avcodec/avcodec: Free frame_thread_encoder on avcodec_open2() error

[ffmpeg] / libavcodec / mips / hevc_lpf_sao_msa.c

diff --git a/libavcodec/mips/hevc_lpf_sao_msa.c b/libavcodec/mips/hevc_lpf_sao_msa.c
index ac218064043ea49a56e41e66c2487115df7d955b..26663dd89bc30ec5b4d146b66d25cfe5a7b20ca0 100644 (file)
--- a/libavcodec/mips/hevc_lpf_sao_msa.c
+++ b/libavcodec/mips/hevc_lpf_sao_msa.c
@@ -140,19 +140,19 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             temp1 = ((p3_src + p2_src) << 1) + p2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst0 = (v16u8) (temp2 + (v8i16) p2_src);
 
             temp1 = temp0 + p2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - p1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst1 = (v16u8) (temp2 + (v8i16) p1_src);
 
             temp1 = (temp0 << 1) + p2_src + q1_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst2 = (v16u8) (temp2 + (v8i16) p0_src);
 
             dst0 = __msa_bmz_v(dst0, (v16u8) p2_src, (v16u8) p_is_pcm_vec);
@@ -165,19 +165,19 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             temp1 = ((q3_src + q2_src) << 1) + q2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst5 = (v16u8) (temp2 + (v8i16) q2_src);
 
             temp1 = temp0 + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - q1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst4 = (v16u8) (temp2 + (v8i16) q1_src);
 
             temp1 = (temp0 << 1) + p1_src + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst3 = (v16u8) (temp2 + (v8i16) q0_src);
 
             dst3 = __msa_bmz_v(dst3, (v16u8) q0_src, (v16u8) q_is_pcm_vec);
@@ -218,15 +218,15 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             abs_delta0 = __msa_add_a_h(delta0, (v8i16) zero);
             abs_delta0 = (v8u16) abs_delta0 < temp1;
 
-            delta0 = CLIP_SH(delta0, tc_neg, tc_pos);
+            CLIP_SH(delta0, tc_neg, tc_pos);
 
-            temp0 = (v8u16) (delta0 + p0_src);
-            temp0 = (v8u16) CLIP_SH_0_255(temp0);
-            temp0 = (v8u16) __msa_bmz_v((v16u8) temp0, (v16u8) p0_src,
+            temp2 = (v8i16) (delta0 + p0_src);
+            CLIP_SH_0_255(temp2);
+            temp0 = (v8u16) __msa_bmz_v((v16u8) temp2, (v16u8) p0_src,
                                         (v16u8) p_is_pcm_vec);
 
             temp2 = (v8i16) (q0_src - delta0);
-            temp2 = CLIP_SH_0_255(temp2);
+            CLIP_SH_0_255(temp2);
             temp2 = (v8i16) __msa_bmz_v((v16u8) temp2, (v16u8) q0_src,
                                         (v16u8) q_is_pcm_vec);
 
@@ -252,9 +252,9 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             delta1 -= (v8i16) p1_src;
             delta1 += delta0;
             delta1 >>= 1;
-            delta1 = CLIP_SH(delta1, tc_neg, tc_pos);
+            CLIP_SH(delta1, tc_neg, tc_pos);
             delta1 = (v8i16) p1_src + (v8i16) delta1;
-            delta1 = CLIP_SH_0_255(delta1);
+            CLIP_SH_0_255(delta1);
             delta1 = (v8i16) __msa_bmnz_v((v16u8) delta1, (v16u8) p1_src,
                                           (v16u8) p_is_pcm_vec);
 
@@ -262,9 +262,9 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             delta2 = delta2 - (v8i16) q1_src;
             delta2 = delta2 - delta0;
             delta2 = delta2 >> 1;
-            delta2 = CLIP_SH(delta2, tc_neg, tc_pos);
+            CLIP_SH(delta2, tc_neg, tc_pos);
             delta2 = (v8i16) q1_src + (v8i16) delta2;
-            delta2 = CLIP_SH_0_255(delta2);
+            CLIP_SH_0_255(delta2);
             delta2 = (v8i16) __msa_bmnz_v((v16u8) delta2, (v16u8) q1_src,
                                           (v16u8) q_is_pcm_vec);
 
