libvpx: Pull from upstream

source/libvpx/vp9/common/vp9_rtcd_defs.pl

 sub vp9_common_forward_decls() {
 print <<EOF
 /*
  * VP9
  */
 
 #include "vpx/vpx_integer.h"
 #include "vp9/common/vp9_common.h"
 #include "vp9/common/vp9_enums.h"
 
@@ skipping 42 matching lines @@
   $avx2_ssse3 = '';
 }
 
 #
 # RECON
 #
 add_proto qw/void vp9_d207_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d207_predictor_4x4/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d45_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_d45_predictor_4x4/, "$ssse3_x86inc";
+specialize qw/vp9_d45_predictor_4x4 neon/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d63_predictor_4x4/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_h_predictor_4x4 neon dspr2/, "$ssse3_x86inc";
+specialize qw/vp9_h_predictor_4x4 neon dspr2 msa/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d117_predictor_4x4/;
 
 add_proto qw/void vp9_d135_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_d135_predictor_4x4/;
+specialize qw/vp9_d135_predictor_4x4 neon/;
 
 add_proto qw/void vp9_d153_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d153_predictor_4x4/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_v_predictor_4x4 neon/, "$sse_x86inc";
+specialize qw/vp9_v_predictor_4x4 neon msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_tm_predictor_4x4 neon dspr2/, "$sse_x86inc";
+specialize qw/vp9_tm_predictor_4x4 neon dspr2 msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_predictor_4x4 dspr2/, "$sse_x86inc";
+specialize qw/vp9_dc_predictor_4x4 dspr2 msa neon/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_top_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_top_predictor_4x4/, "$sse_x86inc";
+specialize qw/vp9_dc_top_predictor_4x4 msa neon/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_left_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_left_predictor_4x4/, "$sse_x86inc";
+specialize qw/vp9_dc_left_predictor_4x4 msa neon/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_128_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_128_predictor_4x4/, "$sse_x86inc";
+specialize qw/vp9_dc_128_predictor_4x4 msa neon/, "$sse_x86inc";
 
 add_proto qw/void vp9_d207_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d207_predictor_8x8/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d45_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_d45_predictor_8x8/, "$ssse3_x86inc";
+specialize qw/vp9_d45_predictor_8x8 neon/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d63_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d63_predictor_8x8/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_h_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_h_predictor_8x8 neon dspr2/, "$ssse3_x86inc";
+specialize qw/vp9_h_predictor_8x8 neon dspr2 msa/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d117_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d117_predictor_8x8/;
 
 add_proto qw/void vp9_d135_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d135_predictor_8x8/;
 
 add_proto qw/void vp9_d153_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d153_predictor_8x8/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_v_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_v_predictor_8x8 neon/, "$sse_x86inc";
+specialize qw/vp9_v_predictor_8x8 neon msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_tm_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_tm_predictor_8x8 neon dspr2/, "$sse2_x86inc";
+specialize qw/vp9_tm_predictor_8x8 neon dspr2 msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_predictor_8x8 dspr2 neon/, "$sse_x86inc";
+specialize qw/vp9_dc_predictor_8x8 dspr2 neon msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_top_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_top_predictor_8x8 neon/, "$sse_x86inc";
+specialize qw/vp9_dc_top_predictor_8x8 neon msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_left_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_left_predictor_8x8 neon/, "$sse_x86inc";
+specialize qw/vp9_dc_left_predictor_8x8 neon msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_dc_128_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_128_predictor_8x8 neon/, "$sse_x86inc";
+specialize qw/vp9_dc_128_predictor_8x8 neon msa/, "$sse_x86inc";
 
 add_proto qw/void vp9_d207_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d207_predictor_16x16/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d45_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_d45_predictor_16x16/, "$ssse3_x86inc";
+specialize qw/vp9_d45_predictor_16x16 neon/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d63_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d63_predictor_16x16/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_h_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_h_predictor_16x16 neon dspr2/, "$ssse3_x86inc";
+specialize qw/vp9_h_predictor_16x16 neon dspr2 msa/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d117_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d117_predictor_16x16/;
 
 add_proto qw/void vp9_d135_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d135_predictor_16x16/;
 
