[FFmpeg-devel] [PATCH v2] avutil/mips: refactor msa SLDI_Bn_0 and SLDI_Bn macros.
Shiyou Yin
yinshiyou-hf at loongson.cn
Wed Aug 7 05:49:19 EEST 2019
LGTM.
>-----Original Message-----
>From: ffmpeg-devel-bounces at ffmpeg.org [mailto:ffmpeg-devel-bounces at ffmpeg.org] On Behalf Of gxw
>Sent: Tuesday, August 6, 2019 7:11 PM
>To: ffmpeg-devel at ffmpeg.org
>Subject: [FFmpeg-devel] [PATCH v2] avutil/mips: refactor msa SLDI_Bn_0 and SLDI_Bn macros.
>
>Changing details as following:
>1. The previous order of parameters are irregular and difficult to
> understand. Adjust the order of the parameters according to the
> rule: (RTYPE, input registers, input mask/input index/..., output registers).
> Most of the existing msa macros follow the rule.
>2. Remove the redundant macro SLDI_Bn_0 and use SLDI_Bn instead.
>---
> libavcodec/mips/h264dsp_msa.c | 9 ++--
> libavcodec/mips/h264qpel_msa.c | 64 ++++++++++++++--------------
> libavcodec/mips/hevc_lpf_sao_msa.c | 70 ++++++++++++++++---------------
> libavcodec/mips/hevcpred_msa.c | 30 ++++++-------
> libavcodec/mips/hpeldsp_msa.c | 66 ++++++++++++++++++-----------
> libavcodec/mips/me_cmp_msa.c | 8 ++--
> libavcodec/mips/qpeldsp_msa.c | 84 ++++++++++++++++++-------------------
> libavcodec/mips/vp8_mc_msa.c | 4 +-
> libavcodec/mips/vp9_idct_msa.c | 3 +-
> libavcodec/mips/vp9_lpf_msa.c | 3 +-
> libavcodec/mips/vp9_mc_msa.c | 16 +++----
> libavutil/mips/generic_macros_msa.h | 80 ++++++++++++++---------------------
> 12 files changed, 222 insertions(+), 215 deletions(-)
>
>diff --git a/libavcodec/mips/h264dsp_msa.c b/libavcodec/mips/h264dsp_msa.c
>index 89fe399..c4ba8c4 100644
>--- a/libavcodec/mips/h264dsp_msa.c
>+++ b/libavcodec/mips/h264dsp_msa.c
>@@ -620,7 +620,7 @@ static void avc_biwgt_8x16_msa(uint8_t *src, uint8_t *dst, int32_t stride,
> \
> out0 = (v16u8) __msa_ilvr_b((v16i8) in1, (v16i8) in0); \
> out1 = (v16u8) __msa_sldi_b(zero_m, (v16i8) out0, 2); \
>- SLDI_B2_0_UB(out1, out2, out2, out3, 2); \
>+ SLDI_B2_UB(zero_m, out1, zero_m, out2, 2, out2, out3); \
> }
>
> #define AVC_LPF_H_2BYTE_CHROMA_422(src, stride, tc_val, alpha, beta, res) \
>@@ -1025,7 +1025,8 @@ static void avc_h_loop_filter_luma_mbaff_intra_msa(uint8_t *src, int32_t
>stride,
>
> ILVR_W2_SB(tmp2, tmp0, tmp3, tmp1, src6, src3);
> ILVL_W2_SB(tmp2, tmp0, tmp3, tmp1, src1, src5);
>- SLDI_B4_0_SB(src6, src1, src3, src5, src0, src2, src4, src7, 8);
>+ SLDI_B4_SB(zeros, src6, zeros, src1, zeros, src3, zeros, src5,
>+ 8, src0, src2, src4, src7);
>
> p0_asub_q0 = __msa_asub_u_b((v16u8) src2, (v16u8) src3);
> p1_asub_p0 = __msa_asub_u_b((v16u8) src1, (v16u8) src2);
>@@ -1116,10 +1117,10 @@ static void avc_h_loop_filter_luma_mbaff_intra_msa(uint8_t *src, int32_t
>stride,
> ILVRL_H2_SH(zeros, dst2_x, tmp2, tmp3);
>
> ILVR_W2_UB(tmp2, tmp0, tmp3, tmp1, dst0, dst4);
>- SLDI_B2_0_UB(dst0, dst4, dst1, dst5, 8);
>+ SLDI_B2_UB(zeros, dst0, zeros, dst4, 8, dst1, dst5);
> dst2_x = (v16u8) __msa_ilvl_w((v4i32) tmp2, (v4i32) tmp0);
> dst2_y = (v16u8) __msa_ilvl_w((v4i32) tmp3, (v4i32) tmp1);
>- SLDI_B2_0_UB(dst2_x, dst2_y, dst3_x, dst3_y, 8);
>+ SLDI_B2_UB(zeros, dst2_x, zeros, dst2_y, 8, dst3_x, dst3_y);
>
> out0 = __msa_copy_u_w((v4i32) dst0, 0);
> out1 = __msa_copy_u_h((v8i16) dst0, 2);
>diff --git a/libavcodec/mips/h264qpel_msa.c b/libavcodec/mips/h264qpel_msa.c
>index df7e3e2..e435c18 100644
>--- a/libavcodec/mips/h264qpel_msa.c
>+++ b/libavcodec/mips/h264qpel_msa.c
>@@ -790,8 +790,8 @@ void ff_put_h264_qpel16_mc10_msa(uint8_t *dst, const uint8_t *src,
> minus5b, res4, res5, res6, res7);
> DPADD_SB4_SH(vec2, vec5, vec8, vec11, plus20b, plus20b, plus20b,
> plus20b, res4, res5, res6, res7);
>- SLDI_B2_SB(src1, src3, src0, src2, src0, src2, 2);
>- SLDI_B2_SB(src5, src7, src4, src6, src4, src6, 2);
>+ SLDI_B4_SB(src1, src0, src3, src2, src5, src4, src7, src6, 2,
>+ src0, src2, src4, src6);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
> SRARI_H4_SH(res4, res5, res6, res7, 5);
> SAT_SH4_SH(res0, res1, res2, res3, 7);
>@@ -858,8 +858,8 @@ void ff_put_h264_qpel16_mc30_msa(uint8_t *dst, const uint8_t *src,
> minus5b, res4, res5, res6, res7);
> DPADD_SB4_SH(vec2, vec5, vec8, vec11, plus20b, plus20b, plus20b,
> plus20b, res4, res5, res6, res7);
>- SLDI_B2_SB(src1, src3, src0, src2, src0, src2, 3);
>- SLDI_B2_SB(src5, src7, src4, src6, src4, src6, 3);
>+ SLDI_B4_SB(src1, src0, src3, src2, src5, src4, src7, src6, 3,
>+ src0, src2, src4, src6);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
> SRARI_H4_SH(res4, res5, res6, res7, 5);
> SAT_SH4_SH(res0, res1, res2, res3, 7);
>@@ -911,10 +911,10 @@ void ff_put_h264_qpel8_mc10_msa(uint8_t *dst, const uint8_t *src,
> VSHF_B2_SB(src6, src6, src7, src7, mask2, mask2, vec10, vec11);
> DPADD_SB4_SH(vec8, vec9, vec10, vec11, plus20b, plus20b, plus20b, plus20b,
> res4, res5, res6, res7);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 2);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 2);
>- SLDI_B2_SB(src4, src5, src4, src5, src4, src5, 2);
>- SLDI_B2_SB(src6, src7, src6, src7, src6, src7, 2);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 2,
>+ src0, src1, src2, src3);
>+ SLDI_B4_SB(src4, src4, src5, src5, src6, src6, src7, src7, 2,
>+ src4, src5, src6, src7);
> PCKEV_D2_SB(src1, src0, src3, src2, src0, src1);
> PCKEV_D2_SB(src5, src4, src7, src6, src4, src5);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
>@@ -966,10 +966,10 @@ void ff_put_h264_qpel8_mc30_msa(uint8_t *dst, const uint8_t *src,
> VSHF_B2_SB(src6, src6, src7, src7, mask2, mask2, vec10, vec11);
> DPADD_SB4_SH(vec8, vec9, vec10, vec11, plus20b, plus20b, plus20b, plus20b,
> res4, res5, res6, res7);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 3);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 3);
>- SLDI_B2_SB(src4, src5, src4, src5, src4, src5, 3);
>- SLDI_B2_SB(src6, src7, src6, src7, src6, src7, 3);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 3,
>+ src0, src1, src2, src3);
>+ SLDI_B4_SB(src4, src4, src5, src5, src6, src6, src7, src7, 3,
>+ src4, src5, src6, src7);
> PCKEV_D2_SB(src1, src0, src3, src2, src0, src1);
> PCKEV_D2_SB(src5, src4, src7, src6, src4, src5);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
>@@ -1007,8 +1007,8 @@ void ff_put_h264_qpel4_mc10_msa(uint8_t *dst, const uint8_t *src,
> SRARI_H2_SH(res0, res1, 5);
> SAT_SH2_SH(res0, res1, 7);
> res = __msa_pckev_b((v16i8) res1, (v16i8) res0);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 2);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 2);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 2,
>+ src0, src1, src2, src3);
> src0 = (v16i8) __msa_insve_w((v4i32) src0, 1, (v4i32) src1);
> src1 = (v16i8) __msa_insve_w((v4i32) src2, 1, (v4i32) src3);
> src0 = (v16i8) __msa_insve_d((v2i64) src0, 1, (v2i64) src1);
>@@ -1038,8 +1038,8 @@ void ff_put_h264_qpel4_mc30_msa(uint8_t *dst, const uint8_t *src,
