Hi Quentin,
On Mon, 11 Nov 2024 at 05:55, Quentin Schulz <quentin.sch...@cherry.de> wrote:
>
> Hi Simon,
>
> On 11/11/24 1:41 PM, Simon Glass wrote:
> > Hi Quentin,
> >
> > On Mon, 11 Nov 2024 at 03:03, Quentin Schulz <quentin.sch...@cherry.de>
> > wrote:
> >>
> >> Hi Simon,
> >>
> >> On 11/7/24 10:31 PM, Simon Glass wrote:
> >>> When the --native flag is given, pretend to be running the host
> >>> architecture rather than sandbox.
> >>>
> >>> Allow the same control for PXE too.
> >>>
> >>
> >> Commit 7506c1566998 ("sandbox: Report host default-filename in native
> >> mode") breaks building jaguar-rk3588_defconfig, I believe it's this patch?
> >>
> >> Reverting it makes the build pass again.
> >>
> >> HOST_ARCH is empty in include/generated/version_autogenerated.h for some
> >> reason. Though it was also empty before.
> >>
> >> The error is:
> >>
> >> """
> >> lib/efi_loader/efi_helper.c:26:15: error: operator '==' has no left operand
> >> 26 | #if HOST_ARCH == HOST_ARCH_X86_64
> >> | ^~
> >> lib/efi_loader/efi_helper.c:29:17: error: operator '==' has no left operand
> >> 29 | #elif HOST_ARCH == HOST_ARCH_X86
> >> | ^~
> >> lib/efi_loader/efi_helper.c:32:17: error: operator '==' has no left operand
> >> 32 | #elif HOST_ARCH == HOST_ARCH_AARCH64
> >> | ^~
> >> lib/efi_loader/efi_helper.c:35:17: error: operator '==' has no left operand
> >> 35 | #elif HOST_ARCH == HOST_ARCH_ARM
> >> | ^~
> >> lib/efi_loader/efi_helper.c:38:17: error: operator '==' has no left operand
> >> 38 | #elif HOST_ARCH == HOST_ARCH_RISCV32
> >> | ^~
> >> lib/efi_loader/efi_helper.c:41:17: error: operator '==' has no left operand
> >> 41 | #elif HOST_ARCH == HOST_ARCH_RISCV64
> >> | ^~
> >> lib/efi_loader/efi_helper.c:45:2: error: #error Unsupported Host
> >> architecture
> >> 45 | #error Unsupported Host architecture
> >> | ^~~~~
> >> lib/efi_loader/efi_helper.c: In function ‘efi_get_basename’:
> >> lib/efi_loader/efi_helper.c:73:38: error: ‘HOST_BOOTEFI_NAME’ undeclared
> >> (first use in this function); did you mean ‘BOOTEFI_NAME’?
> >> 73 | return efi_use_host_arch() ? HOST_BOOTEFI_NAME :
> >> BOOTEFI_NAME;
> >> | ^~~~~~~~~~~~~~~~~
> >> | BOOTEFI_NAME
> >> lib/efi_loader/efi_helper.c:73:38: note: each undeclared identifier is
> >> reported only once for each function it appears in
> >> lib/efi_loader/efi_helper.c: In function ‘efi_get_pxe_arch’:
> >> lib/efi_loader/efi_helper.c:79:24: error: ‘HOST_PXE_ARCH’ undeclared
> >> (first use in this function); did you mean ‘HOST_ARCH’?
> >> 79 | return HOST_PXE_ARCH;
> >> | ^~~~~~~~~~~~~
> >> | HOST_ARCH
> >> lib/efi_loader/efi_helper.c: In function ‘efi_get_basename’:
> >> lib/efi_loader/efi_helper.c:74:1: warning: control reaches end of
> >> non-void function [-Wreturn-type]
> >> 74 | }
> >> | ^
> >> make[2]: *** [scripts/Makefile.build:256: lib/efi_loader/efi_helper.o]
> >> Error 1
> >> make[1]: *** [scripts/Makefile.build:398: lib/efi_loader] Error 2
> >> make: *** [Makefile:1914: lib] Error 2
> >> make: *** Waiting for unfinished jobs....
> >> """
> >>
> >> The build host for me is Fedora 41.
> >
> > Which cross-toolchain are you using? That variable is set at the top
> > of Makefile so it would be worth seeing what MK_ARCH is returning.
> >
>
> Ahah! I had forgotten I use ccache so my CROSS_COMPILE is "ccache
> aarch64-linux-gnu-", without "ccache" in there it works.
>
> I can suggest:
> """
> diff --git a/Makefile b/Makefile
> index 7275a02f24c..2aa115f2097 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -21,7 +21,7 @@ include include/host_arch.h
> ifeq ("", "$(CROSS_COMPILE)")
> MK_ARCH="${shell uname -m}"
> else
> - MK_ARCH="${shell echo $(CROSS_COMPILE) | sed -n
> 's/^[[:space:]]*\([^\/]*\/\)*\([^-]*\)-[^[:space:]]*/\2/p'}"
> + MK_ARCH="${shell echo $(CROSS_COMPILE) | sed -n 's/^\(ccache
> \)\?[[:space:]]*\([^\/]*\/\)*\([^-]*\)-[^[:space:]]*/\3/p'}"
> endif
> unexport HOST_ARCH
> ifeq ("x86_64", $(MK_ARCH))
> """
>
> an opinion on this or do I just send this as a patch and it can be
> discussed on the ML?
>
> This works for me both with and without ccache in CROSS_COMILE.
Yes, please send as a patch, looks OK to me.
I often wonder if we could have tests for the Makefile logic...
Regards,
Simon
