math/mpx-mul4-x86-sse2.S: Use `movdqa' to move between XMM registers.

Not `movdqu'.  I don't think there's a performance difference (any
more), but it's better style.

diff --git a/math/mpx-mul4-x86-sse2.S b/math/mpx-mul4-x86-sse2.S
index ab238862f5a5c09d77357296802c4fed5478eeae..0e87ff58263e086c39c919d94afbcd2dc2d861e1 100644 (file)
--- a/math/mpx-mul4-x86-sse2.S
+++ b/math/mpx-mul4-x86-sse2.S
@@ -276,7 +276,7 @@
        psrldq  \c1, 8                  // (0, y^*_1, 0, 0)
        psrldq  \c2, 4                  // (0, 0, y^*_2, 0)
   .ifnes "\h", "nil"
        psrldq  \c1, 8                  // (0, y^*_1, 0, 0)
        psrldq  \c2, 4                  // (0, 0, y^*_2, 0)
   .ifnes "\h", "nil"

-       movdqu  \t, \c3
+       movdqa  \t, \c3

        pxor    \u, \u
   .endif
        pslldq  \c3, 12                 // (0, 0, 0, y^*_3)
        pxor    \u, \u
   .endif
        pslldq  \c3, 12                 // (0, 0, 0, y^*_3)