> SRARI_H2_SH(res0, res1, 5);
> SAT_SH2_SH(res0, res1, 7);
> res = __msa_pckev_b((v16i8) res1, (v16i8) res0);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 3);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 3);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 3,
>+ src0, src1, src2, src3);
> src0 = (v16i8) __msa_insve_w((v4i32) src0, 1, (v4i32) src1);
> src1 = (v16i8) __msa_insve_w((v4i32) src2, 1, (v4i32) src3);
> src0 = (v16i8) __msa_insve_d((v2i64) src0, 1, (v2i64) src1);
>@@ -3194,8 +3194,8 @@ void ff_avg_h264_qpel16_mc10_msa(uint8_t *dst, const uint8_t *src,
> minus5b, res4, res5, res6, res7);
> DPADD_SB4_SH(vec2, vec5, vec8, vec11, plus20b, plus20b, plus20b,
> plus20b, res4, res5, res6, res7);
>- SLDI_B2_SB(src1, src3, src0, src2, src0, src2, 2);
>- SLDI_B2_SB(src5, src7, src4, src6, src4, src6, 2);
>+ SLDI_B4_SB(src1, src0, src3, src2, src5, src4, src7, src6, 2,
>+ src0, src2, src4, src6);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
> SRARI_H4_SH(res4, res5, res6, res7, 5);
> SAT_SH4_SH(res0, res1, res2, res3, 7);
>@@ -3266,8 +3266,8 @@ void ff_avg_h264_qpel16_mc30_msa(uint8_t *dst, const uint8_t *src,
> minus5b, res4, res5, res6, res7);
> DPADD_SB4_SH(vec2, vec5, vec8, vec11, plus20b, plus20b, plus20b,
> plus20b, res4, res5, res6, res7);
>- SLDI_B2_SB(src1, src3, src0, src2, src0, src2, 3);
>- SLDI_B2_SB(src5, src7, src4, src6, src4, src6, 3);
>+ SLDI_B4_SB(src1, src0, src3, src2, src5, src4, src7, src6, 3,
>+ src0, src2, src4, src6);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
> SRARI_H4_SH(res4, res5, res6, res7, 5);
> SAT_SH4_SH(res0, res1, res2, res3, 7);
>@@ -3323,10 +3323,10 @@ void ff_avg_h264_qpel8_mc10_msa(uint8_t *dst, const uint8_t *src,
> VSHF_B2_SB(src6, src6, src7, src7, mask2, mask2, vec10, vec11);
> DPADD_SB4_SH(vec8, vec9, vec10, vec11, plus20b, plus20b, plus20b, plus20b,
> res4, res5, res6, res7);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 2);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 2);
>- SLDI_B2_SB(src4, src5, src4, src5, src4, src5, 2);
>- SLDI_B2_SB(src6, src7, src6, src7, src6, src7, 2);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 2,
>+ src0, src1, src2, src3);
>+ SLDI_B4_SB(src4, src4, src5, src5, src6, src6, src7, src7, 2,
>+ src4, src5, src6, src7);
> PCKEV_D2_SB(src1, src0, src3, src2, src0, src1);
> PCKEV_D2_SB(src5, src4, src7, src6, src4, src5);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
>@@ -3388,10 +3388,10 @@ void ff_avg_h264_qpel8_mc30_msa(uint8_t *dst, const uint8_t *src,
> VSHF_B2_SB(src6, src6, src7, src7, mask2, mask2, vec10, vec11);
> DPADD_SB4_SH(vec8, vec9, vec10, vec11, plus20b, plus20b, plus20b, plus20b,
> res4, res5, res6, res7);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 3);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 3);
>- SLDI_B2_SB(src4, src5, src4, src5, src4, src5, 3);
>- SLDI_B2_SB(src6, src7, src6, src7, src6, src7, 3);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 3,
>+ src0, src1, src2, src3);
>+ SLDI_B4_SB(src4, src4, src5, src5, src6, src6, src7, src7, 3,
>+ src4, src5, src6, src7);
> PCKEV_D2_SB(src1, src0, src3, src2, src0, src1);
> PCKEV_D2_SB(src5, src4, src7, src6, src4, src5);
> SRARI_H4_SH(res0, res1, res2, res3, 5);
>@@ -3439,8 +3439,8 @@ void ff_avg_h264_qpel4_mc10_msa(uint8_t *dst, const uint8_t *src,
> SRARI_H2_SH(out0, out1, 5);
> SAT_SH2_SH(out0, out1, 7);
> res = __msa_pckev_b((v16i8) out1, (v16i8) out0);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 2);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 2);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 2,
>+ src0, src1, src2, src3);
> src0 = (v16i8) __msa_insve_w((v4i32) src0, 1, (v4i32) src1);
> src1 = (v16i8) __msa_insve_w((v4i32) src2, 1, (v4i32) src3);
> src0 = (v16i8) __msa_insve_d((v2i64) src0, 1, (v2i64) src1);
>@@ -3475,8 +3475,8 @@ void ff_avg_h264_qpel4_mc30_msa(uint8_t *dst, const uint8_t *src,
> SRARI_H2_SH(out0, out1, 5);
> SAT_SH2_SH(out0, out1, 7);
> res = __msa_pckev_b((v16i8) out1, (v16i8) out0);
>- SLDI_B2_SB(src0, src1, src0, src1, src0, src1, 3);
>- SLDI_B2_SB(src2, src3, src2, src3, src2, src3, 3);
>+ SLDI_B4_SB(src0, src0, src1, src1, src2, src2, src3, src3, 3,
>+ src0, src1, src2, src3);
> src0 = (v16i8) __msa_insve_w((v4i32) src0, 1, (v4i32) src1);
> src1 = (v16i8) __msa_insve_w((v4i32) src2, 1, (v4i32) src3);
> src0 = (v16i8) __msa_insve_d((v2i64) src0, 1, (v2i64) src1);
>diff --git a/libavcodec/mips/hevc_lpf_sao_msa.c b/libavcodec/mips/hevc_lpf_sao_msa.c
>index ac21806..791ddb3 100644
>--- a/libavcodec/mips/hevc_lpf_sao_msa.c
>+++ b/libavcodec/mips/hevc_lpf_sao_msa.c
>@@ -1357,6 +1357,7 @@ static void hevc_sao_edge_filter_0degree_8width_msa(uint8_t *dst,
> v16u8 cmp_minus10, diff_minus10, diff_minus11;
> v16u8 src0, src1, dst0, src_minus10, src_minus11, src_plus10, src_plus11;
> v16i8 offset, sao_offset = LD_SB(sao_offset_val);
>+ v16i8 zeros = { 0 };
>
> sao_offset = __msa_pckev_b(sao_offset, sao_offset);
> src -= 1;
>@@ -1367,8 +1368,8 @@ static void hevc_sao_edge_filter_0degree_8width_msa(uint8_t *dst,
> for (height -= 2; height; height -= 2) {
> src += (src_stride << 1);
>
>- SLDI_B2_0_UB(src_minus10, src_minus11, src0, src1, 1);
>- SLDI_B2_0_UB(src_minus10, src_minus11, src_plus10, src_plus11, 2);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 1, src0, src1);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 2, src_plus10, src_plus11);
>
> PCKEV_D2_UB(src_minus11, src_minus10, src_plus11, src_plus10,
> src_minus10, src_plus10);
>@@ -1404,8 +1405,8 @@ static void hevc_sao_edge_filter_0degree_8width_msa(uint8_t *dst,
> dst += dst_stride;
> }
>
>- SLDI_B2_0_UB(src_minus10, src_minus11, src0, src1, 1);
>- SLDI_B2_0_UB(src_minus10, src_minus11, src_plus10, src_plus11, 2);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 1, src0, src1);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 2, src_plus10, src_plus11);
>
> PCKEV_D2_UB(src_minus11, src_minus10, src_plus11, src_plus10, src_minus10,
> src_plus10);
>@@ -1473,14 +1474,12 @@ static void hevc_sao_edge_filter_0degree_16multiple_msa(uint8_t *dst,
> dst_ptr = dst + v_cnt;
> LD_UB4(src_minus1, src_stride, src10, src11, src12, src13);
>
>- SLDI_B2_SB(src10, src11, src_minus10, src_minus11, src_zero0,
>- src_zero1, 1);
>- SLDI_B2_SB(src12, src13, src_minus12, src_minus13, src_zero2,
>- src_zero3, 1);
>- SLDI_B2_SB(src10, src11, src_minus10, src_minus11, src_plus10,
>- src_plus11, 2);
>- SLDI_B2_SB(src12, src13, src_minus12, src_minus13, src_plus12,
>- src_plus13, 2);
>+ SLDI_B4_SB(src10, src_minus10, src11, src_minus11,
>+ src12, src_minus12, src13, src_minus13, 1,
>+ src_zero0, src_zero1, src_zero2, src_zero3);
>+ SLDI_B4_SB(src10, src_minus10, src11, src_minus11,
>+ src12, src_minus12, src13, src_minus13, 2,
>+ src_plus10, src_plus11, src_plus12, src_plus13);
>