@@ -298,19 +298,19 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             temp1 = ((p3_src + p2_src) << 1) + p2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst0 = (v16u8) (temp2 + (v8i16) p2_src);
 
             temp1 = temp0 + p2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - p1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst1 = (v16u8) (temp2 + (v8i16) p1_src);
 
             temp1 = (temp0 << 1) + p2_src + q1_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst2 = (v16u8) (temp2 + (v8i16) p0_src);
 
             dst0 = __msa_bmz_v(dst0, (v16u8) p2_src, (v16u8) p_is_pcm_vec);
@@ -323,19 +323,19 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             temp1 = ((q3_src + q2_src) << 1) + q2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst5 = (v16u8) (temp2 + (v8i16) q2_src);
 
             temp1 = temp0 + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - q1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst4 = (v16u8) (temp2 + (v8i16) q1_src);
 
             temp1 = (temp0 << 1) + p1_src + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst3 = (v16u8) (temp2 + (v8i16) q0_src);
 
             dst3 = __msa_bmz_v(dst3, (v16u8) q0_src, (v16u8) q_is_pcm_vec);
@@ -362,15 +362,15 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             abs_delta0 = __msa_add_a_h(delta0, (v8i16) zero);
             abs_delta0 = (v8u16) abs_delta0 < temp1;
 
-            delta0 = CLIP_SH(delta0, tc_neg, tc_pos);
+            CLIP_SH(delta0, tc_neg, tc_pos);
 
-            temp0 = (v8u16) (delta0 + p0_src);
-            temp0 = (v8u16) CLIP_SH_0_255(temp0);
-            temp0 = (v8u16) __msa_bmz_v((v16u8) temp0, (v16u8) p0_src,
+            temp2 = (v8i16) (delta0 + p0_src);
+            CLIP_SH_0_255(temp2);
+            temp0 = (v8u16) __msa_bmz_v((v16u8) temp2, (v16u8) p0_src,
                                         (v16u8) p_is_pcm_vec);
 
             temp2 = (v8i16) (q0_src - delta0);
-            temp2 = CLIP_SH_0_255(temp2);
+            CLIP_SH_0_255(temp2);
             temp2 = (v8i16) __msa_bmz_v((v16u8) temp2, (v16u8) q0_src,
                                         (v16u8) q_is_pcm_vec);
 
@@ -394,9 +394,9 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             delta1 -= (v8i16) p1_src;
             delta1 += delta0;
             delta1 >>= 1;
-            delta1 = CLIP_SH(delta1, tc_neg, tc_pos);
+            CLIP_SH(delta1, tc_neg, tc_pos);
             delta1 = (v8i16) p1_src + (v8i16) delta1;
-            delta1 = CLIP_SH_0_255(delta1);
+            CLIP_SH_0_255(delta1);
             delta1 = (v8i16) __msa_bmnz_v((v16u8) delta1, (v16u8) p1_src,
                                           (v16u8) p_is_pcm_vec);
 
@@ -404,9 +404,9 @@ static void hevc_loopfilter_luma_hor_msa(uint8_t *src, int32_t stride,
             delta2 = delta2 - (v8i16) q1_src;
             delta2 = delta2 - delta0;
             delta2 = delta2 >> 1;
-            delta2 = CLIP_SH(delta2, tc_neg, tc_pos);
+            CLIP_SH(delta2, tc_neg, tc_pos);
             delta2 = (v8i16) q1_src + (v8i16) delta2;
-            delta2 = CLIP_SH_0_255(delta2);
+            CLIP_SH_0_255(delta2);
             delta2 = (v8i16) __msa_bmnz_v((v16u8) delta2, (v16u8) q1_src,
                                           (v16u8) q_is_pcm_vec);
 
@@ -561,19 +561,19 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             temp1 = ((p3_src + p2_src) << 1) + p2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst0 = (v16u8) (temp2 + (v8i16) p2_src);
 
             temp1 = temp0 + p2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - p1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst1 = (v16u8) (temp2 + (v8i16) p1_src);
 
             temp1 = (temp0 << 1) + p2_src + q1_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst2 = (v16u8) (temp2 + (v8i16) p0_src);
 
             dst0 = __msa_bmz_v(dst0, (v16u8) p2_src, (v16u8) p_is_pcm_vec);
@@ -585,19 +585,19 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             temp1 = ((q3_src + q2_src) << 1) + q2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst5 = (v16u8) (temp2 + (v8i16) q2_src);
 
             temp1 = temp0 + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - q1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst4 = (v16u8) (temp2 + (v8i16) q1_src);
 
             temp1 = (temp0 << 1) + p1_src + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst3 = (v16u8) (temp2 + (v8i16) q0_src);
 
             dst3 = __msa_bmz_v(dst3, (v16u8) q0_src, (v16u8) q_is_pcm_vec);
@@ -620,14 +620,14 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             abs_delta0 = __msa_add_a_h(delta0, (v8i16) zero);
             abs_delta0 = (v8u16) abs_delta0 < temp1;
 
