Change DSPContext.vector_fmul() from dst=dst*src to dest=src0*src1.

Signed-off-by: Mans Rullgard <mans@mansr.com>

13 files changed, 67 insertions, 68 deletions

diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c
index f8f0eb45f6..6a113ef30b 100644
--- a/libavcodec/aacenc.c
+++ b/libavcodec/aacenc.c
@@ -256,7 +256,7 @@ static void apply_window_and_mdct(AVCodecContext *avctx, AACEncContext *s,
                 s->output[i - 448 - k] = (i < 1024)
                                          ? sce->saved[i]
                                          : audio[(i-1024)*chans];
-            s->dsp.vector_fmul        (s->output,     k ?  swindow : pwindow, 128);
+            s->dsp.vector_fmul        (s->output,     s->output, k ?  swindow : pwindow, 128);
             s->dsp.vector_fmul_reverse(s->output+128, s->output+128, swindow, 128);
             ff_mdct_calc(&s->mdct128, sce->coeffs + k, s->output);
         }
diff --git a/libavcodec/arm/dsputil_init_neon.c b/libavcodec/arm/dsputil_init_neon.c
index 04ebb00576..221183cef8 100644
--- a/libavcodec/arm/dsputil_init_neon.c
+++ b/libavcodec/arm/dsputil_init_neon.c
@@ -138,7 +138,7 @@ void ff_avg_h264_chroma_mc2_neon(uint8_t *, uint8_t *, int, int, int, int);
 void ff_vp3_v_loop_filter_neon(uint8_t *, int, int *);
 void ff_vp3_h_loop_filter_neon(uint8_t *, int, int *);
 
-void ff_vector_fmul_neon(float *dst, const float *src, int len);
+void ff_vector_fmul_neon(float *dst, const float *src0, const float *src1, int len);
 void ff_vector_fmul_window_neon(float *dst, const float *src0,
                                 const float *src1, const float *win,
                                 float add_bias, int len);
diff --git a/libavcodec/arm/dsputil_init_vfp.c b/libavcodec/arm/dsputil_init_vfp.c
index 9f8c1b7840..76ef6b4171 100644
--- a/libavcodec/arm/dsputil_init_vfp.c
+++ b/libavcodec/arm/dsputil_init_vfp.c
@@ -21,7 +21,8 @@
 #include "libavcodec/dsputil.h"
 #include "dsputil_arm.h"
 
-void ff_vector_fmul_vfp(float *dst, const float *src, int len);
+void ff_vector_fmul_vfp(float *dst, const float *src0,
+                        const float *src1, int len);
 void ff_vector_fmul_reverse_vfp(float *dst, const float *src0,
                                 const float *src1, int len);
 void ff_float_to_int16_vfp(int16_t *dst, const float *src, long len);
diff --git a/libavcodec/arm/dsputil_neon.S b/libavcodec/arm/dsputil_neon.S
index 2bcdb397f9..42fb38de52 100644
--- a/libavcodec/arm/dsputil_neon.S
+++ b/libavcodec/arm/dsputil_neon.S
@@ -738,42 +738,41 @@ function ff_float_to_int16_interleave_neon, export=1
 endfunc
 
