[Bug other/123841] [16 regression] Failed bootstrap (Error: invalid operands (.text and .text.unlikely sections) for `-') since r16-7073

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123841

--- Comment #9 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
> Trying:
> --- gcc/acinclude.m4.jj 2026-01-27 22:03:07.406974596 +0100
> +++ gcc/acinclude.m4    2026-01-27 22:06:46.560267046 +0100
> @@ -478,7 +478,7 @@ AC_DEFUN([gcc_GAS_FLAGS],
>      as_32_opt=-m32
>      as_64_opt=-m64
>      ;;
> -  i?86-*-*:yes | x86_64-*-*:yes )
> +  i?86-*-*:* | x86_64-*-*:* )
>      dnl Override the default, which may be incompatible.
>      case "$target" in
>        i?86-*-*)

same here.

> now, for solaris2 I think this should make no difference, because the
> *solaris2*:no case matches earlier, for linux restore previous behavior (at
> least, so far both i686-linux and x86_64-linux stage1 auto-host.h is the same

Me too, stage2 as well.  The differences to the gas/gld 2.45 builds from
before on Ubuntu 24.04 with the bundled gas/gld 2.39 are minimaland
expected.

> with the patch as was yesterday).  Unsure about other x86 targets though (and
> whether that makes a difference).
> If it would make a difference, then perhaps
>   i?86-*-solaris2*:* | x86_64-*-solaris2*:* | i?86-*-linux*:* |
> x86_64-*-linux*:* )

Maybe we should go the other way round here: just do it for all
i?86/x86_64 targets except those known not to need it (i.e. Darwin,
maybe more)?  That's why I also tested FreeBSD and Darwin.

But obviously it's crucial to get Linux/x86 back on track first; we can
worry about the rest later if need be.