On 2024-12-16, Herman Rimm wrote:
> On Sun, Dec 15, 2024 at 12:04:06AM -0800, Vagrant Cascadian wrote:
>> On 2024-12-13, Herman Rimm wrote:
>> > * gnu/packages/firmware.scm (make-arm-trusted-firmware): Use guix-name.
>> > (arm-trusted-firmware-sun50i-a64): Return base package directly.
>> 
>> With this patch applied, I get:
>
> Is [PATCH 3/6] also applied?  What command do you use?

Apparently not, because I did not receive the patch!  I had noticed it
was missing, but misread your 0/6 comment about it...

Presumably etc/teams.scm did not include me in CC because it did not
match any of the file patterns I am on a team for... so I had to go
fishing in the bug tracker to get the full patch series. Kind of kills
most of the benefit of getting CCed on a patch series, in my opinion,
but that is outside the scope of this issue... :)

With 3/6 applied, it gets further... but still fails building on
aarch64-linux...


> Applying this patch series to 65892d9f20aa7a8649797507bb1ec867285083fb,
> creating a branch and setting up a channel for it, running this command
> on a x86_64 system succeeds:
>
> guix time-machine -C channels.scm -- build arm-trusted-firmware-rk3399

On aarch64-linux, with patches applied on top of
07b4b1d055c36c6c61d39273c26974771dbfe805, it still fails with errors
like this:

/gnu/store/d7q6yazfa1bxf4x7p9n3b99hn97q6bvw-bash-minimal-5.1.16/bin/sh: line 1: aarch64-none-elf-gcc: command not found
make: *** [lib/libfdt/libfdt.mk:23: /tmp/guix-build-arm-trusted-firmware-sun50i-a64-2.12.drv-0/source/build/sun50i_a64/debug/libfdt/fdt_strerror.o] Error 127

Maybe it builds correctly on x86_64, but given that most targets are
aarch64 targets, it ought to be able to build natively too.

As mentioned earlier, I would guess this is either a bug in upstream, or
in the way guix is passing a cross-toolchain of some kind...


live well,
  vagrant