GMP 5.0.5 build fails with tcc due to x86_64 asm (unknown register...)

[Richard Guenther]

On Tue, 26 Jun 2012, Torbjorn Granlund wrote:

> Richard Guenther <rguenther at suse.de> writes:
> 
>   But the failure mode is assembling assembler source - GMP should
>   arrange to not use the compiler driver to assemble but to call
>   an assembler directly for assembler source.
> 
> Are you volunteering to do this work?

No ;)

> It will be a massive undertaking to bring things back to the current
> level of functionality.
> 
> Why is that?  Because you will need to pass lots of different ISA and
> ABI slecting options to the assemblers.  And many systems have several
> "as" which ar incompatible.
>
> You should expect to spend at least one month on this project.
> 
>   (yes, the compiler
>   driver knows which assembler to use, so it's certainly convenient
>   to use it - but I can see people using a compiler with a built-in
>   assembler that expects Intel assembler operand ordering instead
>   of AT&T?)
>   
> I can "see" many hypothetical problems too.  Since I co-maintain a
> complex piece of software, I need to spend my time on real problems,
> though...

It was just a suggestion - if I'd implement it I would default to
the compiler driver for the assembler call but allow a different
assembler to be specified via configure.  tcc people can then
tell configure to use gcc as assembler.

Richard.

-- 
Richard Guenther <rguenther at suse.de>
SUSE / SUSE Labs
SUSE LINUX Products GmbH - Nuernberg - AG Nuernberg - HRB 16746
GF: Jeff Hawn, Jennifer Guild, Felix Imend