 function ff_vector_fmul_neon, export=1
-        mov             r3,  r0
-        subs            r2,  r2,  #8
-        vld1.64         {d0-d3},  [r0,:128]!
-        vld1.64         {d4-d7},  [r1,:128]!
+        subs            r3,  r3,  #8
+        vld1.64         {d0-d3},  [r1,:128]!
+        vld1.64         {d4-d7},  [r2,:128]!
         vmul.f32        q8,  q0,  q2
         vmul.f32        q9,  q1,  q3
         beq             3f
-        bics            ip,  r2,  #15
+        bics            ip,  r3,  #15
         beq             2f
 1:      subs            ip,  ip,  #16
-        vld1.64         {d0-d1},  [r0,:128]!
-        vld1.64         {d4-d5},  [r1,:128]!
+        vld1.64         {d0-d1},  [r1,:128]!
+        vld1.64         {d4-d5},  [r2,:128]!
         vmul.f32        q10, q0,  q2
-        vld1.64         {d2-d3},  [r0,:128]!
-        vld1.64         {d6-d7},  [r1,:128]!
+        vld1.64         {d2-d3},  [r1,:128]!
+        vld1.64         {d6-d7},  [r2,:128]!
         vmul.f32        q11, q1,  q3
-        vst1.64         {d16-d19},[r3,:128]!
-        vld1.64         {d0-d1},  [r0,:128]!
-        vld1.64         {d4-d5},  [r1,:128]!
+        vst1.64         {d16-d19},[r0,:128]!
+        vld1.64         {d0-d1},  [r1,:128]!
+        vld1.64         {d4-d5},  [r2,:128]!
         vmul.f32        q8,  q0,  q2
-        vld1.64         {d2-d3},  [r0,:128]!
-        vld1.64         {d6-d7},  [r1,:128]!
+        vld1.64         {d2-d3},  [r1,:128]!
+        vld1.64         {d6-d7},  [r2,:128]!
         vmul.f32        q9,  q1,  q3
-        vst1.64         {d20-d23},[r3,:128]!
+        vst1.64         {d20-d23},[r0,:128]!
         bne             1b
-        ands            r2,  r2,  #15
+        ands            r3,  r3,  #15
         beq             3f
-2:      vld1.64         {d0-d1},  [r0,:128]!
-        vld1.64         {d4-d5},  [r1,:128]!
-        vst1.64         {d16-d17},[r3,:128]!
+2:      vld1.64         {d0-d1},  [r1,:128]!
+        vld1.64         {d4-d5},  [r2,:128]!
+        vst1.64         {d16-d17},[r0,:128]!
         vmul.f32        q8,  q0,  q2
-        vld1.64         {d2-d3},  [r0,:128]!
-        vld1.64         {d6-d7},  [r1,:128]!
-        vst1.64         {d18-d19},[r3,:128]!
+        vld1.64         {d2-d3},  [r1,:128]!
+        vld1.64         {d6-d7},  [r2,:128]!
+        vst1.64         {d18-d19},[r0,:128]!
         vmul.f32        q9,  q1,  q3
-3:      vst1.64         {d16-d19},[r3,:128]!
+3:      vst1.64         {d16-d19},[r0,:128]!
         bx              lr
 endfunc
 
diff --git a/libavcodec/arm/dsputil_vfp.S b/libavcodec/arm/dsputil_vfp.S
index b704ba9144..a65b69e20a 100644
--- a/libavcodec/arm/dsputil_vfp.S
+++ b/libavcodec/arm/dsputil_vfp.S
@@ -41,34 +41,33 @@
  * ARM VFP optimized implementation of 'vector_fmul_c' function.
  * Assume that len is a positive number and is multiple of 8
  */
-@ void ff_vector_fmul_vfp(float *dst, const float *src, int len)
+@ void ff_vector_fmul_vfp(float *dst, const float *src0, const float *src1, int len)
 function ff_vector_fmul_vfp, export=1
         vpush           {d8-d15}
-        mov             r3,  r0
         fmrx            r12, fpscr
         orr             r12, r12, #(3 << 16) /* set vector size to 4 */
         fmxr            fpscr, r12
 
-        vldmia          r3!, {s0-s3}
-        vldmia          r1!, {s8-s11}
-        vldmia          r3!, {s4-s7}
-        vldmia          r1!, {s12-s15}
+        vldmia          r1!, {s0-s3}
+        vldmia          r2!, {s8-s11}
+        vldmia          r1!, {s4-s7}
+        vldmia          r2!, {s12-s15}
         vmul.f32        s8,  s0,  s8
 1:
-        subs            r2,  r2,  #16
+        subs            r3,  r3,  #16
         vmul.f32        s12, s4,  s12
-        vldmiage        r3!, {s16-s19}
-        vldmiage        r1!, {s24-s27}
-        vldmiage        r3!, {s20-s23}
-        vldmiage        r1!, {s28-s31}
+        vldmiage        r1!, {s16-s19}
+        vldmiage        r2!, {s24-s27}
+        vldmiage        r1!, {s20-s23}
+        vldmiage        r2!, {s28-s31}
         vmulge.f32      s24, s16, s24
         vstmia          r0!, {s8-s11}
         vstmia          r0!, {s12-s15}
         vmulge.f32      s28, s20, s28
-        vldmiagt        r3!, {s0-s3}
-        vldmiagt        r1!, {s8-s11}
-        vldmiagt        r3!, {s4-s7}
-        vldmiagt        r1!, {s12-s15}
+        vldmiagt        r1!, {s0-s3}
+        vldmiagt        r2!, {s8-s11}
+        vldmiagt        r1!, {s4-s7}
+        vldmiagt        r2!, {s12-s15}
         vmulge.f32      s8,  s0,  s8
         vstmiage        r0!, {s24-s27}
         vstmiage        r0!, {s28-s31}
diff --git a/libavcodec/atrac3.c b/libavcodec/atrac3.c
index 797e1f1992..cc13b730ab 100644
--- a/libavcodec/atrac3.c
+++ b/libavcodec/atrac3.c
@@ -159,7 +159,7 @@ static void IMLT(ATRAC3Context *q, float *pInput, float *pOutput, int odd_band)
     ff_imdct_calc(&q->mdct_ctx,pOutput,pInput);
 