 add_proto qw/void vp9_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d153_predictor_16x16/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_v_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_v_predictor_16x16 neon/, "$sse2_x86inc";
+specialize qw/vp9_v_predictor_16x16 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_tm_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_tm_predictor_16x16 neon/, "$sse2_x86inc";
+specialize qw/vp9_tm_predictor_16x16 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_predictor_16x16 dspr2 neon/, "$sse2_x86inc";
+specialize qw/vp9_dc_predictor_16x16 dspr2 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_top_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_top_predictor_16x16 neon/, "$sse2_x86inc";
+specialize qw/vp9_dc_top_predictor_16x16 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_left_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_left_predictor_16x16 neon/, "$sse2_x86inc";
+specialize qw/vp9_dc_left_predictor_16x16 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_128_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_128_predictor_16x16 neon/, "$sse2_x86inc";
+specialize qw/vp9_dc_128_predictor_16x16 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_d207_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d207_predictor_32x32/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d45_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d45_predictor_32x32/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d63_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d63_predictor_32x32/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_h_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_h_predictor_32x32 neon/, "$ssse3_x86inc";
+specialize qw/vp9_h_predictor_32x32 neon msa/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_d117_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d117_predictor_32x32/;
 
 add_proto qw/void vp9_d135_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vp9_d135_predictor_32x32/;
 
 add_proto qw/void vp9_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_d153_predictor_32x32/;
+specialize qw/vp9_d153_predictor_32x32/, "$ssse3_x86inc";
 
 add_proto qw/void vp9_v_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_v_predictor_32x32 neon/, "$sse2_x86inc";
+specialize qw/vp9_v_predictor_32x32 neon msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_tm_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_tm_predictor_32x32 neon/, "$sse2_x86_64";
+specialize qw/vp9_tm_predictor_32x32 neon msa/, "$sse2_x86_64";
 
 add_proto qw/void vp9_dc_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_predictor_32x32/, "$sse2_x86inc";
+specialize qw/vp9_dc_predictor_32x32 msa neon/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_top_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_top_predictor_32x32/, "$sse2_x86inc";
+specialize qw/vp9_dc_top_predictor_32x32 msa neon/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_left_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_left_predictor_32x32/, "$sse2_x86inc";
+specialize qw/vp9_dc_left_predictor_32x32 msa neon/, "$sse2_x86inc";
 
 add_proto qw/void vp9_dc_128_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vp9_dc_128_predictor_32x32/, "$sse2_x86inc";
+specialize qw/vp9_dc_128_predictor_32x32 msa neon/, "$sse2_x86inc";
 
 #
 # Loopfilter
 #
 add_proto qw/void vp9_lpf_vertical_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 specialize qw/vp9_lpf_vertical_16 sse2 neon_asm dspr2 msa/;
 $vp9_lpf_vertical_16_neon_asm=vp9_lpf_vertical_16_neon;
 
 add_proto qw/void vp9_lpf_vertical_16_dual/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 specialize qw/vp9_lpf_vertical_16_dual sse2 neon_asm dspr2 msa/;
@@ skipping 45 matching lines @@
 
 add_proto qw/void vp9_post_proc_down_and_across/, "const uint8_t *src_ptr, uint8_t *dst_ptr, int src_pixels_per_line, int dst_pixels_per_line, int rows, int cols, int flimit";
 specialize qw/vp9_post_proc_down_and_across sse2/;
 $vp9_post_proc_down_and_across_sse2=vp9_post_proc_down_and_across_xmm;
 
 add_proto qw/void vp9_plane_add_noise/, "uint8_t *Start, char *noise, char blackclamp[16], char whiteclamp[16], char bothclamp[16], unsigned int Width, unsigned int Height, int Pitch";
 specialize qw/vp9_plane_add_noise sse2/;
 $vp9_plane_add_noise_sse2=vp9_plane_add_noise_wmt;
 
 add_proto qw/void vp9_filter_by_weight16x16/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int src_weight";
-specialize qw/vp9_filter_by_weight16x16 sse2/;
+specialize qw/vp9_filter_by_weight16x16 sse2 msa/;
 
 add_proto qw/void vp9_filter_by_weight8x8/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int src_weight";
-specialize qw/vp9_filter_by_weight8x8 sse2/;
+specialize qw/vp9_filter_by_weight8x8 sse2 msa/;
 }
 