> cmp_minus10 = ((v16u8) src_zero0 == src_minus10);
> cmp_plus10 = ((v16u8) src_zero0 == (v16u8) src_plus10);
>@@ -1880,6 +1879,7 @@ static void hevc_sao_edge_filter_45degree_4width_msa(uint8_t *dst,
> v16u8 src_minus11, src10, src11;
> v16i8 src_plus0, src_zero0, src_plus1, src_zero1, dst0;
> v8i16 offset_mask0, offset_mask1;
>+ v16i8 zeros = { 0 };
>
> sao_offset = __msa_pckev_b(sao_offset, sao_offset);
>
>@@ -1892,8 +1892,8 @@ static void hevc_sao_edge_filter_45degree_4width_msa(uint8_t *dst,
> for (height -= 2; height; height -= 2) {
> src_orig += (src_stride << 1);
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_SB(src10, src11, src_plus0, src_plus1, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_SB(zeros, src10, zeros, src11, 2, src_plus0, src_plus1);
>
> ILVR_B2_UB(src_plus0, src_minus10, src_plus1, src_minus11, src_minus10,
> src_minus11);
>@@ -1938,8 +1938,8 @@ static void hevc_sao_edge_filter_45degree_4width_msa(uint8_t *dst,
> dst += dst_stride;
> }
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_SB(src10, src11, src_plus0, src_plus1, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_SB(zeros, src10, zeros, src11, 2, src_plus0, src_plus1);
>
> ILVR_B2_UB(src_plus0, src_minus10, src_plus1, src_minus11, src_minus10,
> src_minus11);
>@@ -1992,6 +1992,7 @@ static void hevc_sao_edge_filter_45degree_8width_msa(uint8_t *dst,
> v16u8 src_minus10, src10, src_minus11, src11;
> v16i8 src_zero0, src_plus10, src_zero1, src_plus11, dst0;
> v8i16 offset_mask0, offset_mask1;
>+ v16i8 zeros = { 0 };
>
> sao_offset = __msa_pckev_b(sao_offset, sao_offset);
> src_orig = src - 1;
>@@ -2003,8 +2004,8 @@ static void hevc_sao_edge_filter_45degree_8width_msa(uint8_t *dst,
> for (height -= 2; height; height -= 2) {
> src_orig += (src_stride << 1);
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_SB(src10, src11, src_plus10, src_plus11, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_SB(zeros, src10, zeros, src11, 2, src_plus10, src_plus11);
>
> ILVR_B2_UB(src_plus10, src_minus10, src_plus11, src_minus11,
> src_minus10, src_minus11);
>@@ -2048,8 +2049,8 @@ static void hevc_sao_edge_filter_45degree_8width_msa(uint8_t *dst,
> dst += dst_stride;
> }
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_SB(src10, src11, src_plus10, src_plus11, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_SB(zeros, src10, zeros, src11, 2, src_plus10, src_plus11);
> ILVR_B2_UB(src_plus10, src_minus10, src_plus11, src_minus11, src_minus10,
> src_minus11);
> ILVR_B2_SB(src_zero0, src_zero0, src_zero1, src_zero1, src_zero0,
>@@ -2130,12 +2131,11 @@ static void hevc_sao_edge_filter_45degree_16multiple_msa(uint8_t *dst,
> src_plus13 = LD_UB(src + 1 + v_cnt + (src_stride << 2));
> src_orig += 16;
>
>- SLDI_B2_SB(src10, src11, src_minus11, src_minus12, src_zero0,
>- src_zero1, 1);
>- SLDI_B2_SB(src12, src13, src_minus13, src_minus14, src_zero2,
>- src_zero3, 1);
>- SLDI_B2_SB(src11, src12, src_minus12, src_minus13, src_plus10,
>- src_plus11, 2);
>+ SLDI_B4_SB(src10, src_minus11, src11, src_minus12,
>+ src12, src_minus13, src13, src_minus14, 1,
>+ src_zero0, src_zero1, src_zero2, src_zero3);
>+ SLDI_B2_SB(src11, src_minus12, src12, src_minus13, 2, src_plus10,
>+ src_plus11);
>
> src_plus12 = __msa_sldi_b((v16i8) src13, (v16i8) src_minus14, 2);
>
>@@ -2228,6 +2228,7 @@ static void hevc_sao_edge_filter_135degree_4width_msa(uint8_t *dst,
> v16u8 cmp_minus10, diff_minus10, cmp_minus11, diff_minus11;
> v16u8 src_minus10, src10, src_minus11, src11;
> v8i16 offset_mask0, offset_mask1;
>+ v16i8 zeros = { 0 };
>
> sao_offset = __msa_pckev_b(sao_offset, sao_offset);
> src_orig = src - 1;
>@@ -2239,8 +2240,8 @@ static void hevc_sao_edge_filter_135degree_4width_msa(uint8_t *dst,
> for (height -= 2; height; height -= 2) {
> src_orig += (src_stride << 1);
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_UB(src_minus10, src_minus11, src_minus10, src_minus11, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 2, src_minus10, src_minus11);
>
> ILVR_B2_UB(src10, src_minus10, src11, src_minus11, src_minus10,
> src_minus11);
>@@ -2286,8 +2287,8 @@ static void hevc_sao_edge_filter_135degree_4width_msa(uint8_t *dst,
> dst += dst_stride;
> }
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_UB(src_minus10, src_minus11, src_minus10, src_minus11, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 2, src_minus10, src_minus11);
>
> ILVR_B2_UB(src10, src_minus10, src11, src_minus11, src_minus10,
> src_minus11);
>@@ -2342,6 +2343,7 @@ static void hevc_sao_edge_filter_135degree_8width_msa(uint8_t *dst,
> v16u8 src_minus10, src10, src_minus11, src11;
> v16i8 src_zero0, src_zero1, dst0;
> v8i16 offset_mask0, offset_mask1;
>+ v16i8 zeros = { 0 };
>
> sao_offset = __msa_pckev_b(sao_offset, sao_offset);
> src_orig = src - 1;
>@@ -2353,8 +2355,8 @@ static void hevc_sao_edge_filter_135degree_8width_msa(uint8_t *dst,
> for (height -= 2; height; height -= 2) {
> src_orig += (src_stride << 1);
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_UB(src_minus10, src_minus11, src_minus10, src_minus11, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 2, src_minus10, src_minus11);
> ILVR_B2_UB(src10, src_minus10, src11, src_minus11, src_minus10,
> src_minus11);
> ILVR_B2_SB(src_zero0, src_zero0, src_zero1, src_zero1, src_zero0,
>@@ -2398,8 +2400,8 @@ static void hevc_sao_edge_filter_135degree_8width_msa(uint8_t *dst,
> dst += dst_stride;
> }
>
>- SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1);
>- SLDI_B2_0_UB(src_minus10, src_minus11, src_minus10, src_minus11, 2);
>+ SLDI_B2_SB(zeros, src_minus11, zeros, src10, 1, src_zero0, src_zero1);
>+ SLDI_B2_UB(zeros, src_minus10, zeros, src_minus11, 2, src_minus10, src_minus11);
> ILVR_B2_UB(src10, src_minus10, src11, src_minus11, src_minus10,
> src_minus11);
> ILVR_B2_SB(src_zero0, src_zero0, src_zero1, src_zero1, src_zero0,
>diff --git a/libavcodec/mips/hevcpred_msa.c b/libavcodec/mips/hevcpred_msa.c
>index b8df089..909b62d 100644
>--- a/libavcodec/mips/hevcpred_msa.c
>+++ b/libavcodec/mips/hevcpred_msa.c
>@@ -998,7 +998,8 @@ static void hevc_intra_pred_angular_upper_4width_msa(const uint8_t *src_top,
> ILVR_D2_SH(fact3, fact1, fact7, fact5, fact1, fact3);
> ILVR_B4_SH(zero, top0, zero, top1, zero, top2, zero, top3,
> diff0, diff2, diff4, diff6);
>- SLDI_B4_0_SH(diff0, diff2, diff4, diff6, diff1, diff3, diff5, diff7, 2);
>+ SLDI_B4_SH(zero, diff0, zero, diff2, zero, diff4, zero, diff6, 2,
>+ diff1, diff3, diff5, diff7);
> ILVR_D2_SH(diff2, diff0, diff6, diff4, diff0, diff2);
> ILVR_D2_SH(diff3, diff1, diff7, diff5, diff1, diff3);
> MUL2(diff1, fact0, diff3, fact2, diff1, diff3);
>@@ -1093,8 +1094,8 @@ static void hevc_intra_pred_angular_upper_8width_msa(const uint8_t
>*src_top,
> UNPCK_UB_SH(top2, diff4, diff5);
> UNPCK_UB_SH(top3, diff6, diff7);
>
>- SLDI_B2_SH(diff1, diff3, diff0, diff2, diff1, diff3, 2);