     /* Perform windowing on the output. */
-    dsp.vector_fmul(pOutput,mdct_window,512);
+    dsp.vector_fmul(pOutput, pOutput, mdct_window, 512);
 
 }
 
diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 57b2640686..2ed0052977 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -3750,10 +3750,10 @@ WRAPPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
 WRAPPER8_16_SQ(rd8x8_c, rd16_c)
 WRAPPER8_16_SQ(bit8x8_c, bit16_c)
 
-static void vector_fmul_c(float *dst, const float *src, int len){
+static void vector_fmul_c(float *dst, const float *src0, const float *src1, int len){
     int i;
     for(i=0; i<len; i++)
-        dst[i] *= src[i];
+        dst[i] = src0[i] * src1[i];
 }
 
 static void vector_fmul_reverse_c(float *dst, const float *src0, const float *src1, int len){
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index 842d72746d..baa68bebe7 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -375,7 +375,7 @@ typedef struct DSPContext {
     void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
     void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
     /* assume len is a multiple of 8, and arrays are 16-byte aligned */
-    void (*vector_fmul)(float *dst, const float *src, int len);
+    void (*vector_fmul)(float *dst, const float *src0, const float *src1, int len);
     void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
     /* assume len is a multiple of 8, and src arrays are 16-byte aligned */
     void (*vector_fmul_add)(float *dst, const float *src0, const float *src1, const float *src2, int len);
diff --git a/libavcodec/nellymoserenc.c b/libavcodec/nellymoserenc.c
index b3f6aa31d2..03b647e297 100644
--- a/libavcodec/nellymoserenc.c
+++ b/libavcodec/nellymoserenc.c
@@ -113,13 +113,13 @@ static const uint8_t quant_lut_offset[8] = { 0, 0, 1, 4, 11, 32, 81, 230 };
 
 static void apply_mdct(NellyMoserEncodeContext *s)
 {
-    memcpy(s->in_buff, s->buf[s->bufsel], NELLY_BUF_LEN * sizeof(float));
-    s->dsp.vector_fmul(s->in_buff, ff_sine_128, NELLY_BUF_LEN);
+    s->dsp.vector_fmul(s->in_buff, s->buf[s->bufsel], ff_sine_128, NELLY_BUF_LEN);
     s->dsp.vector_fmul_reverse(s->in_buff + NELLY_BUF_LEN, s->buf[s->bufsel] + NELLY_BUF_LEN, ff_sine_128,
                                NELLY_BUF_LEN);
     ff_mdct_calc(&s->mdct_ctx, s->mdct_out, s->in_buff);
 
-    s->dsp.vector_fmul(s->buf[s->bufsel] + NELLY_BUF_LEN, ff_sine_128, NELLY_BUF_LEN);
+    s->dsp.vector_fmul(s->buf[s->bufsel] + NELLY_BUF_LEN, s->buf[s->bufsel] + NELLY_BUF_LEN,
+                       ff_sine_128, NELLY_BUF_LEN);
     s->dsp.vector_fmul_reverse(s->buf[s->bufsel] + 2 * NELLY_BUF_LEN, s->buf[1 - s->bufsel], ff_sine_128,
                                NELLY_BUF_LEN);
     ff_mdct_calc(&s->mdct_ctx, s->mdct_out + NELLY_BUF_LEN, s->buf[s->bufsel] + NELLY_BUF_LEN);
diff --git a/libavcodec/ppc/float_altivec.c b/libavcodec/ppc/float_altivec.c
index d1f9f1ade3..188e03ea2d 100644
--- a/libavcodec/ppc/float_altivec.c
+++ b/libavcodec/ppc/float_altivec.c
@@ -23,16 +23,16 @@
 #include "dsputil_altivec.h"
 #include "util_altivec.h"
 
-static void vector_fmul_altivec(float *dst, const float *src, int len)
+static void vector_fmul_altivec(float *dst, const float *src0, const float *src1, int len)
 {
     int i;
     vector float d0, d1, s, zero = (vector float)vec_splat_u32(0);
     for(i=0; i<len-7; i+=8) {
-        d0 = vec_ld(0, dst+i);
-        s = vec_ld(0, src+i);
-        d1 = vec_ld(16, dst+i);
+        d0 = vec_ld(0, src0+i);
+        s = vec_ld(0, src1+i);
+        d1 = vec_ld(16, src0+i);
         d0 = vec_madd(d0, s, zero);
-        d1 = vec_madd(d1, vec_ld(16,src+i), zero);
+        d1 = vec_madd(d1, vec_ld(16,src1+i), zero);
         vec_st(d0, 0, dst+i);
         vec_st(d1, 16, dst+i);
     }
diff --git a/libavcodec/twinvq.c b/libavcodec/twinvq.c
index 3d26c6e3cb..15907aebcf 100644
--- a/libavcodec/twinvq.c
+++ b/libavcodec/twinvq.c
@@ -783,7 +783,7 @@ static void read_and_decode_spectrum(TwinContext *tctx, GetBitContext *gb,
             dec_bark_env(tctx, bark1[i][j], bark_use_hist[i][j], i,
                          tctx->tmp_buf, gain[sub*i+j], ftype);
 
