aboutsummaryrefslogtreecommitdiff
path: root/vp8/common
diff options
context:
space:
mode:
Diffstat (limited to 'vp8/common')
-rw-r--r--vp8/common/arm/neon/sixtappredict_neon.c9
-rw-r--r--vp8/common/blockd.h2
-rw-r--r--vp8/common/loongarch/sixtap_filter_lsx.c23
-rw-r--r--vp8/common/mips/msa/sixtap_filter_msa.c181
-rw-r--r--vp8/common/mips/msa/vp8_macros_msa.h258
-rw-r--r--vp8/common/onyx.h33
-rw-r--r--vp8/common/rtcd_defs.pl6
-rw-r--r--vp8/common/vp8_loopfilter.c2
8 files changed, 271 insertions, 243 deletions
diff --git a/vp8/common/arm/neon/sixtappredict_neon.c b/vp8/common/arm/neon/sixtappredict_neon.c
index 48e86d327..ee3c281f0 100644
--- a/vp8/common/arm/neon/sixtappredict_neon.c
+++ b/vp8/common/arm/neon/sixtappredict_neon.c
@@ -16,7 +16,7 @@
#include "vpx_ports/mem.h"
static const int8_t vp8_sub_pel_filters[8][8] = {
- { 0, 0, 128, 0, 0, 0, 0, 0 }, /* note that 1/8 pel positionyys are */
+ { 0, 0, 128, 0, 0, 0, 0, 0 }, /* note that 1/8 pel positions are */
{ 0, -6, 123, 12, -1, 0, 0, 0 }, /* just as per alpha -0.5 bicubic */
{ 2, -11, 108, 36, -8, 1, 0, 0 }, /* New 1/4 pel 6 tap filter */
{ 0, -9, 93, 50, -6, 0, 0, 0 },
@@ -781,7 +781,6 @@ void vp8_sixtap_predict8x4_neon(unsigned char *src_ptr, int src_pixels_per_line,
vst1_u8(dst_ptr, d8u8);
dst_ptr += dst_pitch;
vst1_u8(dst_ptr, d9u8);
- return;
}
void vp8_sixtap_predict8x8_neon(unsigned char *src_ptr, int src_pixels_per_line,
@@ -1250,7 +1249,6 @@ void vp8_sixtap_predict8x8_neon(unsigned char *src_ptr, int src_pixels_per_line,
vst1_u8(dst_ptr, d9u8);
dst_ptr += dst_pitch;
}
- return;
}
void vp8_sixtap_predict16x16_neon(unsigned char *src_ptr,
@@ -1504,7 +1502,9 @@ void vp8_sixtap_predict16x16_neon(unsigned char *src_ptr,
src += src_pixels_per_line;
d12u8 = vld1_u8(src);
d13u8 = vld1_u8(src + 8);
- d14u8 = vld1_u8(src + 16);
+ // Only 5 pixels are needed, avoid a potential out of bounds read.
+ d14u8 = vld1_u8(src + 13);
+ d14u8 = vext_u8(d14u8, d14u8, 3);
src += src_pixels_per_line;
__builtin_prefetch(src);
@@ -1726,5 +1726,4 @@ void vp8_sixtap_predict16x16_neon(unsigned char *src_ptr,
dst += dst_pitch;
}
}
- return;
}
diff --git a/vp8/common/blockd.h b/vp8/common/blockd.h
index 405443449..8300aad94 100644
--- a/vp8/common/blockd.h
+++ b/vp8/common/blockd.h
@@ -251,7 +251,7 @@ typedef struct macroblockd {
unsigned char update_mb_segmentation_data;
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
- unsigned char mb_segement_abs_delta;
+ unsigned char mb_segment_abs_delta;
/* Per frame flags that define which MB level features (such as quantizer or
* loop filter level) */
diff --git a/vp8/common/loongarch/sixtap_filter_lsx.c b/vp8/common/loongarch/sixtap_filter_lsx.c
index cd7ba5474..986763341 100644
--- a/vp8/common/loongarch/sixtap_filter_lsx.c
+++ b/vp8/common/loongarch/sixtap_filter_lsx.c
@@ -1706,21 +1706,22 @@ void vp8_sixtap_predict4x4_lsx(uint8_t *RESTRICT src, int32_t src_stride,
switch (xoffset) {
case 0: {
__m128i tp0;
- tp0 = __lsx_vinsgr2vr_w(tp0, src, 0);