>- SLDI_B2_SH(diff5, diff7, diff4, diff6, diff5, diff7, 2);
>+ SLDI_B4_SH(diff1, diff0, diff3, diff2, diff5, diff4, diff7, diff6, 2,
>+ diff1, diff3, diff5, diff7);
> MUL4(diff1, fact0, diff3, fact2, diff5, fact4, diff7, fact6,
> diff1, diff3, diff5, diff7);
>
>@@ -1186,8 +1187,8 @@ static void hevc_intra_pred_angular_upper_16width_msa(const uint8_t
>*src_top,
> fact6 = __msa_fill_h(fact_val3);
> fact7 = __msa_fill_h(32 - fact_val3);
>
>- SLDI_B2_UB(top1, top3, top0, top2, top1, top3, 1);
>- SLDI_B2_UB(top5, top7, top4, top6, top5, top7, 1);
>+ SLDI_B4_UB(top1, top0, top3, top2, top5, top4, top7, top6, 1,
>+ top1, top3, top5, top7);
> UNPCK_UB_SH(top0, diff0, diff1);
> UNPCK_UB_SH(top1, diff2, diff3);
> UNPCK_UB_SH(top2, diff4, diff5);
>@@ -1297,8 +1298,8 @@ static void hevc_intra_pred_angular_upper_32width_msa(const uint8_t
>*src_top,
> top2 = top1;
> top6 = top5;
>
>- SLDI_B2_UB(top1, top3, top0, top2, top1, top3, 1);
>- SLDI_B2_UB(top5, top7, top4, top6, top5, top7, 1);
>+ SLDI_B4_UB(top1, top0, top3, top2, top5, top4, top7, top6, 1,
>+ top1, top3, top5, top7);
> UNPCK_UB_SH(top0, diff0, diff1);
> UNPCK_UB_SH(top1, diff2, diff3);
> UNPCK_UB_SH(top2, diff4, diff5);
>@@ -1407,7 +1408,8 @@ static void hevc_intra_pred_angular_lower_4width_msa(const uint8_t
>*src_top,
> ILVR_D2_SH(fact3, fact1, fact7, fact5, fact1, fact3);
> ILVR_B4_SH(zero, top0, zero, top1, zero, top2, zero, top3,
> diff0, diff2, diff4, diff6);
>- SLDI_B4_0_SH(diff0, diff2, diff4, diff6, diff1, diff3, diff5, diff7, 2);
>+ SLDI_B4_SH(zero, diff0, zero, diff2, zero, diff4, zero, diff6, 2,
>+ diff1, diff3, diff5, diff7);
> ILVR_D2_SH(diff2, diff0, diff6, diff4, diff0, diff2);
> ILVR_D2_SH(diff3, diff1, diff7, diff5, diff1, diff3);
> MUL2(diff1, fact0, diff3, fact2, diff1, diff3);
>@@ -1511,8 +1513,8 @@ static void hevc_intra_pred_angular_lower_8width_msa(const uint8_t
>*src_top,
> UNPCK_UB_SH(top1, diff2, diff3);
> UNPCK_UB_SH(top2, diff4, diff5);
> UNPCK_UB_SH(top3, diff6, diff7);
>- SLDI_B2_SH(diff1, diff3, diff0, diff2, diff1, diff3, 2);
>- SLDI_B2_SH(diff5, diff7, diff4, diff6, diff5, diff7, 2);
>+ SLDI_B4_SH(diff1, diff0, diff3, diff2, diff5, diff4, diff7, diff6, 2,
>+ diff1, diff3, diff5, diff7);
> MUL4(diff1, fact0, diff3, fact2, diff5, fact4, diff7, fact6,
> diff1, diff3, diff5, diff7);
>
>@@ -1606,8 +1608,8 @@ static void hevc_intra_pred_angular_lower_16width_msa(const uint8_t
>*src_top,
> fact6 = __msa_fill_h(fact_val3);
> fact7 = __msa_fill_h(32 - fact_val3);
>
>- SLDI_B2_SB(top1, top3, top0, top2, top1, top3, 1);
>- SLDI_B2_SB(top5, top7, top4, top6, top5, top7, 1);
>+ SLDI_B4_SB(top1, top0, top3, top2, top5, top4, top7, top6, 1,
>+ top1, top3, top5, top7);
>
> UNPCK_UB_SH(top0, diff0, diff1);
> UNPCK_UB_SH(top1, diff2, diff3);
>@@ -1713,8 +1715,8 @@ static void hevc_intra_pred_angular_lower_32width_msa(const uint8_t
>*src_top,
> top2 = top1;
> top6 = top5;
>
>- SLDI_B2_SB(top1, top3, top0, top2, top1, top3, 1);
>- SLDI_B2_SB(top5, top7, top4, top6, top5, top7, 1);
>+ SLDI_B4_SB(top1, top0, top3, top2, top5, top4, top7, top6, 1,
>+ top1, top3, top5, top7);
>
> UNPCK_UB_SH(top0, diff0, diff1);
> UNPCK_UB_SH(top1, diff2, diff3);
>diff --git a/libavcodec/mips/hpeldsp_msa.c b/libavcodec/mips/hpeldsp_msa.c
>index ad92f8f..2bbe477 100644
>--- a/libavcodec/mips/hpeldsp_msa.c
>+++ b/libavcodec/mips/hpeldsp_msa.c
>@@ -59,12 +59,13 @@ static void common_hz_bil_4w_msa(const uint8_t *src, int32_t src_stride,
> uint8_t loop_cnt;
> uint32_t out0, out1;
> v16u8 src0, src1, src0_sld1, src1_sld1, res0, res1;
>+ v16i8 zeros = { 0 };
>
> for (loop_cnt = (height >> 1); loop_cnt--;) {
> LD_UB2(src, src_stride, src0, src1);
> src += (2 * src_stride);
>
>- SLDI_B2_0_UB(src0, src1, src0_sld1, src1_sld1, 1);
>+ SLDI_B2_UB(zeros, src0, zeros, src1, 1, src0_sld1, src1_sld1);
> AVER_UB2_UB(src0_sld1, src0, src1_sld1, src1, res0, res1);
>
> out0 = __msa_copy_u_w((v4i32) res0, 0);
>@@ -82,13 +83,14 @@ static void common_hz_bil_8w_msa(const uint8_t *src, int32_t src_stride,
> {
> uint8_t loop_cnt;
> v16i8 src0, src1, src2, src3, src0_sld1, src1_sld1, src2_sld1, src3_sld1;
>+ v16i8 zeros = { 0 };
>
> for (loop_cnt = (height >> 2); loop_cnt--;) {
> LD_SB4(src, src_stride, src0, src1, src2, src3);
> src += (4 * src_stride);
>
>- SLDI_B4_0_SB(src0, src1, src2, src3,
>- src0_sld1, src1_sld1, src2_sld1, src3_sld1, 1);
>+ SLDI_B4_SB(zeros, src0, zeros, src1, zeros, src2, zeros, src3, 1,
>+ src0_sld1, src1_sld1, src2_sld1, src3_sld1);
> AVER_ST8x4_UB(src0, src0_sld1, src1, src1_sld1,
> src2, src2_sld1, src3, src3_sld1, dst, dst_stride);
> dst += (4 * dst_stride);
>@@ -125,14 +127,15 @@ static void common_hz_bil_no_rnd_8x8_msa(const uint8_t *src, int32_t
>src_stride,
> v16i8 src0, src1, src2, src3, src4, src5, src6, src7;
> v16i8 src0_sld1, src1_sld1, src2_sld1, src3_sld1;
> v16i8 src4_sld1, src5_sld1, src6_sld1, src7_sld1;
>+ v16i8 zeros = { 0 };
>
> LD_SB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
> src += (8 * src_stride);
>
>- SLDI_B4_0_SB(src0, src1, src2, src3,
>- src0_sld1, src1_sld1, src2_sld1, src3_sld1, 1);
>- SLDI_B4_0_SB(src4, src5, src6, src7,
>- src4_sld1, src5_sld1, src6_sld1, src7_sld1, 1);
>+ SLDI_B4_SB(zeros, src0, zeros, src1, zeros, src2, zeros, src3, 1,
>+ src0_sld1, src1_sld1, src2_sld1, src3_sld1);
>+ SLDI_B4_SB(zeros, src4, zeros, src5, zeros, src6, zeros, src7, 1,
>+ src4_sld1, src5_sld1, src6_sld1, src7_sld1);
>
> AVE_ST8x4_UB(src0, src0_sld1, src1, src1_sld1,
> src2, src2_sld1, src3, src3_sld1, dst, dst_stride);
>@@ -145,10 +148,11 @@ static void common_hz_bil_no_rnd_4x8_msa(const uint8_t *src, int32_t
>src_stride,
> uint8_t *dst, int32_t dst_stride)
> {
> v16i8 src0, src1, src2, src3, src0_sld1, src1_sld1, src2_sld1, src3_sld1;
>+ v16i8 zeros = { 0 };
>
> LD_SB4(src, src_stride, src0, src1, src2, src3);
>- SLDI_B4_0_SB(src0, src1, src2, src3,
>- src0_sld1, src1_sld1, src2_sld1, src3_sld1, 1);
>+ SLDI_B4_SB(zeros, src0, zeros, src1, zeros, src2, zeros, src3, 1,
>+ src0_sld1, src1_sld1, src2_sld1, src3_sld1);
> AVE_ST8x4_UB(src0, src0_sld1, src1, src1_sld1,
> src2, src2_sld1, src3, src3_sld1, dst, dst_stride);
> }
>@@ -216,12 +220,13 @@ static void common_hz_bil_and_aver_dst_4w_msa(const uint8_t *src,
> v16u8 src0, src1, src0_sld1, src1_sld1, res0, res1;
> v16u8 tmp0 = { 0 };
> v16u8 tmp1 = { 0 };
>+ v16i8 zeros = { 0 };
>
> for (loop_cnt = (height >> 1); loop_cnt--;) {
> LD_UB2(src, src_stride, src0, src1);
> src += (2 * src_stride);
>
>- SLDI_B2_0_UB(src0, src1, src0_sld1, src1_sld1, 1);
>+ SLDI_B2_UB(zeros, src0, zeros, src1, 1, src0_sld1, src1_sld1);
>
> dst0 = LW(dst);
> dst1 = LW(dst + dst_stride);
>@@ -247,13 +252,14 @@ static void common_hz_bil_and_aver_dst_8w_msa(const uint8_t *src,
> {
> uint8_t loop_cnt;
> v16i8 src0, src1, src2, src3, src0_sld1, src1_sld1, src2_sld1, src3_sld1;
>+ v16i8 zeros = { 0 };
>