 #
 # Sub Pixel Filters
 #
 add_proto qw/void vp9_convolve_copy/, "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";
 specialize qw/vp9_convolve_copy neon dspr2 msa/, "$sse2_x86inc";
 
 add_proto qw/void vp9_convolve_avg/, "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";
 specialize qw/vp9_convolve_avg neon dspr2 msa/, "$sse2_x86inc";
@@ skipping 578 matching lines @@
 specialize qw/vp9_sub_pixel_avg_variance4x8/, "$sse_x86inc", "$ssse3_x86inc";
 
 add_proto qw/unsigned int vp9_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
 specialize qw/vp9_sub_pixel_variance4x4/, "$sse_x86inc", "$ssse3_x86inc";
 #vp9_sub_pixel_variance4x4_sse2=vp9_sub_pixel_variance4x4_wmt
 
 add_proto qw/unsigned int vp9_sub_pixel_avg_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
 specialize qw/vp9_sub_pixel_avg_variance4x4/, "$sse_x86inc", "$ssse3_x86inc";
 
 add_proto qw/unsigned int vp9_avg_8x8/, "const uint8_t *, int p";
-specialize qw/vp9_avg_8x8 sse2 neon/;
+specialize qw/vp9_avg_8x8 sse2 neon msa/;
 
 add_proto qw/unsigned int vp9_avg_4x4/, "const uint8_t *, int p";
-specialize qw/vp9_avg_4x4 sse2/;
+specialize qw/vp9_avg_4x4 sse2 msa/;
 
 add_proto qw/void vp9_minmax_8x8/, "const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max";
 specialize qw/vp9_minmax_8x8 sse2/;
 
 add_proto qw/void vp9_hadamard_8x8/, "int16_t const *src_diff, int src_stride, int16_t *coeff";
 specialize qw/vp9_hadamard_8x8 sse2/, "$ssse3_x86_64";
 
 add_proto qw/void vp9_hadamard_16x16/, "int16_t const *src_diff, int src_stride, int16_t *coeff";
 specialize qw/vp9_hadamard_16x16 sse2/;
 
@@ skipping 14 matching lines @@
   specialize qw/vp9_highbd_avg_8x8/;
   add_proto qw/unsigned int vp9_highbd_avg_4x4/, "const uint8_t *, int p";
   specialize qw/vp9_highbd_avg_4x4/;
   add_proto qw/void vp9_highbd_minmax_8x8/, "const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max";
   specialize qw/vp9_highbd_minmax_8x8/;
 }
 
 # ENCODEMB INVOKE
 
 add_proto qw/void vp9_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride";
-specialize qw/vp9_subtract_block neon/, "$sse2_x86inc";
+specialize qw/vp9_subtract_block neon msa/, "$sse2_x86inc";
 
 #
 # Denoiser
 #
 if (vpx_config("CONFIG_VP9_TEMPORAL_DENOISING") eq "yes") {
   add_proto qw/int vp9_denoiser_filter/, "const uint8_t *sig, int sig_stride, const uint8_t *mc_avg, int mc_avg_stride, uint8_t *avg, int avg_stride, int increase_denoising, BLOCK_SIZE bs, int motion_magnitude";
   specialize qw/vp9_denoiser_filter sse2/;
 }
 
 if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
@@ skipping 11 matching lines @@
   add_proto qw/void vp9_quantize_b/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
   specialize qw/vp9_quantize_b/;
 
   add_proto qw/void vp9_quantize_b_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
   specialize qw/vp9_quantize_b_32x32/;
 
   add_proto qw/void vp9_fdct8x8_quant/, "const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
   specialize qw/vp9_fdct8x8_quant/;
 } else {
   add_proto qw/int64_t vp9_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz";
-  specialize qw/vp9_block_error avx2/, "$sse2_x86inc";
+  specialize qw/vp9_block_error avx2 msa/, "$sse2_x86inc";
 
   add_proto qw/int64_t vp9_block_error_fp/, "const int16_t *coeff, const int16_t *dqcoeff, int block_size";
   specialize qw/vp9_block_error_fp sse2/;
 
   add_proto qw/void vp9_quantize_fp/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
   specialize qw/vp9_quantize_fp neon sse2/, "$ssse3_x86_64";
 
   add_proto qw/void vp9_quantize_fp_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
   specialize qw/vp9_quantize_fp_32x32/, "$ssse3_x86_64";
 
