ffmpeg roll of source to mar 9 version...

patched-ffmpeg-mt/libavcodec/ppc/dsputil_altivec.c

 /*
  * Copyright (c) 2002 Brian Foley
  * Copyright (c) 2002 Dieter Shirley
  * Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
  *
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
@@ skipping 10 matching lines @@
  */
 
 #include "config.h"
 #if HAVE_ALTIVEC_H
 #include <altivec.h>
 #endif
 #include "libavcodec/dsputil.h"
 #include "dsputil_ppc.h"
 #include "util_altivec.h"
 #include "types_altivec.h"
+#include "dsputil_altivec.h"
 
-int sad16_x2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sad16_x2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
     vector unsigned char *tv;
     vector unsigned char pix1v, pix2v, pix2iv, avgv, t5;
     vector unsigned int sad;
     vector signed int sumdiffs;
 
     s = 0;
@@ skipping 24 matching lines @@
         pix2 += line_size;
     }
     /* Sum up the four partial sums, and put the result into s */
     sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
     sumdiffs = vec_splat(sumdiffs, 3);
     vec_ste(sumdiffs, 0, &s);
 
     return s;
 }
 
-int sad16_y2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sad16_y2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
     vector unsigned char *tv;
     vector unsigned char pix1v, pix2v, pix3v, avgv, t5;
     vector unsigned int sad;
     vector signed int sumdiffs;
     uint8_t *pix3 = pix2 + line_size;
 
@@ skipping 35 matching lines @@
 
     }
 
     /* Sum up the four partial sums, and put the result into s */
     sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
     sumdiffs = vec_splat(sumdiffs, 3);
     vec_ste(sumdiffs, 0, &s);
     return s;
 }
 
-int sad16_xy2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sad16_xy2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     uint8_t *pix3 = pix2 + line_size;
     const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
     const vector unsigned short two = (const vector unsigned short)vec_splat_u16(2);
     vector unsigned char *tv, avgv, t5;
     vector unsigned char pix1v, pix2v, pix3v, pix2iv, pix3iv;
     vector unsigned short pix2lv, pix2hv, pix2ilv, pix2ihv;
     vector unsigned short pix3lv, pix3hv, pix3ilv, pix3ihv;
@@ skipping 74 matching lines @@
         t2 = t4;
     }
     /* Sum up the four partial sums, and put the result into s */
     sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
     sumdiffs = vec_splat(sumdiffs, 3);
     vec_ste(sumdiffs, 0, &s);
 
     return s;
 }
 
-int sad16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sad16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
     vector unsigned char perm1, perm2, *pix1v, *pix2v;
     vector unsigned char t1, t2, t3,t4, t5;
     vector unsigned int sad;
     vector signed int sumdiffs;
 
     sad = (vector unsigned int)vec_splat_u32(0);
@@ skipping 21 matching lines @@
     }
 
     /* Sum up the four partial sums, and put the result into s */
     sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
     sumdiffs = vec_splat(sumdiffs, 3);
     vec_ste(sumdiffs, 0, &s);
 
     return s;
 }
 
-int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
     vector unsigned char perm1, perm2, permclear, *pix1v, *pix2v;
     vector unsigned char t1, t2, t3,t4, t5;
     vector unsigned int sad;
     vector signed int sumdiffs;
 
     sad = (vector unsigned int)vec_splat_u32(0);
@@ skipping 24 matching lines @@
     }
 