> for (loop_cnt = (height >> 2); loop_cnt--;) {
> LD_SB4(src, src_stride, src0, src1, src2, src3);
> src += (4 * src_stride);
>
>- SLDI_B4_0_SB(src0, src1, src2, src3,
>- src0_sld1, src1_sld1, src2_sld1, src3_sld1, 1);
>+ SLDI_B4_SB(zeros, src0, zeros, src1, zeros, src2, zeros, src3, 1,
>+ src0_sld1, src1_sld1, src2_sld1, src3_sld1);
>
> AVER_DST_ST8x4_UB(src0, src0_sld1, src1, src1_sld1, src2, src2_sld1,
> src3, src3_sld1, dst, dst_stride);
>@@ -529,6 +535,7 @@ static void common_hv_bil_4w_msa(const uint8_t *src, int32_t src_stride,
> v16i8 src0, src1, src2, src0_sld1, src1_sld1, src2_sld1;
> v16u8 src0_r, src1_r, src2_r, res;
> v8u16 add0, add1, add2, sum0, sum1;
>+ v16i8 zeros = { 0 };
>
> src0 = LD_SB(src);
> src += src_stride;
>@@ -537,7 +544,8 @@ static void common_hv_bil_4w_msa(const uint8_t *src, int32_t src_stride,
> LD_SB2(src, src_stride, src1, src2);
> src += (2 * src_stride);
>
>- SLDI_B3_0_SB(src0, src1, src2, src0_sld1, src1_sld1, src2_sld1, 1);
>+ SLDI_B3_SB(zeros, src0, zeros, src1, zeros, src2, 1, src0_sld1,
>+ src1_sld1, src2_sld1);
> ILVR_B3_UB(src0_sld1, src0, src1_sld1, src1, src2_sld1, src2,
> src0_r, src1_r, src2_r);
> HADD_UB3_UH(src0_r, src1_r, src2_r, add0, add1, add2);
>@@ -565,6 +573,7 @@ static void common_hv_bil_8w_msa(const uint8_t *src, int32_t src_stride,
> v16u8 src0_r, src1_r, src2_r, src3_r, src4_r;
> v8u16 add0, add1, add2, add3, add4;
> v8u16 sum0, sum1, sum2, sum3;
>+ v16i8 zeros = { 0 };
>
> src0 = LD_SB(src);
> src += src_stride;
>@@ -573,8 +582,9 @@ static void common_hv_bil_8w_msa(const uint8_t *src, int32_t src_stride,
> LD_SB4(src, src_stride, src1, src2, src3, src4);
> src += (4 * src_stride);
>
>- SLDI_B3_0_SB(src0, src1, src2, src0_sld1, src1_sld1, src2_sld1, 1);
>- SLDI_B2_0_SB(src3, src4, src3_sld1, src4_sld1, 1);
>+ SLDI_B3_SB(zeros, src0, zeros, src1, zeros, src2, 1, src0_sld1,
>+ src1_sld1, src2_sld1);
>+ SLDI_B2_SB(zeros, src3, zeros, src4, 1, src3_sld1, src4_sld1);
> ILVR_B3_UB(src0_sld1, src0, src1_sld1, src1, src2_sld1, src2, src0_r,
> src1_r, src2_r);
> ILVR_B2_UB(src3_sld1, src3, src4_sld1, src4, src3_r, src4_r);
>@@ -659,15 +669,17 @@ static void common_hv_bil_no_rnd_8x8_msa(const uint8_t *src, int32_t
>src_stride,
> v8u16 add0, add1, add2, add3, add4, add5, add6, add7, add8;
> v8u16 sum0, sum1, sum2, sum3, sum4, sum5, sum6, sum7;
> v16i8 out0, out1;
>+ v16i8 zeros = { 0 };
>
> LD_UB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
> src += (8 * src_stride);
> src8 = LD_UB(src);
>
>- SLDI_B4_0_UB(src0, src1, src2, src3, src0_sld1, src1_sld1, src2_sld1,
>- src3_sld1, 1);
>- SLDI_B3_0_UB(src4, src5, src6, src4_sld1, src5_sld1, src6_sld1, 1);
>- SLDI_B2_0_UB(src7, src8, src7_sld1, src8_sld1, 1);
>+ SLDI_B4_UB(zeros, src0, zeros, src1, zeros, src2, zeros, src3, 1,
>+ src0_sld1, src1_sld1, src2_sld1, src3_sld1);
>+ SLDI_B3_UB(zeros, src4, zeros, src5, zeros, src6, 1, src4_sld1,
>+ src5_sld1, src6_sld1);
>+ SLDI_B2_UB(zeros, src7, zeros, src8, 1, src7_sld1, src8_sld1);
> ILVR_B4_UH(src0_sld1, src0, src1_sld1, src1, src2_sld1, src2, src3_sld1,
> src3, src0_r, src1_r, src2_r, src3_r);
> ILVR_B3_UH(src4_sld1, src4, src5_sld1, src5, src6_sld1, src6, src4_r,
>@@ -703,13 +715,15 @@ static void common_hv_bil_no_rnd_4x8_msa(const uint8_t *src, int32_t
>src_stride,
> v8u16 add0, add1, add2, add3, add4;
> v8u16 sum0, sum1, sum2, sum3;
> v16i8 out0, out1;
>+ v16i8 zeros = { 0 };
>
> LD_SB4(src, src_stride, src0, src1, src2, src3);
> src += (4 * src_stride);
> src4 = LD_SB(src);
>
>- SLDI_B3_0_SB(src0, src1, src2, src0_sld1, src1_sld1, src2_sld1, 1);
>- SLDI_B2_0_SB(src3, src4, src3_sld1, src4_sld1, 1);
>+ SLDI_B3_SB(zeros, src0, zeros, src1, zeros, src2, 1, src0_sld1,
>+ src1_sld1, src2_sld1);
>+ SLDI_B2_SB(zeros, src3, zeros, src4, 1, src3_sld1, src4_sld1);
> ILVR_B3_UH(src0_sld1, src0, src1_sld1, src1, src2_sld1, src2, src0_r,
> src1_r, src2_r);
> ILVR_B2_UH(src3_sld1, src3, src4_sld1, src4, src3_r, src4_r);
>@@ -918,6 +932,7 @@ static void common_hv_bil_and_aver_dst_4w_msa(const uint8_t *src,
> v16u8 src0_r, src1_r, src2_r;
> v8u16 add0, add1, add2, sum0, sum1;
> v16u8 dst0, dst1, res0, res1;
>+ v16i8 zeros = { 0 };
>
> src0 = LD_SB(src);
> src += src_stride;
>@@ -927,7 +942,8 @@ static void common_hv_bil_and_aver_dst_4w_msa(const uint8_t *src,
> src += (2 * src_stride);
>
> LD_UB2(dst, dst_stride, dst0, dst1);
>- SLDI_B3_0_SB(src0, src1, src2, src0_sld1, src1_sld1, src2_sld1, 1);
>+ SLDI_B3_SB(zeros, src0, zeros, src1, zeros, src2, 1, src0_sld1,
>+ src1_sld1, src2_sld1);
> ILVR_B3_UB(src0_sld1, src0, src1_sld1, src1, src2_sld1, src2, src0_r,
> src1_r, src2_r);
> HADD_UB3_UH(src0_r, src1_r, src2_r, add0, add1, add2);
>@@ -959,6 +975,7 @@ static void common_hv_bil_and_aver_dst_8w_msa(const uint8_t *src,
> v16u8 src0_r, src1_r, src2_r, src3_r, src4_r;
> v8u16 add0, add1, add2, add3, add4;
> v8u16 sum0, sum1, sum2, sum3;
>+ v16i8 zeros = { 0 };
>
> src0 = LD_SB(src);
> src += src_stride;
>@@ -968,8 +985,9 @@ static void common_hv_bil_and_aver_dst_8w_msa(const uint8_t *src,
> src += (4 * src_stride);
>
> LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
>- SLDI_B3_0_SB(src0, src1, src2, src0_sld1, src1_sld1, src2_sld1, 1);
>- SLDI_B2_0_SB(src3, src4, src3_sld1, src4_sld1, 1);
>+ SLDI_B3_SB(zeros, src0, zeros, src1, zeros, src2, 1, src0_sld1,
>+ src1_sld1, src2_sld1);
>+ SLDI_B2_SB(zeros, src3, zeros, src4, 1, src3_sld1, src4_sld1);
> ILVR_B3_UB(src0_sld1, src0, src1_sld1, src1, src2_sld1, src2, src0_r,
> src1_r, src2_r);
> ILVR_B2_UB(src3_sld1, src3, src4_sld1, src4, src3_r, src4_r);
>diff --git a/libavcodec/mips/me_cmp_msa.c b/libavcodec/mips/me_cmp_msa.c
>index 0e3165c..7cb7af0 100644
>--- a/libavcodec/mips/me_cmp_msa.c
>+++ b/libavcodec/mips/me_cmp_msa.c
>@@ -87,8 +87,8 @@ static uint32_t sad_horiz_bilinear_filter_8width_msa(uint8_t *src,
>
> PCKEV_D2_UB(src1, src0, src3, src2, src0, src1);
> PCKEV_D2_UB(ref1, ref0, ref3, ref2, ref4, ref5);
>- SLDI_B2_UB(ref0, ref1, ref0, ref1, ref0, ref1, 1);
>- SLDI_B2_UB(ref2, ref3, ref2, ref3, ref2, ref3, 1);
>+ SLDI_B4_UB(ref0, ref0, ref1, ref1, ref2, ref2, ref3, ref3, 1,
>+ ref0, ref1, ref2, ref3);
> PCKEV_D2_UB(ref1, ref0, ref3, ref2, ref0, ref1);
> AVER_UB2_UB(ref4, ref0, ref5, ref1, comp0, comp1);
> sad += SAD_UB2_UH(src0, src1, comp0, comp1);
>@@ -100,8 +100,8 @@ static uint32_t sad_horiz_bilinear_filter_8width_msa(uint8_t *src,
>
> PCKEV_D2_UB(src1, src0, src3, src2, src0, src1);
> PCKEV_D2_UB(ref1, ref0, ref3, ref2, ref4, ref5);
>- SLDI_B2_UB(ref0, ref1, ref0, ref1, ref0, ref1, 1);
>- SLDI_B2_UB(ref2, ref3, ref2, ref3, ref2, ref3, 1);
>+ SLDI_B4_UB(ref0, ref0, ref1, ref1, ref2, ref2, ref3, ref3, 1,
>+ ref0, ref1, ref2, ref3);
> PCKEV_D2_UB(ref1, ref0, ref3, ref2, ref0, ref1);
> AVER_UB2_UB(ref4, ref0, ref5, ref1, comp0, comp1);
> sad += SAD_UB2_UH(src0, src1, comp0, comp1);
>diff --git a/libavcodec/mips/qpeldsp_msa.c b/libavcodec/mips/qpeldsp_msa.c
