From aab34ca0c5bbffd06aaa3718151ef1662860da3a Mon Sep 17 00:00:00 2001
From: Michael Niedermayer <michaelni@gmx.at>
Date: Sat, 28 Feb 2004 15:03:53 +0000
Subject: Metrowerks CodeWarrior patches by (John Dalgliesh <johnd at defyne
 dot org>)

Originally committed as revision 2823 to svn://svn.ffmpeg.org/ffmpeg/trunk
---
 libavcodec/ppc/dsputil_altivec.c   | 60 +++++++++++++++++++-------------------
 libavcodec/ppc/dsputil_ppc.c       |  8 +++--
 libavcodec/ppc/gcc_fixes.h         | 12 +++++++-
 libavcodec/ppc/gmc_altivec.c       |  4 +--
 libavcodec/ppc/idct_altivec.c      |  8 +++--
 libavcodec/ppc/mpegvideo_altivec.c | 31 ++++++++++----------
 6 files changed, 71 insertions(+), 52 deletions(-)

diff --git a/libavcodec/ppc/dsputil_altivec.c b/libavcodec/ppc/dsputil_altivec.c
index d1a2943656..633cae68b0 100644
--- a/libavcodec/ppc/dsputil_altivec.c
+++ b/libavcodec/ppc/dsputil_altivec.c
@@ -49,7 +49,7 @@ int sad16_x2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
+    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;
@@ -96,7 +96,7 @@ int sad16_y2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
+    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;
@@ -157,8 +157,8 @@ int sad16_xy2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int
     int i;
     int s __attribute__((aligned(16)));
     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);
+    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;
@@ -257,7 +257,7 @@ int sad16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
+    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;
@@ -299,7 +299,7 @@ int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
+    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;
@@ -344,7 +344,7 @@ int pix_norm1_altivec(uint8_t *pix, int line_size)
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
+    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;
@@ -380,7 +380,7 @@ int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
+    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;
@@ -436,7 +436,7 @@ int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int i;
     int s __attribute__((aligned(16)));
-    const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
+    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;
@@ -480,7 +480,7 @@ int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 
 int pix_sum_altivec(uint8_t * pix, int line_size)
 {
-    const vector unsigned int zero = (const vector unsigned int)vec_splat_u32(0);
+    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;
@@ -515,7 +515,7 @@ void get_pixels_altivec(DCTELEM *restrict block, const uint8_t *pixels, int line
 {
     int i;
     vector unsigned char perm, bytes, *pixv;
-    const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
+    const_vector unsigned char zero = (const_vector unsigned char)vec_splat_u8(0);
     vector signed short shorts;
 
     for(i=0;i<8;i++)
@@ -542,7 +542,7 @@ void diff_pixels_altivec(DCTELEM *restrict block, const uint8_t *s1,
 {
     int i;
     vector unsigned char perm, bytes, *pixv;
-    const vector unsigned char zero = (const vector unsigned char)vec_splat_u8(0);
+    const_vector unsigned char zero = (const_vector unsigned char)vec_splat_u8(0);
     vector signed short shorts1, shorts2;
 
     for(i=0;i<4;i++)
@@ -654,10 +654,10 @@ POWERPC_PERF_DECLARE(altivec_put_pixels16_num, 1);
 POWERPC_PERF_START_COUNT(altivec_put_pixels16_num, 1);
 
     for(i=0; i<h; i++) {
-      *((uint32_t*)(block )) = (((const struct unaligned_32 *) (pixels))->l);
-      *((uint32_t*)(block+4)) = (((const struct unaligned_32 *) (pixels+4))->l);
-      *((uint32_t*)(block+8)) = (((const struct unaligned_32 *) (pixels+8))->l);
-      *((uint32_t*)(block+12)) = (((const struct unaligned_32 *) (pixels+12))->l);
+      *((uint32_t*)(block)) = LD32(pixels);
+      *((uint32_t*)(block+4)) = LD32(pixels+4);
+      *((uint32_t*)(block+8)) = LD32(pixels+8);
+      *((uint32_t*)(block+12)) = LD32(pixels+12);
       pixels+=line_size;
       block +=line_size;
     }
@@ -729,10 +729,10 @@ POWERPC_PERF_DECLARE(altivec_avg_pixels16_num, 1);
 POWERPC_PERF_START_COUNT(altivec_avg_pixels16_num, 1);
 