- src += src_stride;
- tp0 = __lsx_vinsgr2vr_w(tp0, src, 0);
- src += src_stride;
- tp0 = __lsx_vinsgr2vr_w(tp0, src, 0);
- src += src_stride;
- tp0 = __lsx_vinsgr2vr_w(tp0, src, 0);
+ tp0 = __lsx_vldrepl_w(src, 0);
+ src += src_stride;
__lsx_vstelm_w(tp0, dst, 0, 0);
dst += dst_stride;
- __lsx_vstelm_w(tp0, dst, 0, 1);
+ tp0 = __lsx_vldrepl_w(src, 0);
+ src += src_stride;
+ __lsx_vstelm_w(tp0, dst, 0, 0);
dst += dst_stride;
- __lsx_vstelm_w(tp0, dst, 0, 2);
+ tp0 = __lsx_vldrepl_w(src, 0);
+ src += src_stride;
+ __lsx_vstelm_w(tp0, dst, 0, 0);
dst += dst_stride;
- __lsx_vstelm_w(tp0, dst, 0, 3);
+ tp0 = __lsx_vldrepl_w(src, 0);
+ __lsx_vstelm_w(tp0, dst, 0, 0);
+
break;
}
case 2:
@@ -1865,7 +1866,7 @@ void vp8_sixtap_predict16x16_lsx(uint8_t *RESTRICT src, int32_t src_stride,
case 1:
Predict16x16Funcs1[3](src, src_stride, dst, dst_stride,
- h_filter, v_filter + 1, 16);
+ h_filter + 1, v_filter + 1, 16);
break;
}
break;
diff --git a/vp8/common/mips/msa/sixtap_filter_msa.c b/vp8/common/mips/msa/sixtap_filter_msa.c
index b0affcff0..3a1bb7cd5 100644
--- a/vp8/common/mips/msa/sixtap_filter_msa.c
+++ b/vp8/common/mips/msa/sixtap_filter_msa.c
@@ -35,101 +35,134 @@ static const uint8_t vp8_mc_filt_mask_arr[16 * 3] = {
#define HORIZ_6TAP_FILT(src0, src1, mask0, mask1, mask2, filt_h0, filt_h1, \
filt_h2) \
({ \
- v16i8 vec0_m, vec1_m, vec2_m; \
- v8i16 hz_out_m; \
+ v16i8 _6tap_vec0_m, _6tap_vec1_m, _6tap_vec2_m; \
+ v8i16 _6tap_out_m; \
\
VSHF_B3_SB(src0, src1, src0, src1, src0, src1, mask0, mask1, mask2, \
- vec0_m, vec1_m, vec2_m); \
- hz_out_m = \
- DPADD_SH3_SH(vec0_m, vec1_m, vec2_m, filt_h0, filt_h1, filt_h2); \
+ _6tap_vec0_m, _6tap_vec1_m, _6tap_vec2_m); \
+ _6tap_out_m = DPADD_SH3_SH(_6tap_vec0_m, _6tap_vec1_m, _6tap_vec2_m, \
+ filt_h0, filt_h1, filt_h2); \
\
- hz_out_m = __msa_srari_h(hz_out_m, VP8_FILTER_SHIFT); \
- hz_out_m = __msa_sat_s_h(hz_out_m, 7); \
+ _6tap_out_m = __msa_srari_h(_6tap_out_m, VP8_FILTER_SHIFT); \
+ _6tap_out_m = __msa_sat_s_h(_6tap_out_m, 7); \
\
- hz_out_m; \
+ _6tap_out_m; \
})
#define HORIZ_6TAP_4WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
mask2, filt0, filt1, filt2, out0, out1) \
{ \
- v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m; \
+ v16i8 _6tap_4wid_vec0_m, _6tap_4wid_vec1_m, _6tap_4wid_vec2_m, \
+ _6tap_4wid_vec3_m, _6tap_4wid_vec4_m, _6tap_4wid_vec5_m; \
\
- VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, vec0_m, vec1_m); \
- DOTP_SB2_SH(vec0_m, vec1_m, filt0, filt0, out0, out1); \
- VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, vec2_m, vec3_m); \
- DPADD_SB2_SH(vec2_m, vec3_m, filt1, filt1, out0, out1); \
- VSHF_B2_SB(src0, src1, src2, src3, mask2, mask2, vec4_m, vec5_m); \
- DPADD_SB2_SH(vec4_m, vec5_m, filt2, filt2, out0, out1); \
+ VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, _6tap_4wid_vec0_m, \
+ _6tap_4wid_vec1_m); \
+ DOTP_SB2_SH(_6tap_4wid_vec0_m, _6tap_4wid_vec1_m, filt0, filt0, out0, \