>index fba42b30..f9d1a95 100644
>--- a/libavcodec/mips/qpeldsp_msa.c
>+++ b/libavcodec/mips/qpeldsp_msa.c
>@@ -480,8 +480,8 @@ static void horiz_mc_qpel_aver_src1_8width_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3,
> mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B4_UB(inp0, inp0, inp1, inp1, inp2, inp2, inp3, inp3, 1,
>+ inp0, inp1, inp2, inp3);
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> AVER_UB2_UB(inp0, res0, inp2, res1, res0, res1);
>@@ -710,8 +710,8 @@ static void horiz_mc_qpel_no_rnd_aver_src1_8width_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B4_UB(inp0, inp0, inp1, inp1, inp2, inp2, inp3, inp3, 1,
>+ inp0, inp1, inp2, inp3);
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> res0 = __msa_ave_u_b(inp0, res0);
>@@ -948,8 +948,8 @@ static void horiz_mc_qpel_avg_dst_aver_src1_8width_msa(const uint8_t *src,
> mask0, mask1, mask2, mask3,
> const20, const6, const3);
> LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B4_UB(inp0, inp0, inp1, inp1, inp2, inp2, inp3, inp3, 1,
>+ inp0, inp1, inp2, inp3);
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> dst0 = (v16u8) __msa_insve_d((v2i64) dst0, 1, (v2i64) dst1);
>@@ -3094,7 +3094,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src10_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp0, inp1, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz0 = __msa_ave_u_b(inp0, res0);
>@@ -3104,7 +3104,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src10_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz2 = __msa_ave_u_b(inp2, res1);
>@@ -3114,7 +3114,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src10_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp0, inp1, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz4 = __msa_ave_u_b(inp0, res0);
>@@ -3134,7 +3134,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src10_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz6 = __msa_ave_u_b(inp2, res1);
>@@ -3389,7 +3389,7 @@ static void hv_mc_qpel_no_rnd_aver_h_src1_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp0, inp1, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz0 = __msa_ave_u_b(inp0, res0);
>@@ -3399,7 +3399,7 @@ static void hv_mc_qpel_no_rnd_aver_h_src1_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz2 = __msa_ave_u_b(inp2, res1);
>@@ -3409,7 +3409,7 @@ static void hv_mc_qpel_no_rnd_aver_h_src1_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp0, inp1, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz4 = __msa_ave_u_b(inp0, res0);
>@@ -3427,7 +3427,7 @@ static void hv_mc_qpel_no_rnd_aver_h_src1_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz6 = __msa_ave_u_b(inp2, res1);
>@@ -3691,7 +3691,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src11_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp0, inp1, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz0 = __msa_ave_u_b(inp0, res0);
>@@ -3701,7 +3701,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src11_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz2 = __msa_ave_u_b(inp2, res1);
>@@ -3712,7 +3712,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src11_8x8_msa(const uint8_t *src,
> mask2, mask3, const20,
> const6, const3);
>
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz4 = __msa_ave_u_b(inp0, res0);
> horiz5 = (v16u8) __msa_splati_d((v2i64) horiz4, 1);
>@@ -3731,7 +3731,7 @@ static void hv_mc_qpel_no_rnd_aver_hv_src11_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_NO_ROUND_FILTER_8BYTE(inp2, inp3, mask0, mask1,
> mask2, mask3, const20,
> const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz6 = __msa_ave_u_b(inp2, res1);
>@@ -4134,12 +4134,12 @@ static void hv_mc_qpel_aver_hv_src10_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz0 = __msa_aver_u_b(inp0, res0);
> horiz1 = (v16u8) __msa_splati_d((v2i64) horiz0, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz2 = __msa_aver_u_b(inp2, res1);
>@@ -4150,12 +4150,12 @@ static void hv_mc_qpel_aver_hv_src10_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz4 = __msa_aver_u_b(inp0, res0);
> horiz5 = (v16u8) __msa_splati_d((v2i64) horiz4, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz6 = __msa_aver_u_b(inp2, res1);
>@@ -4410,12 +4410,12 @@ static void hv_mc_qpel_aver_h_src1_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz0 = __msa_aver_u_b(inp0, res0);
> horiz1 = (v16u8) __msa_splati_d((v2i64) horiz0, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz2 = __msa_aver_u_b(inp2, res1);
>@@ -4426,12 +4426,12 @@ static void hv_mc_qpel_aver_h_src1_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz4 = __msa_aver_u_b(inp0, res0);
> horiz5 = (v16u8) __msa_splati_d((v2i64) horiz4, 1);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz6 = __msa_aver_u_b(inp2, res1);
>@@ -4690,14 +4690,14 @@ static void hv_mc_qpel_aver_hv_src11_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp0, inp1,
> mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz0 = __msa_aver_u_b(inp0, res0);
> horiz1 = (v16u8) __msa_splati_d((v2i64) horiz0, 1);
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz2 = __msa_aver_u_b(inp2, res1);
>@@ -4706,7 +4706,7 @@ static void hv_mc_qpel_aver_hv_src11_8x8_msa(const uint8_t *src,
> src += (2 * src_stride);
> res0 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp0, inp1, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_insve_d((v2i64) inp0, 1, (v2i64) inp1);
> horiz4 = __msa_aver_u_b(inp0, res0);
>@@ -4725,7 +4725,7 @@ static void hv_mc_qpel_aver_hv_src11_8x8_msa(const uint8_t *src,
>
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_insve_d((v2i64) inp2, 1, (v2i64) inp3);
> horiz6 = __msa_aver_u_b(inp2, res1);
>@@ -5020,7 +5020,7 @@ static void hv_mc_qpel_avg_dst_aver_hv_src10_8x8_msa(const uint8_t *src,
>
> LD_UB2(src, src_stride, inp2, inp3);
> src += (2 * src_stride);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_ilvr_d((v2i64) inp1, (v2i64) inp0);
> horiz0 = __msa_aver_u_b(inp0, res0);
>@@ -5029,7 +5029,7 @@ static void hv_mc_qpel_avg_dst_aver_hv_src10_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> LD_UB2(src, src_stride, inp0, inp1);
> src += (2 * src_stride);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_ilvr_d((v2i64) inp3, (v2i64) inp2);
> horiz2 = __msa_aver_u_b(inp2, res1);
>@@ -5037,7 +5037,7 @@ static void hv_mc_qpel_avg_dst_aver_hv_src10_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp0, inp1, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_ilvr_d((v2i64) inp1, (v2i64) inp0);
> horiz4 = __msa_aver_u_b(inp0, res0);
>@@ -5060,7 +5060,7 @@ static void hv_mc_qpel_avg_dst_aver_hv_src10_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_ilvr_d((v2i64) inp3, (v2i64) inp2);
> horiz6 = __msa_aver_u_b(inp2, res1);
>@@ -5347,7 +5347,7 @@ static void hv_mc_qpel_avg_dst_aver_h_src1_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> LD_UB2(src, src_stride, inp2, inp3);