     for(i=0; i<h; i++) {
-      op_avg(*((uint32_t*)(block)),(((const struct unaligned_32 *)(pixels))->l));
-      op_avg(*((uint32_t*)(block+4)),(((const struct unaligned_32 *)(pixels+4))->l));
-      op_avg(*((uint32_t*)(block+8)),(((const struct unaligned_32 *)(pixels+8))->l));
-      op_avg(*((uint32_t*)(block+12)),(((const struct unaligned_32 *)(pixels+12))->l));
+      op_avg(*((uint32_t*)(block)),LD32(pixels));
+      op_avg(*((uint32_t*)(block+4)),LD32(pixels+4));
+      op_avg(*((uint32_t*)(block+8)),LD32(pixels+8));
+      op_avg(*((uint32_t*)(block+12)),LD32(pixels+12));
       pixels+=line_size;
       block +=line_size;
     }
@@ -878,8 +878,8 @@ POWERPC_PERF_STOP_COUNT(altivec_put_pixels8_xy2_num, 1);
      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);
+   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);
    temp2 = vec_ld(16, pixels);
@@ -993,9 +993,9 @@ POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
      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);
+   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);
    
    temp1 = vec_ld(0, pixels);
    temp2 = vec_ld(16, pixels);
@@ -1109,8 +1109,8 @@ POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_xy2_num, 1);
    register vector unsigned short
      pixelssum1, pixelssum2, temp3,
      pixelssum3, pixelssum4, temp4;
-   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);
+   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);
 
 POWERPC_PERF_START_COUNT(altivec_put_pixels16_xy2_num, 1);
  
@@ -1230,9 +1230,9 @@ POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
    register vector unsigned short
      pixelssum1, pixelssum2, temp3,
      pixelssum3, pixelssum4, temp4;
-   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);
+   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);
 
 POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
  
diff --git a/libavcodec/ppc/dsputil_ppc.c b/libavcodec/ppc/dsputil_ppc.c
index 9882e401f9..691b7725b1 100644
--- a/libavcodec/ppc/dsputil_ppc.c
+++ b/libavcodec/ppc/dsputil_ppc.c
@@ -69,13 +69,13 @@ static unsigned char* perfname[] = {
 void powerpc_display_perf_report(void)
 {
   int i, j;
-  fprintf(stderr, "PowerPC performance report\n Values are from the PMC registers, and represent whatever the registers are set to record.\n");
+  av_log(NULL, AV_LOG_INFO, "PowerPC performance report\n Values are from the PMC registers, and represent whatever the registers are set to record.\n");
   for(i = 0 ; i < powerpc_perf_total ; i++)
   {
     for (j = 0; j < POWERPC_NUM_PMC_ENABLED ; j++)
       {
 	if (perfdata[j][i][powerpc_data_num] != (unsigned long long)0)
-	  fprintf(stderr,
+	  av_log(NULL, AV_LOG_INFO,
 		  " Function \"%s\" (pmc%d):\n\tmin: %llu\n\tmax: %llu\n\tavg: %1.2lf (%llu)\n",
 		  perfname[i],
 		  j+1,
@@ -129,7 +129,11 @@ POWERPC_PERF_START_COUNT(powerpc_clear_blocks_dcbz32, 1);
       i += 16;
     }
     for ( ; i < sizeof(DCTELEM)*6*64 ; i += 32) {
+#ifndef __MWERKS__
       asm volatile("dcbz %0,%1" : : "b" (blocks), "r" (i) : "memory");
+#else
+      __dcbz( blocks, i );
+#endif
     }
     if (misal) {
       ((unsigned long*)blocks)[188] = 0L;
diff --git a/libavcodec/ppc/gcc_fixes.h b/libavcodec/ppc/gcc_fixes.h
index a8e92cb2f1..13d4ff12e7 100644
--- a/libavcodec/ppc/gcc_fixes.h
+++ b/libavcodec/ppc/gcc_fixes.h
@@ -12,7 +12,11 @@
 #endif
 