+ out1); \
+ VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, _6tap_4wid_vec2_m, \
+ _6tap_4wid_vec3_m); \
+ DPADD_SB2_SH(_6tap_4wid_vec2_m, _6tap_4wid_vec3_m, filt1, filt1, out0, \
+ out1); \
+ VSHF_B2_SB(src0, src1, src2, src3, mask2, mask2, _6tap_4wid_vec4_m, \
+ _6tap_4wid_vec5_m); \
+ DPADD_SB2_SH(_6tap_4wid_vec4_m, _6tap_4wid_vec5_m, filt2, filt2, out0, \
+ out1); \
}
-#define HORIZ_6TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
- mask2, filt0, filt1, filt2, out0, out1, \
- out2, out3) \
- { \
- v16i8 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m; \
- \
- VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); \
- VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); \
- DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, \
- out0, out1, out2, out3); \
- VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec0_m, vec1_m); \
- VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec2_m, vec3_m); \
- VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, vec4_m, vec5_m); \
- VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, vec6_m, vec7_m); \
- DPADD_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt1, filt1, filt1, filt1, \
- out0, out1, out2, out3); \
- DPADD_SB4_SH(vec4_m, vec5_m, vec6_m, vec7_m, filt2, filt2, filt2, filt2, \
- out0, out1, out2, out3); \
+#define HORIZ_6TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
+ mask2, filt0, filt1, filt2, out0, out1, \
+ out2, out3) \
+ { \
+ v16i8 _6tap_8wid_vec0_m, _6tap_8wid_vec1_m, _6tap_8wid_vec2_m, \
+ _6tap_8wid_vec3_m, _6tap_8wid_vec4_m, _6tap_8wid_vec5_m, \
+ _6tap_8wid_vec6_m, _6tap_8wid_vec7_m; \
+ \
+ VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, _6tap_8wid_vec0_m, \
+ _6tap_8wid_vec1_m); \
+ VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, _6tap_8wid_vec2_m, \
+ _6tap_8wid_vec3_m); \
+ DOTP_SB4_SH(_6tap_8wid_vec0_m, _6tap_8wid_vec1_m, _6tap_8wid_vec2_m, \
+ _6tap_8wid_vec3_m, filt0, filt0, filt0, filt0, out0, out1, \
+ out2, out3); \
+ VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, _6tap_8wid_vec0_m, \
+ _6tap_8wid_vec1_m); \
+ VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, _6tap_8wid_vec2_m, \
+ _6tap_8wid_vec3_m); \
+ VSHF_B2_SB(src0, src0, src1, src1, mask2, mask2, _6tap_8wid_vec4_m, \
+ _6tap_8wid_vec5_m); \
+ VSHF_B2_SB(src2, src2, src3, src3, mask2, mask2, _6tap_8wid_vec6_m, \
+ _6tap_8wid_vec7_m); \
+ DPADD_SB4_SH(_6tap_8wid_vec0_m, _6tap_8wid_vec1_m, _6tap_8wid_vec2_m, \
+ _6tap_8wid_vec3_m, filt1, filt1, filt1, filt1, out0, out1, \
+ out2, out3); \
+ DPADD_SB4_SH(_6tap_8wid_vec4_m, _6tap_8wid_vec5_m, _6tap_8wid_vec6_m, \
+ _6tap_8wid_vec7_m, filt2, filt2, filt2, filt2, out0, out1, \
+ out2, out3); \
}
-#define FILT_4TAP_DPADD_S_H(vec0, vec1, filt0, filt1) \
- ({ \
- v8i16 tmp0; \
- \
- tmp0 = __msa_dotp_s_h((v16i8)vec0, (v16i8)filt0); \
- tmp0 = __msa_dpadd_s_h(tmp0, (v16i8)vec1, (v16i8)filt1); \
- \
- tmp0; \
- })
-
-#define HORIZ_4TAP_FILT(src0, src1, mask0, mask1, filt_h0, filt_h1) \
