Re: RPI4-4Gb Boot fails from SD card (both my build AND the official 14.3 build)

Now this is interesting.....

The image I have been running (and the one apparently in the official 
releases -- I'll have to go look but I mimic'd what it was doing for 
ARM64 on the Pis) was using the u-boot binary out of the rpi-4 package.

If I copy in the u-boot.bin file out of the rpi-3 package the older Pi4 
I have here boots and runs.  The u-boot.bin files are different in both 
size and content despite allegedly being the same version.

I do not have (yet) one of the newer Pi4s to test with but do have one 
coming, so I'll check that one as well -- and in the meantime, well, 
guess where I'll be pulling the u-boot binary from in my builds since 
this one works with both the "3" and "4"s that I have.....

On 7/27/2025 08:10, Karl Denninger wrote:
> On 7/26/2025 23:40, Mark Millard wrote:
>> On Jul 26, 2025, at 18:24, Karl Denninger<karl@denninger.net> wrote:
>>
>>> Well, the same card (physically the same card) boots in the Pi3 with the same FreeBSD load *and* the same card loaded with an "official" Pi firmware load boots in the 4 and runs.  I have attempted this with two very different SD cards -- one Sandisk Ultra that previously had Kodi on it and was running in that specific Pi4 and the second is one that while slower is one of a dozen here of a series that have served me extremely well in Pis for a very, very long time.
>>> The builds on my box (and apparently the one on the FreeBSD official image page) are all using 2025-04 u-boot as the official FreeBSD load off the distribution page behaves exactly the same way -- the EFI loader boots without incident, gets and loads the kernel, the kernel boots as expected but the mmc driver takes timeouts and cannot see the sd card after the kernel comes up, thus root mount fails.
>> Summary: The below booted from a microsd card just fine
>> on one system --but not on another-- based on:
>>
>> FreeBSD-14.3-STABLE-arm64-aarch64-RPI-20250724-f0a7a1bda375-272016.img.xz
>>
>> It failed for a rather old 4 GByte RPi4B that I got access
>> to (older EEPROM content as well) --but worked on a modern
>> 8 GiByte RPi4B.
>
> Lovely.  So from appearances there's a hardware difference that 
> apparently results in this; this is an older one (quite old) which is 
> the only "4" I have right now.  I guess it get relegated to Kodi use 
> (if ever); I'll see about sourcing some newer ones (whether I borked 
> it when I updated its firmware as Kodi wanted or whether its rot 
> somewhere in the FreeBSD driver code that breaks for this hardware but 
> not newer stuff I don't know.)
>
> I'll play with some older u-boot copies I have laying around but since 
> the EFI loader and then kernel starts I'll wager that ain't the issue 
> and see about sourcing a couple of the newer ones to play with for the 
> purpose I'm screwing with these in the first place.
>
>> ===
>> Mark Millard
>> marklmi at yahoo.com
>>
>>
> -- 
> Karl Denninger
> karl@denninger.net
> /The Market Ticker/
> /[S/MIME encrypted email preferred]/
-- 
Karl Denninger
karl@denninger.net
/The Market Ticker/
/[S/MIME encrypted email preferred]/