libvpx: Pull from upstream

source/libvpx/vp9/encoder/x86/vp9_sad4d_intrin_avx2.c

 /*
  *  Copyright (c) 2014 The WebM 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 <immintrin.h>  // AVX2
@@ skipping 13 matching lines @@
   ref0 = ref[0];
   ref1 = ref[1];
   ref2 = ref[2];
   ref3 = ref[3];
   sum_ref0 = _mm256_set1_epi16(0);
   sum_ref1 = _mm256_set1_epi16(0);
   sum_ref2 = _mm256_set1_epi16(0);
   sum_ref3 = _mm256_set1_epi16(0);
   for (i = 0; i < 32 ; i++) {
     // load src and all refs
-    src_reg = _mm256_load_si256((__m256i *)(src));
+    src_reg = _mm256_loadu_si256((__m256i *)(src));
     ref0_reg = _mm256_loadu_si256((__m256i *) (ref0));
     ref1_reg = _mm256_loadu_si256((__m256i *) (ref1));
     ref2_reg = _mm256_loadu_si256((__m256i *) (ref2));
     ref3_reg = _mm256_loadu_si256((__m256i *) (ref3));
     // sum of the absolute differences between every ref-i to src
     ref0_reg = _mm256_sad_epu8(ref0_reg, src_reg);
     ref1_reg = _mm256_sad_epu8(ref1_reg, src_reg);
     ref2_reg = _mm256_sad_epu8(ref2_reg, src_reg);
     ref3_reg = _mm256_sad_epu8(ref3_reg, src_reg);
     // sum every ref-i
@@ skipping 51 matching lines @@
   ref0 = ref[0];
   ref1 = ref[1];
   ref2 = ref[2];
   ref3 = ref[3];
   sum_ref0 = _mm256_set1_epi16(0);
   sum_ref1 = _mm256_set1_epi16(0);
   sum_ref2 = _mm256_set1_epi16(0);
   sum_ref3 = _mm256_set1_epi16(0);
   for (i = 0; i < 64 ; i++) {
     // load 64 bytes from src and all refs
-    src_reg = _mm256_load_si256((__m256i *)(src));
-    srcnext_reg = _mm256_load_si256((__m256i *)(src + 32));
+    src_reg = _mm256_loadu_si256((__m256i *)(src));
+    srcnext_reg = _mm256_loadu_si256((__m256i *)(src + 32));
     ref0_reg = _mm256_loadu_si256((__m256i *) (ref0));
     ref0next_reg = _mm256_loadu_si256((__m256i *) (ref0 + 32));
     ref1_reg = _mm256_loadu_si256((__m256i *) (ref1));
     ref1next_reg = _mm256_loadu_si256((__m256i *) (ref1 + 32));
     ref2_reg = _mm256_loadu_si256((__m256i *) (ref2));
     ref2next_reg = _mm256_loadu_si256((__m256i *) (ref2 + 32));
     ref3_reg = _mm256_loadu_si256((__m256i *) (ref3));
     ref3next_reg = _mm256_loadu_si256((__m256i *) (ref3 + 32));
     // sum of the absolute differences between every ref-i to src
     ref0_reg = _mm256_sad_epu8(ref0_reg, src_reg);
@@ skipping 40 matching lines @@
     // add the low 64 bit to the high 64 bit
     sum_mlow = _mm256_add_epi32(sum_mlow, sum_mhigh);
 
     // add the low 128 bit to the high 128 bit
     sum = _mm_add_epi32(_mm256_castsi256_si128(sum_mlow),
                         _mm256_extractf128_si256(sum_mlow, 1));
 
     _mm_storeu_si128((__m128i *)(res), sum);
   }
 }