ec9d2633f1
clobber with the 'y' constraint. Otherwise, we get the wrong return type and an assert, because it created a '<1 x i64>' vector type instead of the x86_mmx type. llvm-svn: 127185
23 lines
632 B
C
23 lines
632 B
C
// RUN: %clang -mmmx -ccc-host-triple i386-unknown-unknown -emit-llvm -S %s -o - | FileCheck %s
|
|
// <rdar://problem/9091220>
|
|
#include <mmintrin.h>
|
|
|
|
// CHECK: type { x86_mmx, x86_mmx, x86_mmx, x86_mmx, x86_mmx, x86_mmx, x86_mmx }
|
|
|
|
void foo(long long fill) {
|
|
__m64 vfill = _mm_cvtsi64_m64(fill);
|
|
__m64 v1, v2, v3, v4, v5, v6, v7;
|
|
|
|
__asm__ __volatile__ (
|
|
"\tmovq %7, %0\n"
|
|
"\tmovq %7, %1\n"
|
|
"\tmovq %7, %2\n"
|
|
"\tmovq %7, %3\n"
|
|
"\tmovq %7, %4\n"
|
|
"\tmovq %7, %5\n"
|
|
"\tmovq %7, %6"
|
|
: "=&y" (v1), "=&y" (v2), "=&y" (v3),
|
|
"=&y" (v4), "=&y" (v5), "=&y" (v6), "=y" (v7)
|
|
: "y" (vfill));
|
|
}
|