     /* Sum up the four partial sums, and put the result into s */
     sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
     sumdiffs = vec_splat(sumdiffs, 3);
     vec_ste(sumdiffs, 0, &s);
 
     return s;
 }
 
-int pix_norm1_altivec(uint8_t *pix, int line_size)
+static int pix_norm1_altivec(uint8_t *pix, int line_size)
 {
     int i;
     int s;
     const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
     vector unsigned char *tv;
     vector unsigned char pixv;
     vector unsigned int sv;
     vector signed int sum;
 
     sv = (vector unsigned int)vec_splat_u32(0);
@@ skipping 15 matching lines @@
     vec_ste(sum, 0, &s);
 
     return s;
 }
 
 /**
  * Sum of Squared Errors for a 8x8 block.
  * AltiVec-enhanced.
  * It's the sad8_altivec code above w/ squaring added.
  */
-int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
     vector unsigned char perm1, perm2, permclear, *pix1v, *pix2v;
     vector unsigned char t1, t2, t3,t4, t5;
     vector unsigned int sum;
     vector signed int sumsqr;
 
     sum = (vector unsigned int)vec_splat_u32(0);
@@ skipping 33 matching lines @@
     vec_ste(sumsqr, 0, &s);
 
     return s;
 }
 
 /**
  * Sum of Squared Errors for a 16x16 block.
  * AltiVec-enhanced.
  * It's the sad16_altivec code above w/ squaring added.
  */
-int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
+static int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s;
     const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
     vector unsigned char perm1, perm2, *pix1v, *pix2v;
     vector unsigned char t1, t2, t3,t4, t5;
     vector unsigned int sum;
     vector signed int sumsqr;
 
     sum = (vector unsigned int)vec_splat_u32(0);
@@ skipping 23 matching lines @@
     }
 
     /* Sum up the four partial sums, and put the result into s */
     sumsqr = vec_sums((vector signed int) sum, (vector signed int) zero);
     sumsqr = vec_splat(sumsqr, 3);
     vec_ste(sumsqr, 0, &s);
 
     return s;
 }
 
-int pix_sum_altivec(uint8_t * pix, int line_size)
+static int pix_sum_altivec(uint8_t * pix, int line_size)
 {
     const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
     vector unsigned char perm, *pixv;
     vector unsigned char t1;
     vector unsigned int sad;
     vector signed int sumdiffs;
 
     int i;
     int s;
 
@@ skipping 12 matching lines @@
     }
 
     /* Sum up the four partial sums, and put the result into s */
     sumdiffs = vec_sums((vector signed int) sad, (vector signed int) zero);
     sumdiffs = vec_splat(sumdiffs, 3);
     vec_ste(sumdiffs, 0, &s);
 
     return s;
 }
 
-void get_pixels_altivec(DCTELEM *restrict block, const uint8_t *pixels, int line_size)
+static void get_pixels_altivec(DCTELEM *restrict block, const uint8_t *pixels, int line_size)
 {
     int i;
     vector unsigned char perm, bytes, *pixv;
     const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
     vector signed short shorts;
 
     for (i = 0; i < 8; i++) {
         // Read potentially unaligned pixels.
         // We're reading 16 pixels, and actually only want 8,
         // but we simply ignore the extras.
         perm = vec_lvsl(0, pixels);
         pixv = (vector unsigned char *) pixels;
         bytes = vec_perm(pixv[0], pixv[1], perm);
 
         // convert the bytes into shorts
         shorts = (vector signed short)vec_mergeh(zero, bytes);
 
         // save the data to the block, we assume the block is 16-byte aligned
         vec_st(shorts, i*16, (vector signed short*)block);
 
         pixels += line_size;
     }
 }
 
-void diff_pixels_altivec(DCTELEM *restrict block, const uint8_t *s1,
+static void diff_pixels_altivec(DCTELEM *restrict block, const uint8_t *s1,
         const uint8_t *s2, int stride)
 {
     int i;
     vector unsigned char perm, bytes, *pixv;
     const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
     vector signed short shorts1, shorts2;
 
     for (i = 0; i < 4; i++) {
         // Read potentially unaligned pixels
         // We're reading 16 pixels, and actually only want 8,
@@ skipping 64 matching lines @@
     vec_st(zero_s16v,  16, block);
     vec_st(zero_s16v,  32, block);
     vec_st(zero_s16v,  48, block);
     vec_st(zero_s16v,  64, block);
     vec_st(zero_s16v,  80, block);
     vec_st(zero_s16v,  96, block);
     vec_st(zero_s16v, 112, block);
 }
 
