This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Fix PR inline-asm/16194, register and clobbers
On Mon, 2005-01-17 at 21:37 -0800, Richard Henderson wrote:
> On Mon, Jan 17, 2005 at 09:33:00PM -0700, Jeffrey A Law wrote:
> > Which makes me wonder if we ought to be doing something like
> > while (handled_component_p (val) || TREE_CODE (val) == INDIRECT_REF)
> > val = TREE_OPERAND (val, 0)
> Alternately, we could use a variant of is_gimple_min_lval to
> force this down to "*tmp", where tmp is not a hard register.
Not a terribly bad idea. It has the added benefit of exposing
some additional safe stuff to the optimizers.
> Of even more interest is what happens during tree optimization.
> Do we attempt to propagate the original hard register back into
> the asm argument?
I suspect we would.
I *think* we only avoid propagation into asms when doing so would
replace a hard register in the asm.
In this case we'd be replacing a temporary in the asm with a hard