Width use memory instead of register for 32 bit fpic.

source/row_gcc.cc

 // VERSION 2
 /*
  *  Copyright 2011 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.
  */
@@ skipping 1552 matching lines @@
     "palignr   $0xc,%%xmm0,%%xmm1              \n"
     "movq      %%xmm0," MEMACCESS([dst_rgb24]) "\n"
     "movdqu    %%xmm1," MEMACCESS2(0x8,[dst_rgb24]) "\n"
     "lea       " MEMLEA(0x18,[dst_rgb24]) ",%[dst_rgb24] \n"
     "subl      $0x8,%[width]                   \n"
     "jg        1b                              \n"
   : [y_buf]"+r"(y_buf),    // %[y_buf]
     [u_buf]"+r"(u_buf),    // %[u_buf]
     [v_buf]"+r"(v_buf),    // %[v_buf]
     [dst_rgb24]"+r"(dst_rgb24),  // %[dst_rgb24]
-// TODO(fbarchard): Make width a register for 32 bit.
 #if defined(__i386__) && defined(__pic__)
     [width]"+m"(width)     // %[width]
 #else
     [width]"+rm"(width)    // %[width]
 #endif
   : [yuvconstants]"r"(yuvconstants),  // %[yuvconstants]
     [kShuffleMaskARGBToRGB24_0]"m"(kShuffleMaskARGBToRGB24_0),
     [kShuffleMaskARGBToRGB24]"m"(kShuffleMaskARGBToRGB24)
   : "memory", "cc", NACL_R14
     "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
@@ skipping 24 matching lines @@
     "palignr   $0xc,%%xmm0,%%xmm1              \n"
     "movq      %%xmm0," MEMACCESS([dst_raw]) " \n"
     "movdqu    %%xmm1," MEMACCESS2(0x8,[dst_raw]) "\n"
     "lea       " MEMLEA(0x18,[dst_raw]) ",%[dst_raw] \n"
     "subl      $0x8,%[width]                   \n"
     "jg        1b                              \n"
   : [y_buf]"+r"(y_buf),    // %[y_buf]
     [u_buf]"+r"(u_buf),    // %[u_buf]
     [v_buf]"+r"(v_buf),    // %[v_buf]
     [dst_raw]"+r"(dst_raw),  // %[dst_raw]
-// TODO(fbarchard): Make width a register for 32 bit.
 #if defined(__i386__) && defined(__pic__)
     [width]"+m"(width)    // %[width]
 #else
     [width]"+rm"(width)    // %[width]
 #endif
   : [yuvconstants]"r"(yuvconstants),  // %[yuvconstants]
     [kShuffleMaskARGBToRAW_0]"m"(kShuffleMaskARGBToRAW_0),
     [kShuffleMaskARGBToRAW]"m"(kShuffleMaskARGBToRAW)
   : "memory", "cc", NACL_R14
     "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
@@ skipping 34 matching lines @@
                                      uint8* dst_argb,
                                      struct YuvConstants* yuvconstants,
                                      int width) {
   asm volatile (
     "sub       %[u_buf],%[v_buf]               \n"
     LABELALIGN
   "1:                                          \n"
     READYUVA422
     YUVTORGB(yuvconstants)
     STOREARGB
-    "sub       $0x8,%[width]                   \n"
+    "subl      $0x8,%[width]                   \n"
     "jg        1b                              \n"
   : [y_buf]"+r"(y_buf),    // %[y_buf]
     [u_buf]"+r"(u_buf),    // %[u_buf]
     [v_buf]"+r"(v_buf),    // %[v_buf]
     [a_buf]"+r"(a_buf),    // %[a_buf]
     [dst_argb]"+r"(dst_argb),  // %[dst_argb]
+#if defined(__i386__) && defined(__pic__)
+    [width]"+m"(width)     // %[width]
+#else
     [width]"+rm"(width)    // %[width]
+#endif        
   : [yuvconstants]"r"(yuvconstants)  // %[yuvconstants]
   : "memory", "cc", NACL_R14
     "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
   );
 }
 