 
-void add_bytes_altivec(uint8_t *dst, uint8_t *src, int w) {
+static void add_bytes_altivec(uint8_t *dst, uint8_t *src, int w) {
     register int i;
     register vector unsigned char vdst, vsrc;
 
     /* dst and src are 16 bytes-aligned (guaranteed) */
     for (i = 0 ; (i + 15) < w ; i+=16) {
         vdst = vec_ld(i, (unsigned char*)dst);
         vsrc = vec_ld(i, (unsigned char*)src);
         vdst = vec_add(vsrc, vdst);
         vec_st(vdst, i, (unsigned char*)dst);
     }
@@ skipping 77 matching lines @@
         blockv = vec_avg(blockv,pixelsv);
         vec_st(blockv, 0, (unsigned char*)block);
         pixels+=line_size;
         block +=line_size;
     }
 
 POWERPC_PERF_STOP_COUNT(altivec_avg_pixels16_num, 1);
 }
 
 /* next one assumes that ((line_size % 8) == 0) */
-void avg_pixels8_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
+static void avg_pixels8_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
 {
 POWERPC_PERF_DECLARE(altivec_avg_pixels8_num, 1);
     register vector unsigned char pixelsv1, pixelsv2, pixelsv, blockv;
     int i;
 
 POWERPC_PERF_START_COUNT(altivec_avg_pixels8_num, 1);
 
    for (i = 0; i < h; i++) {
        /* block is 8 bytes-aligned, so we're either in the
           left block (16 bytes-aligned) or in the right block (not) */
@@ skipping 15 matching lines @@
        vec_st(blockv, 0, block);
 
        pixels += line_size;
        block += line_size;
    }
 
 POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_num, 1);
 }
 
 /* next one assumes that ((line_size % 8) == 0) */
-void put_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+static void put_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
 {
 POWERPC_PERF_DECLARE(altivec_put_pixels8_xy2_num, 1);
     register int i;
     register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
     register vector unsigned char blockv, temp1, temp2;
     register vector unsigned short pixelssum1, pixelssum2, temp3;
     register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
     register const vector unsigned short vctwo = (const vector unsigned short)vec_splat_u16(2);
 
     temp1 = vec_ld(0, pixels);
@@ skipping 42 matching lines @@
         vec_st(blockv, 0, block);
 
         block += line_size;
         pixels += line_size;
     }
 
 POWERPC_PERF_STOP_COUNT(altivec_put_pixels8_xy2_num, 1);
 }
 
 /* next one assumes that ((line_size % 8) == 0) */
-void put_no_rnd_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+static void put_no_rnd_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
 {
 POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels8_xy2_num, 1);
     register int i;
     register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
     register vector unsigned char blockv, temp1, temp2;
     register vector unsigned short pixelssum1, pixelssum2, temp3;
     register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
     register const vector unsigned short vcone = (const vector unsigned short)vec_splat_u16(1);
     register const vector unsigned short vctwo = (const vector unsigned short)vec_splat_u16(2);
 
@@ skipping 43 matching lines @@
         vec_st(blockv, 0, block);
 
         block += line_size;
         pixels += line_size;
     }
 
 POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
 }
 
 /* next one assumes that ((line_size % 16) == 0) */
-void put_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
+static void put_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
 {
 POWERPC_PERF_DECLARE(altivec_put_pixels16_xy2_num, 1);
     register int i;
     register vector unsigned char pixelsv1, pixelsv2, pixelsv3, pixelsv4;
     register vector unsigned char blockv, temp1, temp2;
     register vector unsigned short temp3, temp4,
         pixelssum1, pixelssum2, pixelssum3, pixelssum4;
     register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
     register const vector unsigned short vctwo = (const vector unsigned short)vec_splat_u16(2);
 