@@ skipping 54 matching lines @@
   add_proto qw/void vp9_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vp9_fdct32x32_1 sse2/;
 
   add_proto qw/void vp9_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vp9_fdct32x32 sse2/;
 
   add_proto qw/void vp9_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vp9_fdct32x32_rd sse2/;
 } else {
   add_proto qw/void vp9_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  specialize qw/vp9_fht4x4 sse2/;
+  specialize qw/vp9_fht4x4 sse2 msa/;
 
   add_proto qw/void vp9_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  specialize qw/vp9_fht8x8 sse2/;
+  specialize qw/vp9_fht8x8 sse2 msa/;
 
   add_proto qw/void vp9_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  specialize qw/vp9_fht16x16 sse2/;
+  specialize qw/vp9_fht16x16 sse2 msa/;
 
   add_proto qw/void vp9_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fwht4x4/, "$mmx_x86inc";
+  specialize qw/vp9_fwht4x4 msa/, "$mmx_x86inc";
 
   add_proto qw/void vp9_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vp9_fdct4x4_1 sse2/;
 
   add_proto qw/void vp9_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct4x4 sse2/;
+  specialize qw/vp9_fdct4x4 sse2 msa/;
 
   add_proto qw/void vp9_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct8x8_1 sse2 neon/;
+  specialize qw/vp9_fdct8x8_1 sse2 neon msa/;
 
   add_proto qw/void vp9_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct8x8 sse2 neon/, "$ssse3_x86_64";
+  specialize qw/vp9_fdct8x8 sse2 neon msa/, "$ssse3_x86_64";
 
   add_proto qw/void vp9_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct16x16_1 sse2/;
+  specialize qw/vp9_fdct16x16_1 sse2 msa/;
 
   add_proto qw/void vp9_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct16x16 sse2/;
+  specialize qw/vp9_fdct16x16 sse2 msa/;
 
   add_proto qw/void vp9_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct32x32_1 sse2/;
+  specialize qw/vp9_fdct32x32_1 sse2 msa/;
 
   add_proto qw/void vp9_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct32x32 sse2 avx2/;
+  specialize qw/vp9_fdct32x32 sse2 avx2 msa/;
 
   add_proto qw/void vp9_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_fdct32x32_rd sse2 avx2/;
+  specialize qw/vp9_fdct32x32_rd sse2 avx2 msa/;
 }
 
 #
 # Motion search
 #
 add_proto qw/int vp9_full_search_sad/, "const struct macroblock *x, const struct mv *ref_mv, int sad_per_bit, int distance, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv, struct mv *best_mv";
 specialize qw/vp9_full_search_sad sse3 sse4_1/;
 $vp9_full_search_sad_sse3=vp9_full_search_sadx3;
 $vp9_full_search_sad_sse4_1=vp9_full_search_sadx8;
 
 add_proto qw/int vp9_diamond_search_sad/, "const struct macroblock *x, const struct search_site_config *cfg,  struct mv *ref_mv, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv";
 specialize qw/vp9_diamond_search_sad/;
 
 add_proto qw/int vp9_full_range_search/, "const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv";
 specialize qw/vp9_full_range_search/;
 
 add_proto qw/void vp9_temporal_filter_apply/, "uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count";
-specialize qw/vp9_temporal_filter_apply sse2/;
+specialize qw/vp9_temporal_filter_apply sse2 msa/;
 
 if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
 
   add_proto qw/unsigned int vp9_highbd_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
   specialize qw/vp9_highbd_sub_pixel_variance64x64/, "$sse2_x86inc";
 
   add_proto qw/unsigned int vp9_highbd_sub_pixel_avg_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
   specialize qw/vp9_highbd_sub_pixel_avg_variance64x64/, "$sse2_x86inc";
 
   add_proto qw/unsigned int vp9_highbd_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
@@ skipping 292 matching lines @@
 
   add_proto qw/void vp9_highbd_temporal_filter_apply/, "uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count";
   specialize qw/vp9_highbd_temporal_filter_apply/;
 
 }
 # End vp9_high encoder functions
 
 }
 # end encoder functions
 1;