-            delta0 = CLIP_SH(delta0, tc_neg, tc_pos);
-            temp0 = (v8u16) (delta0 + p0_src);
-            temp0 = (v8u16) CLIP_SH_0_255(temp0);
-            temp0 = (v8u16) __msa_bmz_v((v16u8) temp0, (v16u8) p0_src,
+            CLIP_SH(delta0, tc_neg, tc_pos);
+            temp2 = (v8i16) (delta0 + p0_src);
+            CLIP_SH_0_255(temp2);
+            temp0 = (v8u16) __msa_bmz_v((v16u8) temp2, (v16u8) p0_src,
                                         (v16u8) p_is_pcm_vec);
 
             temp2 = (v8i16) (q0_src - delta0);
-            temp2 = CLIP_SH_0_255(temp2);
+            CLIP_SH_0_255(temp2);
             temp2 = (v8i16) __msa_bmz_v((v16u8) temp2, (v16u8) q0_src,
                                         (v16u8) q_is_pcm_vec);
 
@@ -649,9 +649,9 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             delta1 -= (v8i16) p1_src;
             delta1 += delta0;
             delta1 >>= 1;
-            delta1 = CLIP_SH(delta1, tc_neg, tc_pos);
+            CLIP_SH(delta1, tc_neg, tc_pos);
             delta1 = (v8i16) p1_src + (v8i16) delta1;
-            delta1 = CLIP_SH_0_255(delta1);
+            CLIP_SH_0_255(delta1);
             delta1 = (v8i16) __msa_bmnz_v((v16u8) delta1, (v16u8) p1_src,
                                           (v16u8) p_is_pcm_vec);
 
@@ -659,9 +659,9 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             delta2 = delta2 - (v8i16) q1_src;
             delta2 = delta2 - delta0;
             delta2 = delta2 >> 1;
-            delta2 = CLIP_SH(delta2, tc_neg, tc_pos);
+            CLIP_SH(delta2, tc_neg, tc_pos);
             delta2 = (v8i16) q1_src + (v8i16) delta2;
-            delta2 = CLIP_SH_0_255(delta2);
+            CLIP_SH_0_255(delta2);
             delta2 = (v8i16) __msa_bmnz_v((v16u8) delta2, (v16u8) q1_src,
                                           (v16u8) q_is_pcm_vec);
 
@@ -726,19 +726,19 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             temp1 = ((p3_src + p2_src) << 1) + p2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst0 = (v16u8) (temp2 + (v8i16) p2_src);
 
             temp1 = temp0 + p2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - p1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst1 = (v16u8) (temp2 + (v8i16) p1_src);
 
             temp1 = (temp0 << 1) + p2_src + q1_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - p0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst2 = (v16u8) (temp2 + (v8i16) p0_src);
 
             dst0 = __msa_bmz_v(dst0, (v16u8) p2_src, (v16u8) p_is_pcm_vec);
@@ -750,19 +750,19 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             temp1 = ((q3_src + q2_src) << 1) + q2_src + temp0;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q2_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst5 = (v16u8) (temp2 + (v8i16) q2_src);
 
             temp1 = temp0 + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 2);
             temp2 = (v8i16) (temp1 - q1_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst4 = (v16u8) (temp2 + (v8i16) q1_src);
 
             temp1 = (temp0 << 1) + p1_src + q2_src;
             temp1 = (v8u16) __msa_srari_h((v8i16) temp1, 3);
             temp2 = (v8i16) (temp1 - q0_src);
-            temp2 = CLIP_SH(temp2, tc_neg, tc_pos);
+            CLIP_SH(temp2, tc_neg, tc_pos);
             dst3 = (v16u8) (temp2 + (v8i16) q0_src);
 
             dst3 = __msa_bmz_v(dst3, (v16u8) q0_src, (v16u8) q_is_pcm_vec);
@@ -785,15 +785,15 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             abs_delta0 = __msa_add_a_h(delta0, (v8i16) zero);
             abs_delta0 = (v8u16) abs_delta0 < temp1;
 
-            delta0 = CLIP_SH(delta0, tc_neg, tc_pos);
+            CLIP_SH(delta0, tc_neg, tc_pos);
 
