <sdma_v4_0>
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 8 <vcn_v2_0>
Jan 16 23:33:32 localhost kernel: [drm] add ip block number 9 <jpeg_v2_0>
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: Fetched VBIOS from VFCT
Jan 16 23:33:32 localhost kernel: amdgpu: ATOM BIOS: 113-CEZANNE-018
Jan 16 23:33:32 localhost kernel: [drm] VCN decode is enabled in VM mode
Jan 16 23:33:32 localhost kernel: [drm] VCN encode is enabled in VM mode
Jan 16 23:33:32 localhost kernel: [drm] JPEG decode is enabled in VM mode
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: PCIE atomic ops is not supported
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: MODE2 reset
Jan 16 23:33:32 localhost kernel: [drm] vm size is 262144 GB, 4 levels, block size is 9-bit, fragment size is 9-bit
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: VRAM: 512M 0x000000F400000000 - 0x000000F41FFFFFFF (512M used)
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: GART: 1024M 0x0000000000000000 - 0x000000003FFFFFFF
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: AGP: 267419648M 0x000000F800000000 - 0x0000FFFFFFFFFFFF
Jan 16 23:33:32 localhost kernel: [drm] Detected VRAM RAM=512M, BAR=512M
Jan 16 23:33:32 localhost kernel: [drm] RAM width 128bits DDR4
Jan 16 23:33:32 localhost kernel: [drm] amdgpu: 512M of VRAM memory ready
Jan 16 23:33:32 localhost kernel: [drm] amdgpu: 3072M of GTT memory ready.
Jan 16 23:33:32 localhost kernel: [drm] GART: num cpu pages 262144, num gpu pages 262144
Jan 16 23:33:32 localhost kernel: [drm] PCIE GART of 1024M enabled.
Jan 16 23:33:32 localhost kernel: [drm] PTB located at 0x000000F400900000
Jan 16 23:33:32 localhost kernel: amdgpu 0000:07:00.0: amdgpu: PSP runtime database doesn’t exist

This is the dedicated GPU: ]$ inxi -G
Graphics:
Device-1: AMD Navi 22 [Radeon RX 6700/6700 XT/6750 XT / 6800M] driver: N/A
Device-2: AMD Cezanne driver: amdgpu v: kernel
Device-3: Acer HD Webcam type: USB driver: uvcvideo
Display: wayland server: X.Org v: 1.21.1.3 with: Xwayland v: 21.1.3
compositor: gnome-shell v: 40.10 driver: X: loaded: modesetting
unloaded: fbdev dri: radeonsi gpu: amdgpu resolution: 1920x1080~240Hz
API: OpenGL v: 4.6 Mesa 22.1.5 renderer: AMD RENOIR (LLVM 14.0.6 DRM 3.42
5.14.0-70.30.1.el9_0.x86_64)

2

lol, have you considered an nvidia gpu, this amd one seems to be giving you a lot of hassle! :laughing:

3

I need it for simulation software development/adaption. AMD GPUs have a lot more fp64 performance which is what I need. Unfortunately there are only two notebook models with an AMD GPU with more than 8Gb RAM and I may have picked the wrong one considering the number of BIOS updates and issues like this one.

I made quite some progress on the software development side in the past couple of days so that’s going into the right direction.

1 Like
4

[The Fix]

The MSI Customer Support provided me with a download link to a clean AMD Windows driver installer (they only support Windows but fortunately I have a dual boot setup). It went immediately into a “repair” mode and must have fixed the issue on the mainboard.

3-4 Linux boots later the Notebook was back to normal.

5

The problem is back both on Windows 10 and Linux and is now permanent.

It’s back in the hands of the MSI Customer Support.