@@ skipping 52 matching lines @@
         vec_st(blockv, 0, block);
 
         block += line_size;
         pixels += line_size;
     }
 
 POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_xy2_num, 1);
 }
 
 /* next one assumes that ((line_size % 16) == 0) */
-void put_no_rnd_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
+static void put_no_rnd_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
 {
 POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels16_xy2_num, 1);
     register int i;
     register vector unsigned char pixelsv1, pixelsv2, pixelsv3, pixelsv4;
     register vector unsigned char blockv, temp1, temp2;
     register vector unsigned short temp3, temp4,
         pixelssum1, pixelssum2, pixelssum3, pixelssum4;
     register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
     register const vector unsigned short vcone = (const vector unsigned short)vec_splat_u16(1);
     register const vector unsigned short vctwo = (const vector unsigned short)vec_splat_u16(2);
@@ skipping 52 matching lines @@
 
         vec_st(blockv, 0, block);
 
         block += line_size;
         pixels += line_size;
     }
 
 POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
 }
 
-int hadamard8_diff8x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
+static int hadamard8_diff8x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
 POWERPC_PERF_DECLARE(altivec_hadamard8_diff8x8_num, 1);
     int sum;
     register const vector unsigned char vzero =
                             (const vector unsigned char)vec_splat_u8(0);
     register vector signed short temp0, temp1, temp2, temp3, temp4,
                                  temp5, temp6, temp7;
 POWERPC_PERF_START_COUNT(altivec_hadamard8_diff8x8_num, 1);
     {
     register const vector signed short vprod1 =(const vector signed short)
                                                { 1,-1, 1,-1, 1,-1, 1,-1 };
@@ skipping 300 matching lines @@
     vsum = vec_sum4s(vec_abs(line5CS), vsum);
     vsum = vec_sum4s(vec_abs(line6CS), vsum);
     vsum = vec_sum4s(vec_abs(line7CS), vsum);
     vsum = vec_sums(vsum, (vector signed int)vzero);
     vsum = vec_splat(vsum, 3);
     vec_ste(vsum, 0, &sum);
     }
     return sum;
 }
 
-int hadamard8_diff16_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
+static int hadamard8_diff16_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
 POWERPC_PERF_DECLARE(altivec_hadamard8_diff16_num, 1);
     int score;
 POWERPC_PERF_START_COUNT(altivec_hadamard8_diff16_num, 1);
     score = hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
     if (h==16) {
         dst += 8*stride;
         src += 8*stride;
         score += hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
     }
 POWERPC_PERF_STOP_COUNT(altivec_hadamard8_diff16_num, 1);
@@ skipping 17 matching lines @@
         t0 = (vector bool int)vec_and(a, t1);
         t1 = (vector bool int)vec_andc(a, t1);
         a = vec_sub(m, (vector float)t1);
         m = vec_add(m, (vector float)t0);
         vec_stl(a, 0, ang+i);
         vec_stl(m, 0, mag+i);
     }
 }
 
 /* next one assumes that ((line_size % 8) == 0) */
-void avg_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+static void avg_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
 {
 POWERPC_PERF_DECLARE(altivec_avg_pixels8_xy2_num, 1);
     register int i;
     register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
     register vector unsigned char blockv, temp1, temp2, blocktemp;
     register vector unsigned short pixelssum1, pixelssum2, temp3;
 
     register const vector unsigned char vczero = (const vector unsigned char)
                                         vec_splat_u8(0);
     register const vector unsigned short vctwo = (const vector unsigned short)
@@ skipping 78 matching lines @@
     c->put_pixels_tab[1][3] = put_pixels8_xy2_altivec;
     c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_altivec;
     c->put_pixels_tab[0][3] = put_pixels16_xy2_altivec;
     c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_altivec;
 
     c->hadamard8_diff[0] = hadamard8_diff16_altivec;
     c->hadamard8_diff[1] = hadamard8_diff8x8_altivec;
     if (CONFIG_VORBIS_DECODER)
         c->vorbis_inverse_coupling = vorbis_inverse_coupling_altivec;
 }