> src += (2 * src_stride);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_ilvr_d((v2i64) inp1, (v2i64) inp0);
> horiz0 = __msa_aver_u_b(inp0, res0);
>@@ -5356,7 +5356,7 @@ static void hv_mc_qpel_avg_dst_aver_h_src1_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> LD_UB2(src, src_stride, inp0, inp1);
> src += (2 * src_stride);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_ilvr_d((v2i64) inp3, (v2i64) inp2);
> horiz2 = __msa_aver_u_b(inp2, res1);
>@@ -5364,7 +5364,7 @@ static void hv_mc_qpel_avg_dst_aver_h_src1_8x8_msa(const uint8_t *src,
> res0 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp0, inp1, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_ilvr_d((v2i64) inp1, (v2i64) inp0);
> horiz4 = __msa_aver_u_b(inp0, res0);
>@@ -5385,7 +5385,7 @@ static void hv_mc_qpel_avg_dst_aver_h_src1_8x8_msa(const uint8_t *src,
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_ilvr_d((v2i64) inp3, (v2i64) inp2);
> horiz6 = __msa_aver_u_b(inp2, res1);
>@@ -5684,7 +5684,7 @@ static void hv_mc_qpel_avg_dst_aver_hv_src11_8x8_msa(const uint8_t *src,
> const20, const6, const3);
> LD_UB2(src, src_stride, inp2, inp3);
> src += (2 * src_stride);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_ilvr_d((v2i64) inp1, (v2i64) inp0);
> horiz0 = __msa_aver_u_b(inp0, res0);
>@@ -5693,14 +5693,14 @@ static void hv_mc_qpel_avg_dst_aver_hv_src11_8x8_msa(const uint8_t
>*src,
> const20, const6, const3);
> LD_UB2(src, src_stride, inp0, inp1);
> src += (2 * src_stride);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_ilvr_d((v2i64) inp3, (v2i64) inp2);
> horiz2 = __msa_aver_u_b(inp2, res1);
> horiz3 = (v16u8) __msa_splati_d((v2i64) horiz2, 1);
> res0 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp0, inp1, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp0, inp1, inp0, inp1, inp0, inp1, 1);
>+ SLDI_B2_UB(inp0, inp0, inp1, inp1, 1, inp0, inp1);
>
> inp0 = (v16u8) __msa_ilvr_d((v2i64) inp1, (v2i64) inp0);
> horiz4 = __msa_aver_u_b(inp0, res0);
>@@ -5721,7 +5721,7 @@ static void hv_mc_qpel_avg_dst_aver_hv_src11_8x8_msa(const uint8_t *src,
> src += (2 * src_stride);
> res1 = APPLY_HORIZ_QPEL_FILTER_8BYTE(inp2, inp3, mask0, mask1, mask2, mask3,
> const20, const6, const3);
>- SLDI_B2_UB(inp2, inp3, inp2, inp3, inp2, inp3, 1);
>+ SLDI_B2_UB(inp2, inp2, inp3, inp3, 1, inp2, inp3);
>
> inp2 = (v16u8) __msa_ilvr_d((v2i64) inp3, (v2i64) inp2);
> horiz6 = __msa_aver_u_b(inp2, res1);
>diff --git a/libavcodec/mips/vp8_mc_msa.c b/libavcodec/mips/vp8_mc_msa.c
>index 57af6b4..a613206 100644
>--- a/libavcodec/mips/vp8_mc_msa.c
>+++ b/libavcodec/mips/vp8_mc_msa.c
>@@ -1995,8 +1995,8 @@ static void common_hv_2ht_2vt_4x8_msa(uint8_t *src, int32_t src_stride,
> hz_out4 = HORIZ_2TAP_FILT_UH(src4, src5, mask, filt_hz, 7);
> hz_out6 = HORIZ_2TAP_FILT_UH(src6, src7, mask, filt_hz, 7);
> hz_out8 = HORIZ_2TAP_FILT_UH(src8, src8, mask, filt_hz, 7);
>- SLDI_B3_UH(hz_out2, hz_out4, hz_out6, hz_out0, hz_out2, hz_out4, hz_out1,
>- hz_out3, hz_out5, 8);
>+ SLDI_B3_UH(hz_out2, hz_out0, hz_out4, hz_out2, hz_out6, hz_out4, 8, hz_out1,
>+ hz_out3, hz_out5);
> hz_out7 = (v8u16) __msa_pckod_d((v2i64) hz_out8, (v2i64) hz_out6);
>
> ILVEV_B2_UB(hz_out0, hz_out1, hz_out2, hz_out3, vec0, vec1);
>diff --git a/libavcodec/mips/vp9_idct_msa.c b/libavcodec/mips/vp9_idct_msa.c
>index 1f32770..924b83d 100644
>--- a/libavcodec/mips/vp9_idct_msa.c
>+++ b/libavcodec/mips/vp9_idct_msa.c
>@@ -249,6 +249,7 @@ static const int32_t sinpi_4_9 = 15212;
> v8i16 c0_m, c1_m, c2_m, c3_m; \
> v8i16 step0_m, step1_m; \
> v4i32 tmp0_m, tmp1_m, tmp2_m, tmp3_m; \
>+ v16i8 zeros = { 0 }; \
> \
> c0_m = VP9_SET_COSPI_PAIR(cospi_16_64, cospi_16_64); \
> c1_m = VP9_SET_COSPI_PAIR(cospi_16_64, -cospi_16_64); \
>@@ -262,7 +263,7 @@ static const int32_t sinpi_4_9 = 15212;
> SRARI_W4_SW(tmp0_m, tmp1_m, tmp2_m, tmp3_m, VP9_DCT_CONST_BITS); \
> \
> PCKEV_H2_SW(tmp1_m, tmp0_m, tmp3_m, tmp2_m, tmp0_m, tmp2_m); \
>- SLDI_B2_0_SW(tmp0_m, tmp2_m, tmp1_m, tmp3_m, 8); \
>+ SLDI_B2_SW(zeros, tmp0_m, zeros, tmp2_m, 8, tmp1_m, tmp3_m); \
> BUTTERFLY_4((v8i16) tmp0_m, (v8i16) tmp1_m, \
> (v8i16) tmp2_m, (v8i16) tmp3_m, \
> out0, out1, out2, out3); \
>diff --git a/libavcodec/mips/vp9_lpf_msa.c b/libavcodec/mips/vp9_lpf_msa.c
>index 2450c74..cbb1409 100644
>--- a/libavcodec/mips/vp9_lpf_msa.c
>+++ b/libavcodec/mips/vp9_lpf_msa.c
>@@ -1673,6 +1673,7 @@ static void vp9_transpose_16x8_to_8x16(uint8_t *input, int32_t in_pitch,
> v16u8 p7_org, p6_org, p5_org, p4_org, p3_org, p2_org, p1_org, p0_org;
> v16i8 tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
> v16u8 p7, p6, p5, p4, p3, p2, p1, p0, q0, q1, q2, q3, q4, q5, q6, q7;
>+ v16i8 zeros = { 0 };
>
> LD_UB8(input, in_pitch,
> p7_org, p6_org, p5_org, p4_org, p3_org, p2_org, p1_org, p0_org);
>@@ -1686,7 +1687,7 @@ static void vp9_transpose_16x8_to_8x16(uint8_t *input, int32_t in_pitch,
> ILVL_B2_SB(tmp1, tmp0, tmp3, tmp2, tmp5, tmp7);
> ILVR_W2_UB(tmp6, tmp4, tmp7, tmp5, q0, q4);
> ILVL_W2_UB(tmp6, tmp4, tmp7, tmp5, q2, q6);
>- SLDI_B4_0_UB(q0, q2, q4, q6, q1, q3, q5, q7, 8);
>+ SLDI_B4_UB(zeros, q0, zeros, q2, zeros, q4, zeros, q6, 8, q1, q3, q5, q7);
>
> ST_UB8(p7, p6, p5, p4, p3, p2, p1, p0, output, out_pitch);
> output += (8 * out_pitch);
>diff --git a/libavcodec/mips/vp9_mc_msa.c b/libavcodec/mips/vp9_mc_msa.c
>index 1d8a892..57ea425 100644
>--- a/libavcodec/mips/vp9_mc_msa.c
>+++ b/libavcodec/mips/vp9_mc_msa.c
>@@ -795,7 +795,7 @@ static void common_hv_8ht_8vt_4w_msa(const uint8_t *src, int32_t src_stride,
> filt_hz1, filt_hz2, filt_hz3);
> hz_out5 = HORIZ_8TAP_FILT(src5, src6, mask0, mask1, mask2, mask3, filt_hz0,
> filt_hz1, filt_hz2, filt_hz3);
>- SLDI_B2_SH(hz_out2, hz_out4, hz_out0, hz_out2, hz_out1, hz_out3, 8);
>+ SLDI_B2_SH(hz_out2, hz_out0, hz_out4, hz_out2, 8, hz_out1, hz_out3);
>
> filt = LD_SH(filter_vert);
> SPLATI_H4_SH(filt, 0, 1, 2, 3, filt_vt0, filt_vt1, filt_vt2, filt_vt3);
>@@ -1585,7 +1585,7 @@ static void common_hv_8ht_8vt_and_aver_dst_4w_msa(const uint8_t *src,
> filt_hz1, filt_hz2, filt_hz3);
> hz_out5 = HORIZ_8TAP_FILT(src5, src6, mask0, mask1, mask2, mask3, filt_hz0,
> filt_hz1, filt_hz2, filt_hz3);
>- SLDI_B2_SH(hz_out2, hz_out4, hz_out0, hz_out2, hz_out1, hz_out3, 8);
>+ SLDI_B2_SH(hz_out2, hz_out0, hz_out4, hz_out2, 8, hz_out1, hz_out3);
>
> filt = LD_SH(filter_vert);
> SPLATI_H4_SH(filt, 0, 1, 2, 3, filt_vt0, filt_vt1, filt_vt2, filt_vt3);
>@@ -2093,7 +2093,7 @@ void ff_put_bilin_64h_msa(uint8_t *dst, ptrdiff_t dst_stride,
> src4 = LD_SB(src + 32);
> src6 = LD_SB(src + 48);
> src7 = LD_SB(src + 56);
>- SLDI_B3_SB(src2, src4, src6, src0, src2, src4, src1, src3, src5, 8);
>+ SLDI_B3_SB(src2, src0, src4, src2, src6, src4, 8, src1, src3, src5);
> src += src_stride;
>
> VSHF_B2_UB(src0, src0, src1, src1, mask, mask, vec0, vec1);