-            temp0 = (v8u16) (delta0 + p0_src);
-            temp0 = (v8u16) CLIP_SH_0_255(temp0);
-            temp0 = (v8u16) __msa_bmz_v((v16u8) temp0, (v16u8) p0_src,
+            temp2 = (v8i16) (delta0 + p0_src);
+            CLIP_SH_0_255(temp2);
+            temp0 = (v8u16) __msa_bmz_v((v16u8) temp2, (v16u8) p0_src,
                                         (v16u8) p_is_pcm_vec);
 
             temp2 = (v8i16) (q0_src - delta0);
-            temp2 = CLIP_SH_0_255(temp2);
+            CLIP_SH_0_255(temp2);
             temp2 = (v8i16) __msa_bmz_v((v16u8) temp2, (v16u8) q0_src,
                                         (v16u8) q_is_pcm_vec);
 
@@ -815,9 +815,9 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             delta1 -= (v8i16) p1_src;
             delta1 += delta0;
             delta1 >>= 1;
-            delta1 = CLIP_SH(delta1, tc_neg, tc_pos);
+            CLIP_SH(delta1, tc_neg, tc_pos);
             delta1 = (v8i16) p1_src + (v8i16) delta1;
-            delta1 = CLIP_SH_0_255(delta1);
+            CLIP_SH_0_255(delta1);
             delta1 = (v8i16) __msa_bmnz_v((v16u8) delta1, (v16u8) p1_src,
                                           (v16u8) p_is_pcm_vec);
 
@@ -825,9 +825,9 @@ static void hevc_loopfilter_luma_ver_msa(uint8_t *src, int32_t stride,
             delta2 = delta2 - (v8i16) q1_src;
             delta2 = delta2 - delta0;
             delta2 = delta2 >> 1;
-            delta2 = CLIP_SH(delta2, tc_neg, tc_pos);
+            CLIP_SH(delta2, tc_neg, tc_pos);
             delta2 = (v8i16) q1_src + (v8i16) delta2;
-            delta2 = CLIP_SH_0_255(delta2);
+            CLIP_SH_0_255(delta2);
             delta2 = (v8i16) __msa_bmnz_v((v16u8) delta2, (v16u8) q1_src,
                                           (v16u8) q_is_pcm_vec);
             delta1 = (v8i16) __msa_bmz_v((v16u8) delta1, (v16u8) p1_src,
@@ -955,15 +955,15 @@ static void hevc_loopfilter_chroma_hor_msa(uint8_t *src, int32_t stride,
         temp0 <<= 2;
         temp0 += temp1;
         delta = __msa_srari_h((v8i16) temp0, 3);
-        delta = CLIP_SH(delta, tc_neg, tc_pos);
+        CLIP_SH(delta, tc_neg, tc_pos);
 
         temp0 = (v8i16) ((v8i16) p0 + delta);
-        temp0 = CLIP_SH_0_255(temp0);
+        CLIP_SH_0_255(temp0);
         temp0 = (v8i16) __msa_bmz_v((v16u8) temp0, (v16u8) p0,
                                     (v16u8) p_is_pcm_vec);
 
         temp1 = (v8i16) ((v8i16) q0 - delta);
-        temp1 = CLIP_SH_0_255(temp1);
+        CLIP_SH_0_255(temp1);
         temp1 = (v8i16) __msa_bmz_v((v16u8) temp1, (v16u8) q0,
                                     (v16u8) q_is_pcm_vec);
 
@@ -1014,15 +1014,15 @@ static void hevc_loopfilter_chroma_ver_msa(uint8_t *src, int32_t stride,
         temp0 <<= 2;
         temp0 += temp1;
         delta = __msa_srari_h((v8i16) temp0, 3);
-        delta = CLIP_SH(delta, tc_neg, tc_pos);
+        CLIP_SH(delta, tc_neg, tc_pos);
 
         temp0 = (v8i16) ((v8i16) p0 + delta);
-        temp0 = CLIP_SH_0_255(temp0);
+        CLIP_SH_0_255(temp0);
         temp0 = (v8i16) __msa_bmz_v((v16u8) temp0, (v16u8) p0,
                                     (v16u8) p_is_pcm_vec);
 
         temp1 = (v8i16) ((v8i16) q0 - delta);
-        temp1 = CLIP_SH_0_255(temp1);
+        CLIP_SH_0_255(temp1);
         temp1 = (v8i16) __msa_bmz_v((v16u8) temp1, (v16u8) q0,
                                     (v16u8) q_is_pcm_vec);
 
@@ -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,