+#define FILT_4TAP_DPADD_S_H(vec0, vec1, filt0, filt1) \
({ \
- v16i8 vec0_m, vec1_m; \
- v8i16 hz_out_m; \
- \
- VSHF_B2_SB(src0, src1, src0, src1, mask0, mask1, vec0_m, vec1_m); \
- hz_out_m = FILT_4TAP_DPADD_S_H(vec0_m, vec1_m, filt_h0, filt_h1); \
+ v8i16 _4tap_dpadd_tmp0; \
\
- hz_out_m = __msa_srari_h(hz_out_m, VP8_FILTER_SHIFT); \
- hz_out_m = __msa_sat_s_h(hz_out_m, 7); \
+ _4tap_dpadd_tmp0 = __msa_dotp_s_h((v16i8)vec0, (v16i8)filt0); \
+ _4tap_dpadd_tmp0 = \
+ __msa_dpadd_s_h(_4tap_dpadd_tmp0, (v16i8)vec1, (v16i8)filt1); \
\
- hz_out_m; \
+ _4tap_dpadd_tmp0; \
})
-#define HORIZ_4TAP_4WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
- filt0, filt1, out0, out1) \
- { \
- v16i8 vec0_m, vec1_m, vec2_m, vec3_m; \
- \
- VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, vec0_m, vec1_m); \
- DOTP_SB2_SH(vec0_m, vec1_m, filt0, filt0, out0, out1); \
- VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, vec2_m, vec3_m); \
- DPADD_SB2_SH(vec2_m, vec3_m, filt1, filt1, out0, out1); \
+#define HORIZ_4TAP_FILT(src0, src1, mask0, mask1, filt_h0, filt_h1) \
+ ({ \
+ v16i8 _4tap_vec0_m, _4tap_vec1_m; \
+ v8i16 _4tap_out_m; \
+ \
+ VSHF_B2_SB(src0, src1, src0, src1, mask0, mask1, _4tap_vec0_m, \
+ _4tap_vec1_m); \
+ _4tap_out_m = \
+ FILT_4TAP_DPADD_S_H(_4tap_vec0_m, _4tap_vec1_m, filt_h0, filt_h1); \
+ \
+ _4tap_out_m = __msa_srari_h(_4tap_out_m, VP8_FILTER_SHIFT); \
+ _4tap_out_m = __msa_sat_s_h(_4tap_out_m, 7); \
+ \
+ _4tap_out_m; \
+ })
+
+#define HORIZ_4TAP_4WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
+ filt0, filt1, out0, out1) \
+ { \
+ v16i8 _4tap_4wid_vec0_m, _4tap_4wid_vec1_m, _4tap_4wid_vec2_m, \
+ _4tap_4wid_vec3_m; \
+ \
+ VSHF_B2_SB(src0, src1, src2, src3, mask0, mask0, _4tap_4wid_vec0_m, \
+ _4tap_4wid_vec1_m); \
+ DOTP_SB2_SH(_4tap_4wid_vec0_m, _4tap_4wid_vec1_m, filt0, filt0, out0, \
+ out1); \
+ VSHF_B2_SB(src0, src1, src2, src3, mask1, mask1, _4tap_4wid_vec2_m, \
+ _4tap_4wid_vec3_m); \
+ DPADD_SB2_SH(_4tap_4wid_vec2_m, _4tap_4wid_vec3_m, filt1, filt1, out0, \
+ out1); \
}
-#define HORIZ_4TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
- filt0, filt1, out0, out1, out2, out3) \
- { \
- v16i8 vec0_m, vec1_m, vec2_m, vec3_m; \
- \
- VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, vec0_m, vec1_m); \
- VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, vec2_m, vec3_m); \
- DOTP_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt0, filt0, filt0, filt0, \
- out0, out1, out2, out3); \
- VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, vec0_m, vec1_m); \
- VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, vec2_m, vec3_m); \
- DPADD_SB4_SH(vec0_m, vec1_m, vec2_m, vec3_m, filt1, filt1, filt1, filt1, \
- out0, out1, out2, out3); \
+#define HORIZ_4TAP_8WID_4VECS_FILT(src0, src1, src2, src3, mask0, mask1, \
+ filt0, filt1, out0, out1, out2, out3) \
+ { \
+ v16i8 _4tap_8wid_vec0_m, _4tap_8wid_vec1_m, _4tap_8wid_vec2_m, \
