libvpx: Pull from upstream

source/libvpx/third_party/libyuv/source/compare_win.cc

 /*
  *  Copyright 2012 The LibYuv Project Authors. All rights reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS. All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
 #include "libyuv/basic_types.h"
 #include "libyuv/row.h"
 
 #ifdef __cplusplus
 namespace libyuv {
 extern "C" {
 #endif
 
-#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
+// This module is for Visual C x86.
+#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && \
+    defined(_MSC_VER) && !defined(__clang__)
 
-__declspec(naked) __declspec(align(16))
+__declspec(naked)
 uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) {
   __asm {
     mov        eax, [esp + 4]    // src_a
     mov        edx, [esp + 8]    // src_b
     mov        ecx, [esp + 12]   // count
     pxor       xmm0, xmm0
     pxor       xmm5, xmm5
 
   wloop:
     movdqu     xmm1, [eax]
@@ skipping 20 matching lines @@
     paddd      xmm0, xmm1
     movd       eax, xmm0
     ret
   }
 }
 
 // Visual C 2012 required for AVX2.
 #if _MSC_VER >= 1700
 // C4752: found Intel(R) Advanced Vector Extensions; consider using /arch:AVX.
 #pragma warning(disable: 4752)
-__declspec(naked) __declspec(align(16))
+__declspec(naked)
 uint32 SumSquareError_AVX2(const uint8* src_a, const uint8* src_b, int count) {
   __asm {
     mov        eax, [esp + 4]    // src_a
     mov        edx, [esp + 8]    // src_b
     mov        ecx, [esp + 12]   // count
     vpxor      ymm0, ymm0, ymm0  // sum
     vpxor      ymm5, ymm5, ymm5  // constant 0 for unpck
     sub        edx, eax
 
   wloop:
@@ skipping 53 matching lines @@
 };
 
 // 27: 66 0F 38 40 C6     pmulld      xmm0,xmm6
 // 44: 66 0F 38 40 DD     pmulld      xmm3,xmm5
 // 59: 66 0F 38 40 E5     pmulld      xmm4,xmm5
 // 72: 66 0F 38 40 D5     pmulld      xmm2,xmm5
 // 83: 66 0F 38 40 CD     pmulld      xmm1,xmm5
 #define pmulld(reg) _asm _emit 0x66 _asm _emit 0x0F _asm _emit 0x38 \
     _asm _emit 0x40 _asm _emit reg
 
-__declspec(naked) __declspec(align(16))
+__declspec(naked)
 uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) {
   __asm {
     mov        eax, [esp + 4]    // src
     mov        ecx, [esp + 8]    // count
     movd       xmm0, [esp + 12]  // seed
 
     pxor       xmm7, xmm7        // constant 0 for unpck
     movdqa     xmm6, kHash16x33
 
   wloop:
@@ skipping 30 matching lines @@
     sub        ecx, 16
     jg         wloop
 
     movd       eax, xmm0         // return hash
     ret
   }
 }
 
 // Visual C 2012 required for AVX2.
 #if _MSC_VER >= 1700
-__declspec(naked) __declspec(align(16))
+__declspec(naked)
 uint32 HashDjb2_AVX2(const uint8* src, int count, uint32 seed) {
   __asm {
     mov        eax, [esp + 4]    // src
     mov        ecx, [esp + 8]    // count
     movd       xmm0, [esp + 12]  // seed
     movdqa     xmm6, kHash16x33
 
   wloop:
     vpmovzxbd  xmm3, dword ptr [eax]  // src[0-3]
     pmulld     xmm0, xmm6  // hash *= 33 ^ 16
@@ skipping 14 matching lines @@
     paddd      xmm1, xmm2
     paddd      xmm0, xmm1
     sub        ecx, 16
     jg         wloop
 
     movd       eax, xmm0         // return hash
     ret
   }
 }
 #endif  // _MSC_VER >= 1700
-
-#endif  // !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && defined(_MSC_VER)
+#endif  // !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86)
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif