[PULL 01/16] host/include/x86_64: Use __m128i for "x" constraints

[Richard Henderson <richard.henderson@linaro.org>]

The macOS catalina compiler produces an error for __int128_t
as the type for allocation with SSE inline asm constraint.
Create a new X86Int128Union type and use the vector type for
all SSE register inputs and outputs.

Tested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 host/include/x86_64/host/atomic128-ldst.h     | 25 ++++++++++++-------
 .../x86_64/host/load-extract-al16-al8.h       |  8 +++---
 2 files changed, 20 insertions(+), 13 deletions(-)

diff --git a/host/include/x86_64/host/atomic128-ldst.h b/host/include/x86_64/host/atomic128-ldst.h
index adc9332f91..8d6f909d3c 100644
--- a/host/include/x86_64/host/atomic128-ldst.h
+++ b/host/include/x86_64/host/atomic128-ldst.h
@@ -8,12 +8,19 @@
  * atomic primitive is meant to provide.
  */
 
-#ifndef AARCH64_ATOMIC128_LDST_H
-#define AARCH64_ATOMIC128_LDST_H
+#ifndef X86_64_ATOMIC128_LDST_H
+#define X86_64_ATOMIC128_LDST_H
 
 #ifdef CONFIG_INT128_TYPE
 #include "host/cpuinfo.h"
 #include "tcg/debug-assert.h"
+#include <immintrin.h>
+
+typedef union {
+    __m128i v;
+    __int128_t i;
+    Int128 s;
+} X86Int128Union;
 
 /*
  * Through clang 16, with -mcx16, __atomic_load_n is incorrectly
@@ -25,10 +32,10 @@
 
 static inline Int128 atomic16_read_ro(const Int128 *ptr)
 {
-    Int128Alias r;
+    X86Int128Union r;
 
     tcg_debug_assert(HAVE_ATOMIC128_RO);
-    asm("vmovdqa %1, %0" : "=x" (r.i) : "m" (*ptr));
+    asm("vmovdqa %1, %0" : "=x" (r.v) : "m" (*ptr));
 
     return r.s;
 }
@@ -36,10 +43,10 @@ static inline Int128 atomic16_read_ro(const Int128 *ptr)
 static inline Int128 atomic16_read_rw(Int128 *ptr)
 {
     __int128_t *ptr_align = __builtin_assume_aligned(ptr, 16);
-    Int128Alias r;
+    X86Int128Union r;
 
     if (HAVE_ATOMIC128_RO) {
-        asm("vmovdqa %1, %0" : "=x" (r.i) : "m" (*ptr_align));
+        asm("vmovdqa %1, %0" : "=x" (r.v) : "m" (*ptr_align));
     } else {
         r.i = __sync_val_compare_and_swap_16(ptr_align, 0, 0);
     }
@@ -49,10 +56,10 @@ static inline Int128 atomic16_read_rw(Int128 *ptr)
 static inline void atomic16_set(Int128 *ptr, Int128 val)
 {
     __int128_t *ptr_align = __builtin_assume_aligned(ptr, 16);
-    Int128Alias new = { .s = val };
+    X86Int128Union new = { .s = val };
 
     if (HAVE_ATOMIC128_RO) {
-        asm("vmovdqa %1, %0" : "=m"(*ptr_align) : "x" (new.i));
+        asm("vmovdqa %1, %0" : "=m"(*ptr_align) : "x" (new.v));
     } else {
         __int128_t old;
         do {
@@ -65,4 +72,4 @@ static inline void atomic16_set(Int128 *ptr, Int128 val)
 #include "host/include/generic/host/atomic128-ldst.h"
 #endif
 
-#endif /* AARCH64_ATOMIC128_LDST_H */
+#endif /* X86_64_ATOMIC128_LDST_H */
diff --git a/host/include/x86_64/host/load-extract-al16-al8.h b/host/include/x86_64/host/load-extract-al16-al8.h
index 31b6fe8c45..baa506b7b5 100644
--- a/host/include/x86_64/host/load-extract-al16-al8.h
+++ b/host/include/x86_64/host/load-extract-al16-al8.h
@@ -9,7 +9,7 @@
 #define X86_64_LOAD_EXTRACT_AL16_AL8_H
 
 #ifdef CONFIG_INT128_TYPE
-#include "host/cpuinfo.h"
+#include "host/atomic128-ldst.h"
 
 /**
  * load_atom_extract_al16_or_al8:
@@ -26,7 +26,7 @@ load_atom_extract_al16_or_al8(void *pv, int s)
     uintptr_t pi = (uintptr_t)pv;
     __int128_t *ptr_align = (__int128_t *)(pi & ~7);
     int shr = (pi & 7) * 8;
-    Int128Alias r;
+    X86Int128Union r;
 
     /*
      * ptr_align % 16 is now only 0 or 8.
@@ -35,9 +35,9 @@ load_atom_extract_al16_or_al8(void *pv, int s)
      * when ptr_align % 16 == 0 for 16-byte atomicity.
      */
     if ((cpuinfo & CPUINFO_ATOMIC_VMOVDQU) || (pi & 8)) {
-        asm("vmovdqu %1, %0" : "=x" (r.i) : "m" (*ptr_align));
+        asm("vmovdqu %1, %0" : "=x" (r.v) : "m" (*ptr_align));
     } else {
-        asm("vmovdqa %1, %0" : "=x" (r.i) : "m" (*ptr_align));
+        asm("vmovdqa %1, %0" : "=x" (r.v) : "m" (*ptr_align));
     }
     return int128_getlo(int128_urshift(r.s, shr));
 }
-- 
2.34.1