>@@ -2544,8 +2544,8 @@ static void common_hv_2ht_2vt_4x8_msa(const uint8_t *src, int32_t
>src_stride,
> hz_out4 = HORIZ_2TAP_FILT_UH(src4, src5, mask, filt_hz, 7);
> hz_out6 = HORIZ_2TAP_FILT_UH(src6, src7, mask, filt_hz, 7);
> hz_out8 = HORIZ_2TAP_FILT_UH(src8, src8, mask, filt_hz, 7);
>- SLDI_B3_UH(hz_out2, hz_out4, hz_out6, hz_out0, hz_out2, hz_out4, hz_out1,
>- hz_out3, hz_out5, 8);
>+ SLDI_B3_UH(hz_out2, hz_out0, hz_out4, hz_out2, hz_out6, hz_out4, 8, hz_out1,
>+ hz_out3, hz_out5);
> hz_out7 = (v8u16) __msa_pckod_d((v2i64) hz_out8, (v2i64) hz_out6);
>
> ILVEV_B2_UB(hz_out0, hz_out1, hz_out2, hz_out3, vec0, vec1);
>@@ -3146,7 +3146,7 @@ void ff_avg_bilin_64h_msa(uint8_t *dst, ptrdiff_t dst_stride,
> for (loop_cnt = height; loop_cnt--;) {
> LD_SB4(src, 16, src0, src2, src4, src6);
> src7 = LD_SB(src + 56);
>- SLDI_B3_SB(src2, src4, src6, src0, src2, src4, src1, src3, src5, 8);
>+ SLDI_B3_SB(src2, src0, src4, src2, src6, src4, 8, src1, src3, src5);
> src += src_stride;
>
> VSHF_B2_UB(src0, src0, src1, src1, mask, mask, vec0, vec1);
>@@ -3655,8 +3655,8 @@ static void common_hv_2ht_2vt_and_aver_dst_4x8_msa(const uint8_t *src,
> hz_out4 = HORIZ_2TAP_FILT_UH(src4, src5, mask, filt_hz, 7);
> hz_out6 = HORIZ_2TAP_FILT_UH(src6, src7, mask, filt_hz, 7);
> hz_out8 = HORIZ_2TAP_FILT_UH(src8, src8, mask, filt_hz, 7);
>- SLDI_B3_UH(hz_out2, hz_out4, hz_out6, hz_out0, hz_out2, hz_out4, hz_out1,
>- hz_out3, hz_out5, 8);
>+ SLDI_B3_UH(hz_out2, hz_out0, hz_out4, hz_out2, hz_out6, hz_out4, 8, hz_out1,
>+ hz_out3, hz_out5);
> hz_out7 = (v8u16) __msa_pckod_d((v2i64) hz_out8, (v2i64) hz_out6);
>
> LW4(dst, dst_stride, tp0, tp1, tp2, tp3);
>diff --git a/libavutil/mips/generic_macros_msa.h b/libavutil/mips/generic_macros_msa.h
>index 9ac0583..528f45e 100644
>--- a/libavutil/mips/generic_macros_msa.h
>+++ b/libavutil/mips/generic_macros_msa.h
>@@ -602,67 +602,48 @@
> }
> #define AVER_UB4_UB(...) AVER_UB4(v16u8, __VA_ARGS__)
>
>-/* Description : Immediate number of columns to slide with zero
>- Arguments : Inputs - in0, in1, slide_val
>- Outputs - out0, out1
>+/* Description : Immediate number of columns to slide
>+ Arguments : Inputs - s, d, slide_val
>+ Outputs - out
> Return Type - as per RTYPE
>- Details : Byte elements from 'zero_m' vector are slide into 'in0' by
>+ Details : Byte elements from 'd' vector are slide into 's' by
> number of elements specified by 'slide_val'
> */
>-#define SLDI_B2_0(RTYPE, in0, in1, out0, out1, slide_val) \
>-{ \
>- v16i8 zero_m = { 0 }; \
>- out0 = (RTYPE) __msa_sldi_b((v16i8) zero_m, (v16i8) in0, slide_val); \
>- out1 = (RTYPE) __msa_sldi_b((v16i8) zero_m, (v16i8) in1, slide_val); \
>-}
>-#define SLDI_B2_0_UB(...) SLDI_B2_0(v16u8, __VA_ARGS__)
>-#define SLDI_B2_0_SB(...) SLDI_B2_0(v16i8, __VA_ARGS__)
>-#define SLDI_B2_0_SW(...) SLDI_B2_0(v4i32, __VA_ARGS__)
>-
>-#define SLDI_B3_0(RTYPE, in0, in1, in2, out0, out1, out2, slide_val) \
>-{ \
>- v16i8 zero_m = { 0 }; \
>- SLDI_B2_0(RTYPE, in0, in1, out0, out1, slide_val); \
>- out2 = (RTYPE) __msa_sldi_b((v16i8) zero_m, (v16i8) in2, slide_val); \
>-}
>-#define SLDI_B3_0_UB(...) SLDI_B3_0(v16u8, __VA_ARGS__)
>-#define SLDI_B3_0_SB(...) SLDI_B3_0(v16i8, __VA_ARGS__)
>-
>-#define SLDI_B4_0(RTYPE, in0, in1, in2, in3, \
>- out0, out1, out2, out3, slide_val) \
>-{ \
>- SLDI_B2_0(RTYPE, in0, in1, out0, out1, slide_val); \
>- SLDI_B2_0(RTYPE, in2, in3, out2, out3, slide_val); \
>+#define SLDI_B(RTYPE, d, s, slide_val, out) \
>+{ \
>+ out = (RTYPE) __msa_sldi_b((v16i8) d, (v16i8) s, slide_val); \
> }
>-#define SLDI_B4_0_UB(...) SLDI_B4_0(v16u8, __VA_ARGS__)
>-#define SLDI_B4_0_SB(...) SLDI_B4_0(v16i8, __VA_ARGS__)
>-#define SLDI_B4_0_SH(...) SLDI_B4_0(v8i16, __VA_ARGS__)
>
>-/* Description : Immediate number of columns to slide
>- Arguments : Inputs - in0_0, in0_1, in1_0, in1_1, slide_val
>- Outputs - out0, out1
>- Return Type - as per RTYPE
>- Details : Byte elements from 'in0_0' vector are slide into 'in1_0' by
>- number of elements specified by 'slide_val'
>-*/
>-#define SLDI_B2(RTYPE, in0_0, in0_1, in1_0, in1_1, out0, out1, slide_val) \
>-{ \
>- out0 = (RTYPE) __msa_sldi_b((v16i8) in0_0, (v16i8) in1_0, slide_val); \
>- out1 = (RTYPE) __msa_sldi_b((v16i8) in0_1, (v16i8) in1_1, slide_val); \
>+#define SLDI_B2(RTYPE, d0, s0, d1, s1, slide_val, out0, out1) \
>+{ \
>+ SLDI_B(RTYPE, d0, s0, slide_val, out0) \
>+ SLDI_B(RTYPE, d1, s1, slide_val, out1) \
> }
> #define SLDI_B2_UB(...) SLDI_B2(v16u8, __VA_ARGS__)
> #define SLDI_B2_SB(...) SLDI_B2(v16i8, __VA_ARGS__)
> #define SLDI_B2_SH(...) SLDI_B2(v8i16, __VA_ARGS__)
>+#define SLDI_B2_SW(...) SLDI_B2(v4i32, __VA_ARGS__)
>
>-#define SLDI_B3(RTYPE, in0_0, in0_1, in0_2, in1_0, in1_1, in1_2, \
>- out0, out1, out2, slide_val) \
>-{ \
>- SLDI_B2(RTYPE, in0_0, in0_1, in1_0, in1_1, out0, out1, slide_val) \
>- out2 = (RTYPE) __msa_sldi_b((v16i8) in0_2, (v16i8) in1_2, slide_val); \
>+#define SLDI_B3(RTYPE, d0, s0, d1, s1, d2, s2, slide_val, \
>+ out0, out1, out2) \
>+{ \
>+ SLDI_B2(RTYPE, d0, s0, d1, s1, slide_val, out0, out1) \
>+ SLDI_B(RTYPE, d2, s2, slide_val, out2) \
> }
>+#define SLDI_B3_UB(...) SLDI_B3(v16u8, __VA_ARGS__)
> #define SLDI_B3_SB(...) SLDI_B3(v16i8, __VA_ARGS__)
> #define SLDI_B3_UH(...) SLDI_B3(v8u16, __VA_ARGS__)
>
>+#define SLDI_B4(RTYPE, d0, s0, d1, s1, d2, s2, d3, s3, \
>+ slide_val, out0, out1, out2, out3) \
>+{ \
>+ SLDI_B2(RTYPE, d0, s0, d1, s1, slide_val, out0, out1) \
>+ SLDI_B2(RTYPE, d2, s2, d3, s3, slide_val, out2, out3) \
>+}
>+#define SLDI_B4_UB(...) SLDI_B4(v16u8, __VA_ARGS__)
>+#define SLDI_B4_SB(...) SLDI_B4(v16i8, __VA_ARGS__)
>+#define SLDI_B4_SH(...) SLDI_B4(v8i16, __VA_ARGS__)
>+
> /* Description : Shuffle byte vector elements as per mask vector
> Arguments : Inputs - in0, in1, in2, in3, mask0, mask1
> Outputs - out0, out1
>@@ -2433,6 +2414,7 @@
> { \
> v16i8 tmp0_m, tmp1_m, tmp2_m, tmp3_m; \
> v16i8 tmp4_m, tmp5_m, tmp6_m, tmp7_m; \
>+ v16i8 zeros = { 0 }; \
> \
> ILVR_B4_SB(in2, in0, in3, in1, in6, in4, in7, in5, \
> tmp0_m, tmp1_m, tmp2_m, tmp3_m); \
>@@ -2440,8 +2422,8 @@
> ILVRL_B2_SB(tmp3_m, tmp2_m, tmp6_m, tmp7_m); \
> ILVRL_W2(RTYPE, tmp6_m, tmp4_m, out0, out2); \
> ILVRL_W2(RTYPE, tmp7_m, tmp5_m, out4, out6); \
>- SLDI_B2_0(RTYPE, out0, out2, out1, out3, 8); \
>- SLDI_B2_0(RTYPE, out4, out6, out5, out7, 8); \
>+ SLDI_B4(RTYPE, zeros, out0, zeros, out2, zeros, out4, zeros, out6, \
>+ 8, out1, out3, out5, out7); \
> }
> #define TRANSPOSE8x8_UB_UB(...) TRANSPOSE8x8_UB(v16u8, __VA_ARGS__)
> #define TRANSPOSE8x8_UB_UH(...) TRANSPOSE8x8_UB(v8u16, __VA_ARGS__)
>--
>2.1.0
>
>
>_______________________________________________
>ffmpeg-devel mailing list
>ffmpeg-devel at ffmpeg.org
>https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>To unsubscribe, visit link above, or email
>ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
More information about the ffmpeg-devel
mailing list