 #ifdef CONFIG_DARWIN
-#define AVV(x...) (x)
+# ifndef __MWERKS__
+#  define AVV(x...) (x)
+# else
+#  define AVV
+# endif
 #else
 #define AVV(x...) {x}
 #if (__GNUC__ * 100 + __GNUC_MINOR__ < 303)  
@@ -81,4 +85,10 @@ __ch (__bin_args_eq (vector unsigned int, (a1), vector unsigned int, (a2)), \
 
 #endif /* CONFIG_DARWIN */
 
+#ifndef __MWERKS__
+#define const_vector const vector
+#else
+#define const_vector vector
+#endif
+
 #endif /* _GCC_FIXES_ */
diff --git a/libavcodec/ppc/gmc_altivec.c b/libavcodec/ppc/gmc_altivec.c
index 671ee110aa..3448216859 100644
--- a/libavcodec/ppc/gmc_altivec.c
+++ b/libavcodec/ppc/gmc_altivec.c
@@ -69,8 +69,8 @@ POWERPC_PERF_STOP_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
         (   x16)*(   y16), /* D */
         0, 0, 0, 0         /* padding */
       };
-    register const vector unsigned char vczero = (const vector unsigned char)vec_splat_u8(0);
-    register const vector unsigned short vcsr8 = (const vector unsigned short)vec_splat_u16(8);
+    register const_vector unsigned char vczero = (const_vector unsigned char)vec_splat_u8(0);
+    register const_vector unsigned short vcsr8 = (const_vector unsigned short)vec_splat_u16(8);
     register vector unsigned char dstv, dstv2, src_0, src_1, srcvA, srcvB, srcvC, srcvD;
     register vector unsigned short Av, Bv, Cv, Dv, rounderV, tempA, tempB, tempC, tempD;
     int i;
diff --git a/libavcodec/ppc/idct_altivec.c b/libavcodec/ppc/idct_altivec.c
index d821ecd223..3445adaddf 100644
--- a/libavcodec/ppc/idct_altivec.c
+++ b/libavcodec/ppc/idct_altivec.c
@@ -44,6 +44,7 @@
 #include "dsputil_altivec.h"
 
 #define vector_s16_t vector signed short
+#define const_vector_s16_t const_vector signed short
 #define vector_u16_t vector unsigned short
 #define vector_s8_t vector signed char
 #define vector_u8_t vector unsigned char
@@ -155,7 +156,7 @@
     vx7 = vec_sra (vy7, shift);
 
 
-static const vector_s16_t constants[5] = {
+static const_vector_s16_t constants[5] = {
     (vector_s16_t) AVV(23170, 13573, 6518, 21895, -23170, -21895, 32, 31),
     (vector_s16_t) AVV(16384, 22725, 21407, 19266, 16384, 19266, 21407, 22725),
     (vector_s16_t) AVV(22725, 31521, 29692, 26722, 22725, 26722, 29692, 31521),
@@ -174,8 +175,9 @@ POWERPC_PERF_STOP_COUNT(altivec_idct_put_num, 1);
 #else /* ALTIVEC_USE_REFERENCE_C_CODE */
     vector_u8_t tmp;
 
+#ifdef POWERPC_PERFORMANCE_REPORT
 POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1);
-
+#endif
     IDCT
 
