Re: Boot image for a Pi4 ?

From: Mark Millard <marklmi_at_yahoo.com>
Date: Sun, 07 Jul 2024 16:50:13 UTC
On Jul 7, 2024, at 09:21, bob prohaska <fbsd@www.zefox.net> wrote:

> On Sat, Jul 06, 2024 at 09:38:52PM -0700, Mark Millard wrote:
>> 
>> The RPi4B Rev 1.5 "C0T" wtih 8 GiBytes also lists 2018.
>> 
>> Capture the early serial console output from the
>> RPi* bootloader and firmware and look for lines
>> that contain one of (showing a 8 GiByte Rev 1.5
>> example: d03115):
>> 
>> boardrev d03115
>> 0x00d03115
>> RPI 4 Model B (0xd03115)
>> 
>> That last is actually from U-Boot.
>> The first was toward the end of the "BOOTMODE:" line.
>> The middle was the line after the "Read fixup4.dat bytes" line.
>> 
>> By contrast a 4GiByte Rev 1.1 has for those lines:
>> 
>> boardrev c03111
>> 0x00c03111
>> RPI 4 Model B (0xc03111)
> 
> U-boot reports
> RPI 4 Model B (0xd03114)

So, Rev 1.4 --and most likely a "B0T" part. 8 GiByte
started with Rev 1.4, nothing prior.

Note: The B0T or C0T text is part of what is printed
on the top of the SOC. There might have been some
Rev 1.4 with C0T parts. If I understand right, Rev
1.5 systematically has C0T parts.

> on both machines (8GB). Couldn't find the
> other output. Is the COT vs BOT difference
> still significant?

FreeBSD takes no advantage of "C0T" parts no longer
having the restrictions that the "B0T" parts have:
All RPi4B's are all treated as if they were "B0T"
parts for those issues. (Suboptimal but valid for
"C0T" parts.)

The EDK2 UEFI/ACPI for the RPi4B's does the same
sort of thing: Handle a "C0T" part as if it was a
"B0T" part. That in turn means that whatever OS
is using ACPI also would work that way.

(Plus standard FreeBSD does not handle the ACPI way
of dealing with "B0T" issues correctly, corrupting
some USB media I/O. My personal kernel builds have
the ACPI handling patched such that what EDK2 did
for presenting the "B0T" problem avoidance is also
handled, avoiding the corruptions.)

My stab at making FreeBSD changes in order to have
"C0T" parts avoid "B0T" ways of handling things
for the normal U-Boot style of booting failed.


===
Mark Millard
marklmi at yahoo.com