This is the mail archive of the gcc@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]

Trivial case of inefficial PA code


This is probably trivial to fix for somebody who knows the PA, but hard
for somebody who doesn't.

Consider:

void sub1 (long long *p) { *p = 0; }

This generates:

	copy %r0,%r20
	copy %r0,%r21
	stw %r20,0(%r26)
	stw %r21,4(%r26)
	bv,n %r0(%r2)

As I understand the architecture, this could be done with just

	stw 0,0(%r26)
	stw 0,4(%r26)
	bv,n %r0(%r2)

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