This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: altivec: memory to gpr copies


2002-04-22  Aldy Hernandez  <aldyh@redhat.com>

	* config/rs6000/rs6000.md ("*movv4si_internal"): Change 'm'
	constraint to 'o' for m=r and r=m alternatives.
	("*movv8hi_internal1"): Same.
	("*movv16qi_internal1"): Same.
	("*movv4sf_internal1"): Same.

This patch is okay with me unless someone thinks it is a completely stupid
idea.

	The other approach, if the pattern must accept all MEMs, is to
notice non-offsettable RTL and explicitly increment the base register
during output and restoring the original value of the base register at the
end.

Thanks, David


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]