libvpx: Pull from upstream

source/libvpx/vp9/common/arm/neon/vp9_convolve_neon.c

 /*
  *  Copyright (c) 2013 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 "./vp9_rtcd.h"
 #include "vp9/common/vp9_common.h"
+#include "vpx_ports/mem.h"
 
 void vp9_convolve8_neon(const uint8_t *src, ptrdiff_t src_stride,
                         uint8_t *dst, ptrdiff_t dst_stride,
                         const int16_t *filter_x, int x_step_q4,
                         const int16_t *filter_y, int y_step_q4,
                         int w, int h) {
   /* Given our constraints: w <= 64, h <= 64, taps == 8 we can reduce the
    * maximum buffer size to 64 * 64 + 7 (+ 1 to make it divisible by 4).
    */
-  uint8_t temp[64 * 72];
+  DECLARE_ALIGNED_ARRAY(8, uint8_t, temp, 64 * 72);
 
   // Account for the vertical phase needing 3 lines prior and 4 lines post
   int intermediate_height = h + 7;
 
   if (x_step_q4 != 16 || y_step_q4 != 16)
     return vp9_convolve8_c(src, src_stride,
                            dst, dst_stride,
                            filter_x, x_step_q4,
                            filter_y, y_step_q4,
                            w, h);
@@ skipping 13 matching lines @@
                           dst, dst_stride,
                           filter_x, x_step_q4, filter_y, y_step_q4,
                           w, h);
 }
 
 void vp9_convolve8_avg_neon(const uint8_t *src, ptrdiff_t src_stride,
                             uint8_t *dst, ptrdiff_t dst_stride,
                             const int16_t *filter_x, int x_step_q4,
                             const int16_t *filter_y, int y_step_q4,
                             int w, int h) {
-  uint8_t temp[64 * 72];
+  DECLARE_ALIGNED_ARRAY(8, uint8_t, temp, 64 * 72);
   int intermediate_height = h + 7;
 
   if (x_step_q4 != 16 || y_step_q4 != 16)
     return vp9_convolve8_avg_c(src, src_stride,
                                dst, dst_stride,
                                filter_x, x_step_q4,
                                filter_y, y_step_q4,
                                w, h);
 
   /* This implementation has the same issues as above. In addition, we only want
    * to average the values after both passes.
    */
   vp9_convolve8_horiz_neon(src - src_stride * 3, src_stride,
                            temp, 64,
                            filter_x, x_step_q4, filter_y, y_step_q4,
                            w, intermediate_height);
   vp9_convolve8_avg_vert_neon(temp + 64 * 3,
                               64, dst, dst_stride,
                               filter_x, x_step_q4, filter_y, y_step_q4,
                               w, h);
 }