+ _4tap_8wid_vec3_m; \
+ \
+ VSHF_B2_SB(src0, src0, src1, src1, mask0, mask0, _4tap_8wid_vec0_m, \
+ _4tap_8wid_vec1_m); \
+ VSHF_B2_SB(src2, src2, src3, src3, mask0, mask0, _4tap_8wid_vec2_m, \
+ _4tap_8wid_vec3_m); \
+ DOTP_SB4_SH(_4tap_8wid_vec0_m, _4tap_8wid_vec1_m, _4tap_8wid_vec2_m, \
+ _4tap_8wid_vec3_m, filt0, filt0, filt0, filt0, out0, out1, \
+ out2, out3); \
+ VSHF_B2_SB(src0, src0, src1, src1, mask1, mask1, _4tap_8wid_vec0_m, \
+ _4tap_8wid_vec1_m); \
+ VSHF_B2_SB(src2, src2, src3, src3, mask1, mask1, _4tap_8wid_vec2_m, \
+ _4tap_8wid_vec3_m); \
+ DPADD_SB4_SH(_4tap_8wid_vec0_m, _4tap_8wid_vec1_m, _4tap_8wid_vec2_m, \
+ _4tap_8wid_vec3_m, filt1, filt1, filt1, filt1, out0, out1, \
+ out2, out3); \
}
static void common_hz_6t_4x4_msa(uint8_t *RESTRICT src, int32_t src_stride,
diff --git a/vp8/common/mips/msa/vp8_macros_msa.h b/vp8/common/mips/msa/vp8_macros_msa.h
index 7cb3c9869..cc85b9a1f 100644
--- a/vp8/common/mips/msa/vp8_macros_msa.h
+++ b/vp8/common/mips/msa/vp8_macros_msa.h
@@ -40,160 +40,160 @@
#define ST_SW(...) ST_W(v4i32, __VA_ARGS__)
#if (__mips_isa_rev >= 6)
-#define LW(psrc) \
- ({ \
- const uint8_t *psrc_m = (const uint8_t *)(psrc); \
- uint32_t val_m; \
- \
- asm volatile("lw %[val_m], %[psrc_m] \n\t" \
- \
- : [val_m] "=r"(val_m) \
- : [psrc_m] "m"(*psrc_m)); \
- \
- val_m; \
+#define LW(psrc) \
+ ({ \
+ const uint8_t *lw_psrc_m = (const uint8_t *)(psrc); \
+ uint32_t lw_val_m; \
+ \
+ asm volatile("lw %[lw_val_m], %[lw_psrc_m] \n\t" \
+ \
+ : [lw_val_m] "=r"(lw_val_m) \
+ : [lw_psrc_m] "m"(*lw_psrc_m)); \
+ \
+ lw_val_m; \
})
#if (__mips == 64)
-#define LD(psrc) \
- ({ \
- const uint8_t *psrc_m = (const uint8_t *)(psrc); \
- uint64_t val_m = 0; \
- \
- asm volatile("ld %[val_m], %[psrc_m] \n\t" \
- \
- : [val_m] "=r"(val_m) \
- : [psrc_m] "m"(*psrc_m)); \
- \
- val_m; \
+#define LD(psrc) \
+ ({ \
+ const uint8_t *ld_psrc_m = (const uint8_t *)(psrc); \
+ uint64_t ld_val_m = 0; \
+ \
+ asm volatile("ld %[ld_val_m], %[ld_psrc_m] \n\t" \
+ \
+ : [ld_val_m] "=r"(ld_val_m) \
+ : [ld_psrc_m] "m"(*ld_psrc_m)); \
+ \
+ ld_val_m; \
})
#else // !(__mips == 64)
-#define LD(psrc) \
- ({ \
- const uint8_t *psrc_ld = (const uint8_t *)(psrc); \
- uint32_t val0_m, val1_m; \
- uint64_t val_m = 0; \
- \
- val0_m = LW(psrc_ld); \
- val1_m = LW(psrc_ld + 4); \
- \
- val_m = (uint64_t)(val1_m); \
- val_m = (uint64_t)((val_m << 32) & 0xFFFFFFFF00000000); \
- val_m = (uint64_t)(val_m | (uint64_t)val0_m); \
- \
- val_m; \
+#define LD(psrc) \
+ ({ \
+ const uint8_t *ld_psrc_m = (const uint8_t *)(psrc); \
+ uint32_t ld_val0_m, ld_val1_m; \
+ uint64_t ld_val_m = 0; \
+ \
+ ld_val0_m = LW(ld_psrc_m); \
+ ld_val1_m = LW(ld_psrc_m + 4); \
+ \
+ ld_val_m = (uint64_t)(ld_val1_m); \
+ ld_val_m = (uint64_t)((ld_val_m << 32) & 0xFFFFFFFF00000000); \
+ ld_val_m = (uint64_t)(ld_val_m | (uint64_t)ld_val0_m); \
+ \
+ ld_val_m; \
})
#endif // (__mips == 64)
-#define SH(val, pdst) \
- { \
- uint8_t *pdst_m = (uint8_t *)(pdst); \