 void OMITFP I422AlphaToABGRRow_SSSE3(const uint8* y_buf,
                                      const uint8* u_buf,
                                      const uint8* v_buf,
                                      const uint8* a_buf,
                                      uint8* dst_abgr,
                                      struct YuvConstants* yuvconstants,
                                      int width) {
   asm volatile (
     "sub       %[u_buf],%[v_buf]               \n"
     LABELALIGN
   "1:                                          \n"
     READYUVA422
     YUVTORGB(yuvconstants)
     STOREABGR
-    "sub       $0x8,%[width]                   \n"
+    "subl      $0x8,%[width]                   \n"
     "jg        1b                              \n"
   : [y_buf]"+r"(y_buf),    // %[y_buf]
     [u_buf]"+r"(u_buf),    // %[u_buf]
     [v_buf]"+r"(v_buf),    // %[v_buf]
     [a_buf]"+r"(a_buf),    // %[a_buf]
     [dst_abgr]"+r"(dst_abgr),  // %[dst_abgr]
+#if defined(__i386__) && defined(__pic__)
+    [width]"+m"(width)     // %[width]
+#else
     [width]"+rm"(width)    // %[width]
+#endif    
   : [yuvconstants]"r"(yuvconstants)  // %[yuvconstants]
   : "memory", "cc", NACL_R14
     "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
   );
 }
 
 void OMITFP I411ToARGBRow_SSSE3(const uint8* y_buf,
                                 const uint8* u_buf,
                                 const uint8* v_buf,
                                 uint8* dst_argb,
@@ skipping 362 matching lines @@
                                uint8* dst_argb,
                                struct YuvConstants* yuvconstants,
                                int width) {
   asm volatile (
     "sub       %[u_buf],%[v_buf]               \n"
     LABELALIGN
   "1:                                          \n"
     READYUVA422_AVX2
     YUVTORGB_AVX2(yuvconstants)
     STOREARGB_AVX2
-    "sub       $0x10,%[width]                  \n"
+    "subl      $0x10,%[width]                  \n"
     "jg        1b                              \n"
     "vzeroupper                                \n"
   : [y_buf]"+r"(y_buf),    // %[y_buf]
     [u_buf]"+r"(u_buf),    // %[u_buf]
     [v_buf]"+r"(v_buf),    // %[v_buf]
     [a_buf]"+r"(a_buf),    // %[a_buf]
     [dst_argb]"+r"(dst_argb),  // %[dst_argb]
+#if defined(__i386__) && defined(__pic__)
+    [width]"+m"(width)     // %[width]
+#else
     [width]"+rm"(width)    // %[width]
+#endif    
   : [yuvconstants]"r"(yuvconstants)  // %[yuvconstants]
   : "memory", "cc", NACL_R14
     "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
   );
 }
 #endif  // HAS_I422ALPHATOARGBROW_AVX2
 
 #if defined(HAS_I422ALPHATOABGRROW_AVX2)
 // 16 pixels
 // 8 UV values upsampled to 16 UV, mixed with 16 Y and 16 A producing 16 ABGR.
 void OMITFP I422AlphaToABGRRow_AVX2(const uint8* y_buf,
                                const uint8* u_buf,
                                const uint8* v_buf,
                                const uint8* a_buf,
                                uint8* dst_abgr,
                                struct YuvConstants* yuvconstants,
                                int width) {
   asm volatile (
     "sub       %[u_buf],%[v_buf]               \n"
     LABELALIGN
   "1:                                          \n"
     READYUVA422_AVX2
     YUVTORGB_AVX2(yuvconstants)
     STOREABGR_AVX2
-    "sub       $0x10,%[width]                  \n"
+    "subl      $0x10,%[width]                  \n"
     "jg        1b                              \n"
     "vzeroupper                                \n"
   : [y_buf]"+r"(y_buf),    // %[y_buf]
     [u_buf]"+r"(u_buf),    // %[u_buf]
     [v_buf]"+r"(v_buf),    // %[v_buf]
     [a_buf]"+r"(a_buf),    // %[a_buf]
     [dst_abgr]"+r"(dst_abgr),  // %[dst_abgr]
+#if defined(__i386__) && defined(__pic__)
+    [width]"+m"(width)     // %[width]
+#else
     [width]"+rm"(width)    // %[width]
+#endif    
   : [yuvconstants]"r"(yuvconstants)  // %[yuvconstants]
   : "memory", "cc", NACL_R14
     "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
   );
 }
 #endif  // HAS_I422ALPHATOABGRROW_AVX2
 
 #if defined(HAS_I422TOABGRROW_AVX2)
 // 16 pixels
 // 8 UV values upsampled to 16 UV, mixed with 16 Y producing 16 ABGR (64 bytes).
@@ skipping 3426 matching lines @@
   );
 }
 #endif  // HAS_ARGBLUMACOLORTABLEROW_SSSE3
 
 #endif  // defined(__x86_64__) || defined(__i386__)
 
 #ifdef __cplusplus
 }  // extern "C"
 }  // namespace libyuv
 #endif