[PATCH] Re: loading of zeros into {x,y,z}mm registers (take 2)

[Jakub Jelinek]

On Mon, Dec 04, 2017 at 01:15:46AM -0700, Jan Beulich wrote:
> >>> On 02.12.17 at 01:13, <jakub@redhat.com> wrote:
> > On Fri, Dec 01, 2017 at 02:54:28PM +0100, Jakub Jelinek wrote:
> >> Will try this:
> > 
> > That failed to bootstrap, but here is an updated version that passed
> > bootstrap/regtest on x86_64-linux and i686-linux, ok for trunk?
> 
> Thanks for the quick fix. A related question though: The issue
> could have been spotted (in the provided example) by gas, if
> only it had been given enough information (i.e. that AVX512VL
> insns are fine to use, but AVX512DQ ones aren't). Is there a
> reason gcc doesn't pass any -march= to gas, nor issues any
> .arch directives?

Because we've never done that on i?86/x86_64, so it would confuse users,
plus there can be instructions in inline asm, there is target attribute,
function multiversioning, etc.

A year or more ago I've been regtesting with hacks to let the assembler
complain, I just remember it wasn't that easy, but don't remember the
details anymore.

	Jakub