disable faulty avx2 in argb conversions and box filter. and extend temporary…

disable faulty avx2 in argb conversions and box filter.  and extend temporary buffer to 128 for an avx2 any function.
R=harryjin@google.com
BUG=libyuv:462
TESTED=libyuv_unittest run on haswell laptop

Review URL: https://webrtc-codereview.appspot.com/53759004.

README.chromium

 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 1443
+Version: 1444
 License: BSD
 License File: LICENSE
 

include/libyuv/row.h

@@ -186,8 +186,9 @@ extern "C" {
 
 // The following are available require VS2012.  Port to GCC.
 #if !defined(LIBYUV_DISABLE_X86) && defined(VISUALC_HAS_AVX2)
-#define HAS_ARGB1555TOARGBROW_AVX2
-#define HAS_ARGB4444TOARGBROW_AVX2
+// Some AVX2 versions disabled. See libyuv bug 462.
+// #define HAS_ARGB1555TOARGBROW_AVX2
+// #define HAS_ARGB4444TOARGBROW_AVX2
 #define HAS_ARGBTOARGB1555ROW_AVX2
 #define HAS_ARGBTOARGB4444ROW_AVX2
 #define HAS_ARGBTORGB565DITHERROW_AVX2
@@ -203,7 +204,7 @@ extern "C" {
 #define HAS_NV12TORGB565ROW_AVX2
 #define HAS_NV21TOARGBROW_AVX2
 #define HAS_NV21TORGB565ROW_AVX2
-#define HAS_RGB565TOARGBROW_AVX2
+// #define HAS_RGB565TOARGBROW_AVX2
 #endif
 
 // The following are available on all x86 platforms, but

include/libyuv/scale_row.h

@@ -50,7 +50,8 @@ extern "C" {
 
 // The following are available on VS2012:
 #if !defined(LIBYUV_DISABLE_X86) && defined(VISUALC_HAS_AVX2)
-#define HAS_SCALEADDROW_AVX2
+// Some AVX2 versions disabled. See libyuv bug 462.
+// #define HAS_SCALEADDROW_AVX2
 #define HAS_SCALEROWDOWN2_AVX2
 #define HAS_SCALEROWDOWN4_AVX2
 #endif

include/libyuv/version.h

@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 1443
+#define LIBYUV_VERSION 1444
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT

source/convert.cc

@@ -1232,7 +1232,6 @@ int ARGB1555ToI420(const uint8* src_argb1555, int src_stride_argb1555,
 #else
       ARGB1555ToARGBRow(src_argb1555, row, width);
       ARGB1555ToARGBRow(src_argb1555 + src_stride_argb1555, row + kRowSize,
-
                         width);
       ARGBToUVRow(row, kRowSize, dst_u, dst_v, width);
       ARGBToYRow(row, dst_y, width);

source/row_any.cc

@@ -10,7 +10,7 @@
 
 #include "libyuv/row.h"
 
-#include <string.h>  // For memcpy and memset.
+#include <string.h>  // For memset.
 
 #include "libyuv/basic_types.h"
 

source/row_win.cc

@@ -626,10 +626,10 @@ __declspec(naked)
 void ARGB4444ToARGBRow_AVX2(const uint8* src_argb4444, uint8* dst_argb,
                             int pix) {
   __asm {
-    mov        eax,  0x0f0f0f0f  // generate mask 0x0f0f0f0f
-    vmovd      xmm4, eax
+    mov       eax,  0x0f0f0f0f  // generate mask 0x0f0f0f0f
+    vmovd     xmm4, eax
     vbroadcastss ymm4, xmm4
-    vpslld     ymm5, ymm4, 4    // 0xf0f0f0f0 for high nibbles
+    vpslld    ymm5, ymm4, 4     // 0xf0f0f0f0 for high nibbles
     mov       eax,  [esp + 4]   // src_argb4444
     mov       edx,  [esp + 8]   // dst_argb
     mov       ecx,  [esp + 12]  // pix