- const uint16_t val_m = (val); \
- \
- asm volatile("sh %[val_m], %[pdst_m] \n\t" \
- \
- : [pdst_m] "=m"(*pdst_m) \
- : [val_m] "r"(val_m)); \
+#define SH(val, pdst) \
+ { \
+ uint8_t *sh_pdst_m = (uint8_t *)(pdst); \
+ const uint16_t sh_val_m = (val); \
+ \
+ asm volatile("sh %[sh_val_m], %[sh_pdst_m] \n\t" \
+ \
+ : [sh_pdst_m] "=m"(*sh_pdst_m) \
+ : [sh_val_m] "r"(sh_val_m)); \
}
-#define SW(val, pdst) \
- { \
- uint8_t *pdst_m = (uint8_t *)(pdst); \
- const uint32_t val_m = (val); \
- \
- asm volatile("sw %[val_m], %[pdst_m] \n\t" \
- \
- : [pdst_m] "=m"(*pdst_m) \
- : [val_m] "r"(val_m)); \
+#define SW(val, pdst) \
+ { \
+ uint8_t *sw_pdst_m = (uint8_t *)(pdst); \
+ const uint32_t sw_val_m = (val); \
+ \
+ asm volatile("sw %[sw_val_m], %[sw_pdst_m] \n\t" \
+ \
+ : [sw_pdst_m] "=m"(*sw_pdst_m) \
+ : [sw_val_m] "r"(sw_val_m)); \
}
-#define SD(val, pdst) \
- { \
- uint8_t *pdst_m = (uint8_t *)(pdst); \
- const uint64_t val_m = (val); \
- \
- asm volatile("sd %[val_m], %[pdst_m] \n\t" \
- \
- : [pdst_m] "=m"(*pdst_m) \
- : [val_m] "r"(val_m)); \
+#define SD(val, pdst) \
+ { \
+ uint8_t *sd_pdst_m = (uint8_t *)(pdst); \
+ const uint64_t sd_val_m = (val); \
+ \
+ asm volatile("sd %[sd_val_m], %[sd_pdst_m] \n\t" \
+ \
+ : [sd_pdst_m] "=m"(*sd_pdst_m) \
+ : [sd_val_m] "r"(sd_val_m)); \
}
#else // !(__mips_isa_rev >= 6)
-#define LW(psrc) \
- ({ \
- const uint8_t *psrc_m = (const uint8_t *)(psrc); \
- uint32_t val_m; \
- \
- asm volatile( \
- "lwr %[val_m], 0(%[psrc_m]) \n\t" \
- "lwl %[val_m], 3(%[psrc_m]) \n\t" \
- : [val_m] "=&r"(val_m) \
- : [psrc_m] "r"(psrc_m)); \
- \
- val_m; \
+#define LW(psrc) \
+ ({ \
+ const uint8_t *lw_psrc_m = (const uint8_t *)(psrc); \
+ uint32_t lw_val_m; \
+ \
+ asm volatile( \
+ "lwr %[lw_val_m], 0(%[lw_psrc_m]) \n\t" \
+ "lwl %[lw_val_m], 3(%[lw_psrc_m]) \n\t" \
+ : [lw_val_m] "=&r"(lw_val_m) \
+ : [lw_psrc_m] "r"(lw_psrc_m)); \
+ \
+ lw_val_m; \
})
#if (__mips == 64)
-#define LD(psrc) \
- ({ \
- const uint8_t *psrc_m = (const uint8_t *)(psrc); \
- uint64_t val_m = 0; \
- \
- asm volatile( \
- "ldr %[val_m], 0(%[psrc_m]) \n\t" \
- "ldl %[val_m], 7(%[psrc_m]) \n\t" \
- : [val_m] "=&r"(val_m) \
- : [psrc_m] "r"(psrc_m)); \
- \
- val_m; \
+#define LD(psrc) \
+ ({ \
+ const uint8_t *ld_psrc_m = (const uint8_t *)(psrc); \
+ uint64_t ld_val_m = 0; \
+ \
+ asm volatile( \
+ "ldr %[ld_val_m], 0(%[ld_psrc_m]) \n\t" \
+ "ldl %[ld_val_m], 7(%[ld_psrc_m]) \n\t" \
+ : [ld_val_m] "=&r"(ld_val_m) \
+ : [ld_psrc_m] "r"(ld_psrc_m)); \
+ \
+ ld_val_m; \
})
#else // !(__mips == 64)
-#define LD(psrc) \
- ({ \
- const uint8_t *psrc_m1 = (const uint8_t *)(psrc); \
- uint32_t val0_m, val1_m; \
- uint64_t val_m = 0; \
- \
- val0_m = LW(psrc_m1); \
- val1_m = LW(psrc_m1 + 4); \
- \
- val_m = (uint64_t)(val1_m); \
- val_m = (uint64_t)((val_m << 32) & 0xFFFFFFFF00000000); \
- val_m = (uint64_t)(val_m | (uint64_t)val0_m); \
- \
- val_m; \
+#define LD(psrc) \
+ ({ \
+ const uint8_t *ld_psrc_m1 = (const uint8_t *)(psrc); \
