util/u_math: Use xmmintrin.h whenever possible.

It seems  __builtin_ia32_ldmxcsr is only available on gcc and only when
-msse is used. xmmintrin.h/pmmintrin.h provide portable intrinsics, but
these too are only available with gcc when -msse/-msse3 are set.

scons build always sets -msse on x86 builds, but autotools doesn't seem
to.

We could try to get this working on gcc x86 without -msse by emitting
assembly, but I believe that in this day and age we really should be
building Mesa with -msse and -msse2.

diff --git a/src/gallium/auxiliary/util/u_math.c b/src/gallium/auxiliary/util/u_math.c
index 2487bc7..f3fe392 100644 (file)
--- a/src/gallium/auxiliary/util/u_math.c
+++ b/src/gallium/auxiliary/util/u_math.c
@@ -27,9 +27,17 @@
 
 
 
+#include "pipe/p_config.h"
 #include "util/u_math.h"
 #include "util/u_cpu_detect.h"
 
+#if defined(PIPE_ARCH_SSE)
+#include <xmmintrin.h>
+/* This is defined in pmmintrin.h, but it can only be included when -msse3 is
+ * used, so just define it here to avoid further. */
+#define _MM_DENORMALS_ZERO_MASK        0x0040
+#endif
+
 
 /** 2^x, for x in [-1.0, 1.0) */
 float pow2_table[POW2_TABLE_SIZE];
@@ -81,9 +89,9 @@ util_fpstate_get(void)
 {
    unsigned mxcsr = 0;
 
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+#if defined(PIPE_ARCH_SSE)
    if (util_cpu_caps.has_sse) {
-      mxcsr = __builtin_ia32_stmxcsr();
+      mxcsr = _mm_getcsr();
    }
 #endif
 
@@ -99,13 +107,13 @@ util_fpstate_get(void)
 unsigned
 util_fpstate_set_denorms_to_zero(unsigned current_mxcsr)
 {
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
-#define MXCSR_DAZ (1 << 6)     /* Enable denormals are zero mode */
-#define MXCSR_FTZ (1 << 15)    /* Enable flush to zero mode */
+#if defined(PIPE_ARCH_SSE)
    if (util_cpu_caps.has_sse) {
-      current_mxcsr |= MXCSR_FTZ;
+      /* Enable flush to zero mode */
+      current_mxcsr |= _MM_FLUSH_ZERO_MASK;
       if (util_cpu_caps.has_sse3) {
-         current_mxcsr |= MXCSR_DAZ;
+         /* Enable denormals are zero mode */
+         current_mxcsr |= _MM_DENORMALS_ZERO_MASK;
       }
       util_fpstate_set(current_mxcsr);
    }
@@ -121,9 +129,9 @@ util_fpstate_set_denorms_to_zero(unsigned current_mxcsr)
 void
 util_fpstate_set(unsigned mxcsr)
 {
-#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+#if defined(PIPE_ARCH_SSE)
    if (util_cpu_caps.has_sse) {
-      __builtin_ia32_ldmxcsr(mxcsr);
+      _mm_setcsr(mxcsr);
    }
 #endif
 }