Re: Succeeded to boot on Lenovo Yoga C630

On Dec 5, 2022, at 13:11, Warner Losh <imp@bsdimp.com> wrote:

> On Mon, Dec 5, 2022, 10:40 AM Hiroo Ono (小野寛生) <hiroo.ono+freebsd@gmail.com> wrote:
>> Hello,
>> 
>> I cannot find the original mail in my mailbox, but it is continued
>> from this mail.
>> https://freebsd-arm.freebsd.narkive.com/dBBAi0yX/loader-efi-does-not-boot-on-lenovo-yoga-c630
>> 
>> FreeBSD's bootaa64.efi that is distributed officially does not boot on
>> Lenovo Yoga C630.
>> OpenBSD 7.2's bootaa64.efi booted fine on Yoga C630, so I merged
>> OpenBSD's start.S and ldscript.arm64.
>> boot1.efi booted fine, but loader_lua.efi still needed to be tweaked.
>> 
>> It seems that probing on serial console freezes the loader.
>> Commenting out serialconsole made the loader_lua.efi to boot the kernel.
>> And then, the kernel stopped and complained that it cannot find the
>> device tree blob.
>> 
>> So my questions are:
>> 1. Can I disable loader from probing comconsole by some configuration?
>> (without tweaking the source.)
>> 2. How should I make the loader or kernel to find the dtb file?
> 
> There are some BIOSes that hate our serial code. So far it has just been in the cloud. But I'd just disable serial to confirm it's the same problem.
> 
> Also, the kernel is weird with both DTB and ACPI right now. You have to pick one and it defaults to dtb... and I have systems that require manually setting this to ACPI.
> 

I'll note that if one gets to the boot loader prompt, one can
try following loader command to figure out the ACPI vs. FDT
(Device Tree) type of boot, at least in many cases:

OK configuration
. . .
  ACPI 2.0 Table at 0xef830018
  . . .

vs. (output compare/contrast):

OK configuration
. . .
  FDT Table at 0x7ef0000
  . . .

Although, it is possible to have an environment that
provides with the ACPI Table and the FDT Table. In that
case, what is being used requires more information than
just configuration output.

===
Mark Millard
marklmi at yahoo.com