-            tctx->dsp.vector_fmul(chunk + block_size*j, tctx->tmp_buf,
+            tctx->dsp.vector_fmul(chunk + block_size*j, chunk + block_size*j, tctx->tmp_buf,
                                   block_size);
 
         }
@@ -805,7 +805,7 @@ static void read_and_decode_spectrum(TwinContext *tctx, GetBitContext *gb,
         dec_lpc_spectrum_inv(tctx, lsp, ftype, tctx->tmp_buf);
 
         for (j = 0; j < mtab->fmode[ftype].sub; j++) {
-            tctx->dsp.vector_fmul(chunk, tctx->tmp_buf, block_size);
+            tctx->dsp.vector_fmul(chunk, chunk, tctx->tmp_buf, block_size);
             chunk += block_size;
         }
     }
diff --git a/libavcodec/vorbis_dec.c b/libavcodec/vorbis_dec.c
index 4e16c4a7bc..8f15a2100c 100644
--- a/libavcodec/vorbis_dec.c
+++ b/libavcodec/vorbis_dec.c
@@ -1578,7 +1578,7 @@ static int vorbis_parse_audio_packet(vorbis_context *vc)
     for (j = vc->audio_channels-1;j >= 0; j--) {
         ch_floor_ptr = vc->channel_floors   + j           * blocksize / 2;
         ch_res_ptr   = vc->channel_residues + res_chan[j] * blocksize / 2;
-        vc->dsp.vector_fmul(ch_floor_ptr, ch_res_ptr, blocksize / 2);
+        vc->dsp.vector_fmul(ch_floor_ptr, ch_floor_ptr, ch_res_ptr, blocksize / 2);
         ff_imdct_half(&vc->mdct[blockflag], ch_res_ptr, ch_floor_ptr);
     }
 
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 825149e4a5..5ddfecae24 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -2074,38 +2074,38 @@ static void ac3_downmix_sse(float (*samples)[256], float (*matrix)[2], int out_c
     }
 }
 
-static void vector_fmul_3dnow(float *dst, const float *src, int len){
+static void vector_fmul_3dnow(float *dst, const float *src0, const float *src1, int len){
     x86_reg i = (len-4)*4;
     __asm__ volatile(
         "1: \n\t"
-        "movq    (%1,%0), %%mm0 \n\t"
-        "movq   8(%1,%0), %%mm1 \n\t"
-        "pfmul   (%2,%0), %%mm0 \n\t"
-        "pfmul  8(%2,%0), %%mm1 \n\t"
+        "movq    (%2,%0), %%mm0 \n\t"
+        "movq   8(%2,%0), %%mm1 \n\t"
+        "pfmul   (%3,%0), %%mm0 \n\t"
+        "pfmul  8(%3,%0), %%mm1 \n\t"
         "movq   %%mm0,  (%1,%0) \n\t"
         "movq   %%mm1, 8(%1,%0) \n\t"
         "sub  $16, %0 \n\t"
         "jge 1b \n\t"
         "femms  \n\t"
         :"+r"(i)
-        :"r"(dst), "r"(src)
+        :"r"(dst), "r"(src0), "r"(src1)
         :"memory"
     );
 }
-static void vector_fmul_sse(float *dst, const float *src, int len){
+static void vector_fmul_sse(float *dst, const float *src0, const float *src1, int len){
     x86_reg i = (len-8)*4;
     __asm__ volatile(
         "1: \n\t"
-        "movaps    (%1,%0), %%xmm0 \n\t"
-        "movaps  16(%1,%0), %%xmm1 \n\t"
-        "mulps     (%2,%0), %%xmm0 \n\t"
-        "mulps   16(%2,%0), %%xmm1 \n\t"
+        "movaps    (%2,%0), %%xmm0 \n\t"
+        "movaps  16(%2,%0), %%xmm1 \n\t"
+        "mulps     (%3,%0), %%xmm0 \n\t"
+        "mulps   16(%3,%0), %%xmm1 \n\t"
         "movaps  %%xmm0,   (%1,%0) \n\t"
         "movaps  %%xmm1, 16(%1,%0) \n\t"
         "sub  $32, %0 \n\t"
         "jge 1b \n\t"
         :"+r"(i)
-        :"r"(dst), "r"(src)
+        :"r"(dst), "r"(src0), "r"(src1)
         :"memory"
     );
 }