Cannot have Radeon 5500 XT wotking with kernel 5.9

[feanorelf]

Hallo,

I have been building a LFS based on the current development version with systemd (kernel 5.9.11) and now, one step at a time, I am reconfiguring the kernel to suit my hw.
I have a ASUS mb with chipset z77, an Ivy Bridge 3770 CPU and the Radeon 5500 XT which is being the thorn in my side.
I followed the instructions in https://wiki.gentoo.org/wiki/AMDGPU, downloaded the Navi14 firmware and compiled it with the kernel and what I get now is the screen freezing at boot after showing, as last notification:

fb0: switching to amdgpudrmfb from EFI VGA

journalctl shows errors related to amdgpu from this point on, but looking on the internet, other people could boot notwithstanding these mistakes.

I ave also found suggestion about boot parameters, but none of these worked.

Here are some lines from journalctl after switching to amdgpu

Nov 25 13:14:21 feanorlfs kernel: [drm] amdgpu kernel modesetting enabled.
Nov 25 13:14:21 feanorlfs kernel: CRAT table not found
Nov 25 13:14:21 feanorlfs kernel: Virtual CRAT table created for CPU
Nov 25 13:14:21 feanorlfs kernel: amdgpu: Topology: Add CPU node
Nov 25 13:14:21 feanorlfs kernel: checking generic (e0000000 300000) vs hw (e0000000 10000000)
Nov 25 13:14:21 feanorlfs kernel: fb0: switching to amdgpudrmfb from EFI VGA
Nov 25 13:14:21 feanorlfs kernel: Console: switching to colour dummy device 80x25
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: vgaarb: deactivate vga console
Nov 25 13:14:21 feanorlfs kernel: [drm] initializing kernel modesetting (NAVI14 0x1002:0x7340 0x1DA2:0xE421 0xC5).
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: amdgpu: Trusted Memory Zone (TMZ) feature disabled as experimental (default)
Nov 25 13:14:21 feanorlfs kernel: [drm] register mmio base: 0xF7D00000
Nov 25 13:14:21 feanorlfs kernel: [drm] register mmio size: 524288
Nov 25 13:14:21 feanorlfs kernel: [drm] PCIE atomic ops is not supported
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 0 <nv_common>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 1 <gmc_v10_0>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 2 <navi10_ih>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 3 <psp>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 4 <smu>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 5 <dm>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 6 <gfx_v10_0>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 7 <sdma_v5_0>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 8 <vcn_v2_0>
Nov 25 13:14:21 feanorlfs kernel: [drm] add ip block number 9 <jpeg_v2_0>
Nov 25 13:14:21 feanorlfs kernel: resource sanity check: requesting [mem 0x000c0000-0x000dffff], which spans more than PCI Bus 0000:00 [mem 0x000d0000-0x000d3fff window]
Nov 25 13:14:21 feanorlfs kernel: caller pci_map_rom+0x79/0x1c0 mapping multiple BARs
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: No more image in the PCI ROM
Nov 25 13:14:21 feanorlfs kernel: amdgpu: ATOM BIOS: 113-E4210MB-U0D
Nov 25 13:14:21 feanorlfs kernel: [drm] VCN decode is enabled in VM mode
Nov 25 13:14:21 feanorlfs kernel: [drm] VCN encode is enabled in VM mode
Nov 25 13:14:21 feanorlfs kernel: [drm] JPEG decode is enabled in VM mode
Nov 25 13:14:21 feanorlfs kernel: [drm] vm size is 262144 GB, 4 levels, block size is 9-bit, fragment size is 9-bit
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: amdgpu: VRAM: 8176M 0x0000008000000000 - 0x00000081FEFFFFFF (8176M used)
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: amdgpu: GART: 512M 0x0000000000000000 - 0x000000001FFFFFFF
Nov 25 13:14:21 feanorlfs kernel: [drm] Detected VRAM RAM=8176M, BAR=256M
Nov 25 13:14:21 feanorlfs kernel: [drm] RAM width 128bits GDDR6
Nov 25 13:14:21 feanorlfs kernel: [TTM] Zone kernel: Available graphics memory: 4034110 KiB
Nov 25 13:14:21 feanorlfs kernel: [TTM] Zone dma32: Available graphics memory: 2097152 KiB
Nov 25 13:14:21 feanorlfs kernel: [TTM] Initializing pool allocator
Nov 25 13:14:21 feanorlfs kernel: [TTM] Initializing DMA pool allocator
Nov 25 13:14:21 feanorlfs kernel: [drm] amdgpu: 8176M of VRAM memory ready
Nov 25 13:14:21 feanorlfs kernel: [drm] amdgpu: 5909M of GTT memory ready.
Nov 25 13:14:21 feanorlfs kernel: [drm] GART: num cpu pages 131072, num gpu pages 131072
Nov 25 13:14:21 feanorlfs kernel: [drm] PCIE GART of 512M enabled (table at 0x0000008000300000).
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: amdgpu: failed to init sos firmware
Nov 25 13:14:21 feanorlfs kernel: [drm:psp_sw_init] *ERROR* Failed to load psp firmware!
Nov 25 13:14:21 feanorlfs kernel: [drm:amdgpu_device_init.cold] *ERROR* sw_init of IP block <psp> failed -22
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: amdgpu: amdgpu_device_ip_init failed
Nov 25 13:14:21 feanorlfs kernel: amdgpu 0000:03:00.0: amdgpu: Fatal error during GPU init
Nov 25 13:14:21 feanorlfs kernel: [drm] amdgpu: finishing device.
Nov 25 13:14:21 feanorlfs kernel: ------------[ cut here ]------------

Here the full boot log: https://drive.google.com/file/d/1H3s...ew?usp=sharing

Thank you for helping.

[baryluk]

I have no expirience with LFS, but

https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/tree/

has the firmware.

You might need to put it in the /lib/firmware/

I am not sure if you lack firmware, or it is old, or what.

Maybe it needs to be in initramfs too? I am not sure.


Be sure to have 'Firmware loading facility' enabled in the kernel config.

[feanorelf]

Problem solved.
I do not know for what reason, but the files downloaded from
https://git.kernel.org/pub/scm/linux...it/tree/amdgpu
updated on Nov. 19, 2020, do not work with my system.
I have downloaded the latest tarball of linux firmware, which is dated Nov. 18, and it works.
Do not ask me why...
Thank you all.