[Bug ld/32491] [2.44 Regression] ld-elf/compress.exp test failures on arm-linux-gnueabi*

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32491

--- Comment #12 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> 
---
(In reply to Alan Modra from comment #11)
> Hmm, so to builld previously released versions of gcc with current binutils
> a user would need to apply a patch to gcc/configure.ac and regenerate
> gcc/configure, or modify config.h after configuring gcc.

Indeed. It would be a good idea to backport the GCC configure patch to all 
supported branches.

>  And as far as gas
> is concerned, gcc is asking it whether '.section .rodata.str, "aMS",
> @progbits, 1' assembles, which it now doesn't because everything past the
> '@' is a comment and gas provides 0 when get_absolute_expression parses an
> absent expression.
> 
> I think gas should accept the specific case of an absent entsize with "M",
> and assume 1, but reject a supplied 0.

I don't have an opinion about this, but I'll submit a patch to gas implementing
the change and it can be discussed on the mailing list.

Note that in any case, the GCC patch should still be committed. If nothing else
to avoid the bug proliferating to future configure checks via copy and paste.

-- 
You are receiving this mail because:
You are on the CC list for the bug.