[PATCH 3/3] install: Replace scripts/output-format.sed with objdump -f [BZ #26559]

[Fāng-ruì Sòng]

On Tue, Jan 12, 2021 at 4:00 AM Adhemerval Zanella
<adhemerval.zanella@linaro.org> wrote:
>
>
>
> On 12/01/2021 08:20, Florian Weimer wrote:
> > * Adhemerval Zanella via Libc-alpha:
> >
> >> On 28/12/2020 16:48, Fangrui Song via Libc-alpha wrote:
> >>> GNU ld and gold have supported --print-output-format since 2011. glibc
> >>> requires binutils>=2.25 (2015), so if LD is GNU ld or gold, we can
> >>> assume the option is supported.
> >>>
> >>> lld is by default a cross linker supporting multiple targets. It auto
> >>> detects the file format and does not need OUTPUT_FORMAT. It does not
> >>> support --print-output-format.
> >>>
> >>> By parsing objdump -f, we can support all the three linkers.
> >>
> >> LGTM and this change seems orthogonal to the other lld adjustments.
> >> I have checked that at least with a build for a handful of supported
> >> ABIs I see no difference with ld.bfd (with multiple versions).
> >
> > I see a cross-build failure during the initial static link bootstrap on
> > aarch64-linux-gnu during “make install”:
> >
> > make[2]: Entering directory '/home/bmg/src/glibc'
> > aarch64-glibc-linux-gnu-gcc   -shared  -Wl,-dynamic-linker=/lib/ld-linux-aarch64.so.1 \
> >           -Wl,-z,combreloc -Wl,-z,relro -Wl,--hash-style=both  \
> >           -x c /dev/null -o /home/bmg/build/compilers/aarch64-linux-gnu/glibc/aarch64-linux-gnu/format.lds.so
> > /home/bmg/install/compilers/aarch64-linux-gnu/lib/gcc/aarch64-glibc-linux-gnu/10.2.1/../../../../aarch64-glibc-linux-gnu/bin/ld: cannot find crti.o: No such file or directory
> > /home/bmg/install/compilers/aarch64-linux-gnu/lib/gcc/aarch64-glibc-linux-gnu/10.2.1/../../../../aarch64-glibc-linux-gnu/bin/ld: cannot find crtn.o: No such file or directory
> > collect2: error: ld returned 1 exit status
> > make[2]: *** [Makerules:1070: /home/bmg/build/compilers/aarch64-linux-gnu/glibc/aarch64-linux-gnu/format.lds] Error 1
> >
> > This affects other targets as well.
> >
> > I guess using
> >
> >   $(CC) -nostdlib -nostartfiles -shared -x assembler /dev/null
> >
> > should work on most targets here.
>
> I haven't see it when a stage2 gcc, make install works as intended.  However I
> did see it now with a stage1 gcc with build-many-glibcs.py.
>
> The -nostdlib -nostartfiles seems to fix it indeed, I am checking with a
> bootstrap using build-many-glibcs.py.

Thanks for the fix 0400f928335a5e04c788e1c831d8825d42612c49!