You are not logged in.
Pages: 1
Hi Everyone,
I have a new hardware on which I'd like to install Arch, but the boot media goes into a black srceen at early boot after selecting the linux cmd line.
I added "loglevel=7" to the cmdline to get more details, sorry I don't know how to grab logs for install media boot errors, so I just have pictures:
After this last line
amdgpu 0000:65:00.0: Direct firmware load for amdgpu/isp_4_1_0.bin with error -2
the screen goes black and nothing more happens.
I don't think the system loads in the background, if I plug an ethernet cable, I can't find the device on the network.
The device has an AMD APU Ryzen AI 9 HX 370 (with 890M iGPU), I'm using the latest (12-01) install media with 6.12.1 kernel.
Any idea, should I try to disable some modules ?
Thank you.
Offline
Try whether adding nomodeset to the kernel params helps you here. If that boots the install media, you potentially want to opt for linux-firmware-git and linux-mainline for HW support on the actual install.
Offline
Just to give an update, the install media did boot correctly with the nomodeset flag.
I installed Arch with standards kernel/firmwares at first, and got the exact same error at system startup. I hoped newer versions might help.
So I am now building linux-firmware-git and linux-mainline but it takes forever, I'll give the result tomorrow.
I thought this new AMD CPU was supported since kernel 6.11...
Offline
There's a repo if you just want to test functionality/driver support: https://wiki.archlinux.org/title/Unoffi … ries#miffe
Offline
I used miffe's repo to install linux-mainline, and I built linux-firmware-git, and the same issue happens at system boot.
Searching for information on this recent CPU, I found a thread where someone has the exact same issue, with the same CPU but different hardware :
https://gitlab.freedesktop.org/drm/amd/-/issues/3744
He managed to spot that the hardware works fine on a specific older version of linux-firmware, so I installed Arch's linux-firmware-20240809.59460076-1-any.pkg (found on Archive), and boom, it boots !
There is a discussion with one of the developpers suggesting this might be a BIOS issue, if that's the case I would'nt expect it to be fixed anytime soon, I guess I'll stick with this version for now and follow the thread to see if it gets fixed in the kernel-firmware or by a BIOS update.
I don't think they spotted the culprit commit yet, I could also build the current linux-firmware without the commit...
Offline
Pages: 1