Hi Peter,
Am 02.03.22 um 15:19 schrieb Peter Robinson:
>>>> I wonder if it's a change/regression in the
firmware overlay <->
>>>> kernel interface.
>>> do you use the DTB file from 5.14.x or 5.16x?
>> I'm guessing this is fixed in the RPi firmware DTs with their rebase to
5.15.x?
>>
>> I've not pushed newer firmware/DT to stable Fedora of late as we've
>> had a bunch of regressions on various RPi devices, currently upstream
>> firmware on Fedora doesn't boot on any 32 bit RPi device for example.
> Please never ever use the DTS from the Raspberry Pi tree in combination
> with the mainline kernel. Please use the DTS from mainline tree. Any
> other combination is not tested and has unexpected results.
In the case of arm32 it's all using upstream kernel/DT with RPi
firmware it still has unexpected results. Either way it's a game of
whack-a-mole, we also have cases where the Firmware DT works and the
upstream kernel DT does not [1] but I've not had time to debug that
one yet.
i saw these "invalid bus width" report, but was never able to reproduce
it with my setup (upstream kernel + upstream DTS + RPI OS rootfs for 32
and 64 bit w/o u-boot or EFI). Are you able to reproduce it with any
BCM2711 platform?
Looks like the EFI leaves the SDHCI part in a unexpected state and
kernel doesn't properly handover.
Is there a good starting point which image i should try first?
Nevertheless these many possible sources of DTBs are a nightmare. At
least i plan to submit a patch to print a kernel warning in case of
missing gpio-ranges. So we can identify this issue faster.