+ uint32_t ld_val0_m, ld_val1_m; \
+ uint64_t ld_val_m = 0; \
+ \
+ ld_val0_m = LW(ld_psrc_m1); \
+ ld_val1_m = LW(ld_psrc_m1 + 4); \
+ \
+ ld_val_m = (uint64_t)(ld_val1_m); \
+ ld_val_m = (uint64_t)((ld_val_m << 32) & 0xFFFFFFFF00000000); \
+ ld_val_m = (uint64_t)(ld_val_m | (uint64_t)ld_val0_m); \
+ \
+ ld_val_m; \
})
#endif // (__mips == 64)
-#define SH(val, pdst) \
- { \
- uint8_t *pdst_m = (uint8_t *)(pdst); \
- const uint16_t val_m = (val); \
- \
- asm volatile("ush %[val_m], %[pdst_m] \n\t" \
- \
- : [pdst_m] "=m"(*pdst_m) \
- : [val_m] "r"(val_m)); \
+#define SH(val, pdst) \
+ { \
+ uint8_t *sh_pdst_m = (uint8_t *)(pdst); \
+ const uint16_t sh_val_m = (val); \
+ \
+ asm volatile("ush %[sh_val_m], %[sh_pdst_m] \n\t" \
+ \
+ : [sh_pdst_m] "=m"(*sh_pdst_m) \
+ : [sh_val_m] "r"(sh_val_m)); \
}
-#define SW(val, pdst) \
- { \
- uint8_t *pdst_m = (uint8_t *)(pdst); \
- const uint32_t val_m = (val); \
- \
- asm volatile("usw %[val_m], %[pdst_m] \n\t" \
- \
- : [pdst_m] "=m"(*pdst_m) \
- : [val_m] "r"(val_m)); \
+#define SW(val, pdst) \
+ { \
+ uint8_t *sw_pdst_m = (uint8_t *)(pdst); \
+ const uint32_t sw_val_m = (val); \
+ \
+ asm volatile("usw %[sw_val_m], %[sw_pdst_m] \n\t" \
+ \
+ : [sw_pdst_m] "=m"(*sw_pdst_m) \
+ : [sw_val_m] "r"(sw_val_m)); \
}
-#define SD(val, pdst) \
- { \
- uint8_t *pdst_m1 = (uint8_t *)(pdst); \
- uint32_t val0_m, val1_m; \
- \
- val0_m = (uint32_t)((val)&0x00000000FFFFFFFF); \
- val1_m = (uint32_t)(((val) >> 32) & 0x00000000FFFFFFFF); \
- \
- SW(val0_m, pdst_m1); \
- SW(val1_m, pdst_m1 + 4); \
+#define SD(val, pdst) \
+ { \
+ uint8_t *sd_pdst_m1 = (uint8_t *)(pdst); \
+ uint32_t sd_val0_m, sd_val1_m; \
+ \
+ sd_val0_m = (uint32_t)((val)&0x00000000FFFFFFFF); \
+ sd_val1_m = (uint32_t)(((val) >> 32) & 0x00000000FFFFFFFF); \
+ \
+ SW(sd_val0_m, sd_pdst_m1); \
+ SW(sd_val1_m, sd_pdst_m1 + 4); \
}
#endif // (__mips_isa_rev >= 6)
diff --git a/vp8/common/onyx.h b/vp8/common/onyx.h
index 05c72df3f..1b70ea5db 100644
--- a/vp8/common/onyx.h
+++ b/vp8/common/onyx.h
@@ -27,13 +27,6 @@ struct VP8_COMP;
/* Create/destroy static data structures. */
typedef enum {
- NORMAL = 0,
- FOURFIVE = 1,
- THREEFIVE = 2,
- ONETWO = 3
-} VPX_SCALING;
-
-typedef enum {
USAGE_LOCAL_FILE_PLAYBACK = 0x0,
USAGE_STREAM_FROM_SERVER = 0x1,
USAGE_CONSTRAINED_QUALITY = 0x2,
@@ -58,19 +51,19 @@ typedef enum {
#include <assert.h>
static INLINE void Scale2Ratio(int mode, int *hr, int *hs) {
switch (mode) {
- case NORMAL:
+ case VP8E_NORMAL:
*hr = 1;
*hs = 1;
break;
- case FOURFIVE:
+ case VP8E_FOURFIVE:
*hr = 4;
*hs = 5;
break;
- case THREEFIVE:
+ case VP8E_THREEFIVE:
*hr = 3;
*hs = 5;
break;
- case ONETWO:
+ case VP8E_ONETWO:
*hr = 1;
*hs = 2;
break;
@@ -90,7 +83,14 @@ typedef struct {
int Width;
int Height;
struct vpx_rational timebase;
- unsigned int target_bandwidth; /* kilobits per second */
+ /* In either kilobits per second or bits per second, depending on which
+ * copy of oxcf this is in.
