PC Drivers & Software

[zceemja]

I have reference RX6800XT with vbios 020.001.000.038.015719, that fails to load driver on both linux and fresh windows 10 install two ryzen zen3 systems, but runs on intel system.

Main system under test: Asrock AB350M PRO4 (BIOS P7.40) with Ryzen 9 5900X

Windows AMD driver show error 43. Tested on linux with kernel 6.6.8 & 6.1.69, both throwing some error:

amdgpu 0000:09:00.0: amdgpu: smu driver if version = 0x00000040, smu fw if version = 0x00000041, smu fw program = 0, version = 0x003a5800 (58.88.0)
amdgpu 0000:09:00.0: amdgpu: SMU driver if version not matched
amdgpu 0000:09:00.0: amdgpu: use vbios provided pptable
...skipping...
amdgpu 0000:09:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000006 SMN_C2PMSG_82:0x00000000
amdgpu 0000:09:00.0: amdgpu: Failed to enable requested dpm features!
amdgpu 0000:09:00.0: amdgpu: Failed to setup smc hw!
[drm:amdgpu_device_init [amdgpu]] *ERROR* hw_init of IP block <smu> failed -62
amdgpu 0000:09:00.0: amdgpu: amdgpu_device_ip_init failed
amdgpu 0000:09:00.0: amdgpu: Fatal error during GPU init

BIOS settings:

• CSM: disabled & enabled (tested both)
• Above 4G decoding & ReBar: disabled
• SVM & IOMMU: disabled & enabled (tested both)

Also tested on:

1. Asus ROG STRIX B450-F GAMING (BIOS: tba) with Ryzen 7 5800X on windows 10: same problem, error 43
2. ASRock H170M-ITX/AC (BIOS: P1.1 I should really update that) with i5-6400 on windows 10: driver loads without issues, GPU seem to work without any issue.

So there seems to be some AMD firmware/bios related issue, or GPU related issue that intel has some workaround? I will test on more systems and update this post.

[mengelag]

Why do you have above 4g decoding and rebar set to off?

[zceemja]

Im guessing it shouldn't affect system in any way, worst case it would just make it less compatible. Correct me if Im wrong.

[Ryzen_Warrior]

I have been having the same problem booting into Linux but Windows does work, I've spend hours trying to figure out what the problem is and I found this post. I have a 6950xt that I just got with the "smu if driver does not match" error and when I boot it's a black screen and I am unable to use the terminal. I am able to boot into x11 with "nomodeset" but wayland will not login and I'm limited to a low resolution

Dec 31 08:41:36 GamingPC kernel: hid-steam 0003:28DE:1142.0009: Steam wireless receiver connected
Dec 31 08:41:36 GamingPC kernel: usb 1-6: reset full-speed USB device number 4 using xhci_hcd
Dec 31 08:41:36 GamingPC kernel: usb 1-6: reset full-speed USB device number 4 using xhci_hcd
Dec 31 08:41:36 GamingPC kernel: usb 1-6: reset full-speed USB device number 4 using xhci_hcd
Dec 31 08:41:36 GamingPC kernel: amdgpu 0000:2f:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000006 SMN_C2PMSG_82:0x00000000
Dec 31 08:41:36 GamingPC kernel: amdgpu 0000:2f:00.0: amdgpu: Failed to enable requested dpm features!
Dec 31 08:41:36 GamingPC kernel: amdgpu 0000:2f:00.0: amdgpu: Failed to setup smc hw!
Dec 31 08:41:36 GamingPC kernel: [drm:amdgpu_device_init.cold [amdgpu]] *ERROR* hw_init of IP block <smu> failed -62
Dec 31 08:41:36 GamingPC kernel: amdgpu 0000:2f:00.0: amdgpu: amdgpu_device_ip_init failed
Dec 31 08:41:36 GamingPC kernel: amdgpu 0000:2f:00.0: amdgpu: Fatal error during GPU init
Dec 31 08:41:36 GamingPC kernel: amdgpu 0000:2f:00.0: amdgpu: amdgpu: finishing device.
Dec 31 08:41:36 GamingPC kernel: ------------[ cut here ]------------
Dec 31 08:41:36 GamingPC kernel: WARNING: CPU: 8 PID: 204 at drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c:655 amdgpu_irq_put+0x45/0x70 [amdgpu]
Dec 31 08:41:36 GamingPC kernel: Modules linked in: hid_steam hid_generic usbhid hid amdgpu(+) gpu_sched drm_buddy video i2c_algo_bit drm_display_helper sd_mod cec rc_core drm_ttm_helper ahci ttm libahci xhci_pci drm_kms_helper xhci_hcd nvme libata drm nvme_core r8169 mxm_wmi crc32_pclmul usbcore t10_pi crc32c_intel realtek mdio_devres scsi_mod i2c_piix4 libphy crc64_rocksoft crc64 scsi_common crc_t10dif usb_common crct10dif_generic crct10dif_pclmul crct10dif_common wmi button
Dec 31 08:41:36 GamingPC kernel: CPU: 8 PID: 204 Comm: (udev-worker) Not tainted 6.1.0-16-amd64 #1  Debian 6.1.67-1

[zceemja]

What motherboard & bios version are you using? nomodeset tells kernel not load any video driver so wayland would end up rendered by CPU instead of GPU. Its something similar to vgasave mode on windows.

[zceemja]

Tested on Gigabyte X670 GAMING X AX rev. 1.0 (BIOS: F20) with Ryzen 9 7900X on Windows 11 with latest driver & linux 6.6.8, also does not work.

Also found similar issue reported on linux driver with RX 6600 XT and 12th gen Intel https://gitlab.freedesktop.org/drm/amd/-/issues/1955. Next will try to run on some more intel systems.

[zceemja]

Tested on MSI Z97 GAMING 5 (BIOS: 1.D) with Intel i7-4790k on fresh windows 10. So far I can reliably make GPU run with following steps:

1. Run AMD Cleanup Utility, reboot to safe mode and cleanup all amd software
2. Reboot normally, make sure GPU runs with Microsoft basic adapter driver in device manager
3. Install AMD driver, once loaded it works fine, HDMI and other ports work fine, performance is great.

Now from there on driver may or may not load with error 43 on next reboot. Seems like having display connected over USB-C dock (Im using Thinkpad Docking Station) works mostly consistent with driver working after reboot/power cycle. Connecting HDMI monitor seems to break driver at boot and make it completely stop working until repeating steps 1 to 3, but this is not consistent. 

I also managed to boot linux and load driver but it not always works and Im going to try find some pattern and try other hardware.