+ * - ctx->oxcf.target_bandwidth is in kilobits per second. See
+ * set_vp8e_config().
+ * - ctx->cpi->oxcf.target_bandwidth in is bits per second. See
+ * vp8_change_config().
+ */
+ unsigned int target_bandwidth;
/* Parameter used for applying denoiser.
* For temporal denoiser: noise_sensitivity = 0 means off,
@@ -221,6 +221,7 @@ typedef struct {
/* Temporal scaling parameters */
unsigned int number_of_layers;
+ /* kilobits per second */
unsigned int target_bitrate[VPX_TS_MAX_PERIODICITY];
unsigned int rate_decimator[VPX_TS_MAX_PERIODICITY];
unsigned int periodicity;
@@ -243,11 +244,11 @@ typedef struct {
void vp8_initialize();
-struct VP8_COMP *vp8_create_compressor(VP8_CONFIG *oxcf);
+struct VP8_COMP *vp8_create_compressor(const VP8_CONFIG *oxcf);
void vp8_remove_compressor(struct VP8_COMP **comp);
void vp8_init_config(struct VP8_COMP *onyx, VP8_CONFIG *oxcf);
-void vp8_change_config(struct VP8_COMP *cpi, VP8_CONFIG *oxcf);
+void vp8_change_config(struct VP8_COMP *cpi, const VP8_CONFIG *oxcf);
int vp8_receive_raw_frame(struct VP8_COMP *cpi, unsigned int frame_flags,
YV12_BUFFER_CONFIG *sd, int64_t time_stamp,
@@ -273,8 +274,8 @@ int vp8_set_roimap(struct VP8_COMP *cpi, unsigned char *map, unsigned int rows,
unsigned int threshold[4]);
int vp8_set_active_map(struct VP8_COMP *cpi, unsigned char *map,
unsigned int rows, unsigned int cols);
-int vp8_set_internal_size(struct VP8_COMP *cpi, VPX_SCALING horiz_mode,
- VPX_SCALING vert_mode);
+int vp8_set_internal_size(struct VP8_COMP *cpi, VPX_SCALING_MODE horiz_mode,
+ VPX_SCALING_MODE vert_mode);
int vp8_get_quantizer(struct VP8_COMP *cpi);
#ifdef __cplusplus
diff --git a/vp8/common/rtcd_defs.pl b/vp8/common/rtcd_defs.pl
index 739a61284..12b474d93 100644
--- a/vp8/common/rtcd_defs.pl
+++ b/vp8/common/rtcd_defs.pl
@@ -127,12 +127,6 @@ specialize qw/vp8_copy_mem8x4 mmx neon dspr2 msa mmi/;
#
if (vpx_config("CONFIG_POSTPROC") eq "yes") {
- add_proto qw/void vp8_blend_mb_inner/, "unsigned char *y, unsigned char *u, unsigned char *v, int y_1, int u_1, int v_1, int alpha, int stride";
-
- add_proto qw/void vp8_blend_mb_outer/, "unsigned char *y, unsigned char *u, unsigned char *v, int y_1, int u_1, int v_1, int alpha, int stride";
-
- add_proto qw/void vp8_blend_b/, "unsigned char *y, unsigned char *u, unsigned char *v, int y_1, int u_1, int v_1, int alpha, int stride";
-
add_proto qw/void vp8_filter_by_weight16x16/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight";
specialize qw/vp8_filter_by_weight16x16 sse2 msa/;
diff --git a/vp8/common/vp8_loopfilter.c b/vp8/common/vp8_loopfilter.c
index 9c9e5f351..4576c1853 100644
--- a/vp8/common/vp8_loopfilter.c
+++ b/vp8/common/vp8_loopfilter.c
@@ -111,7 +111,7 @@ void vp8_loop_filter_frame_init(VP8_COMMON *cm, MACROBLOCKD *mbd,
/* Note the baseline filter values for each segment */
if (mbd->segmentation_enabled) {
- if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA) {
+ if (mbd->mb_segment_abs_delta == SEGMENT_ABSDATA) {
lvl_seg = mbd->segment_feature_data[MB_LVL_ALT_LF][seg];
} else { /* Delta Value */
lvl_seg += mbd->segment_feature_data[MB_LVL_ALT_LF][seg];
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-24 19:05:08 +0000