 #define COPY(dest,src)						\
@@ -211,7 +213,9 @@ POWERPC_PERF_STOP_COUNT(altivec_idct_add_num, 1);
     vector_u8_t perm1;
     vector_u8_t p0, p1, p;
 
+#ifdef POWERPC_PERFORMANCE_REPORT
 POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1);
+#endif
 
     IDCT
 
diff --git a/libavcodec/ppc/mpegvideo_altivec.c b/libavcodec/ppc/mpegvideo_altivec.c
index ae3170d91a..51b387792c 100644
--- a/libavcodec/ppc/mpegvideo_altivec.c
+++ b/libavcodec/ppc/mpegvideo_altivec.c
@@ -107,7 +107,9 @@ int dct_quantize_altivec(MpegEncContext* s,
     int lastNonZero;
     vector float row0, row1, row2, row3, row4, row5, row6, row7;
     vector float alt0, alt1, alt2, alt3, alt4, alt5, alt6, alt7;
-    const vector float zero = (const vector float)FOUROF(0.);
+    const_vector float zero = (const_vector float)FOUROF(0.);
+    // used after quantise step
+    int oldBaseValue = 0;
 
     // Load the data into the row/alt vectors
     {
@@ -284,9 +286,6 @@ int dct_quantize_altivec(MpegEncContext* s,
         }
     }
 
-    // used after quantise step
-    int oldBaseValue = 0;
-
     // perform the quantise step, using the floating point data
     // still in the row/alt registers
     {
@@ -414,21 +413,23 @@ int dct_quantize_altivec(MpegEncContext* s,
             data7 = vec_max(vec_min(data7, max_q), min_q);
         }
 
+        {
         vector bool char zero_01, zero_23, zero_45, zero_67;
         vector signed char scanIndices_01, scanIndices_23, scanIndices_45, scanIndices_67;
         vector signed char negOne = vec_splat_s8(-1);
         vector signed char* scanPtr =
                 (vector signed char*)(s->intra_scantable.inverse);
+        signed char lastNonZeroChar;
 
         // Determine the largest non-zero index.
-        zero_01 = vec_pack(vec_cmpeq(data0, (vector short)zero),
-                vec_cmpeq(data1, (vector short)zero));
-        zero_23 = vec_pack(vec_cmpeq(data2, (vector short)zero),
-                vec_cmpeq(data3, (vector short)zero));
-        zero_45 = vec_pack(vec_cmpeq(data4, (vector short)zero),
-                vec_cmpeq(data5, (vector short)zero));
-        zero_67 = vec_pack(vec_cmpeq(data6, (vector short)zero),
-                vec_cmpeq(data7, (vector short)zero));
+        zero_01 = vec_pack(vec_cmpeq(data0, (vector signed short)zero),
+                vec_cmpeq(data1, (vector signed short)zero));
+        zero_23 = vec_pack(vec_cmpeq(data2, (vector signed short)zero),
+                vec_cmpeq(data3, (vector signed short)zero));
+        zero_45 = vec_pack(vec_cmpeq(data4, (vector signed short)zero),
+                vec_cmpeq(data5, (vector signed short)zero));
+        zero_67 = vec_pack(vec_cmpeq(data6, (vector signed short)zero),
+                vec_cmpeq(data7, (vector signed short)zero));
 
         // 64 biggest values
         scanIndices_01 = vec_sel(scanPtr[0], negOne, zero_01);
@@ -461,7 +462,6 @@ int dct_quantize_altivec(MpegEncContext* s,
 
         scanIndices_01 = vec_splat(scanIndices_01, 0);
 
-        signed char lastNonZeroChar;
 
         vec_ste(scanIndices_01, 0, &lastNonZeroChar);
 
@@ -484,6 +484,7 @@ int dct_quantize_altivec(MpegEncContext* s,
         vec_st(data5, 80, data);
         vec_st(data6, 96, data);
         vec_st(data7, 112, data);
+        }
     }
 
     // special handling of block[0]
@@ -562,7 +563,7 @@ POWERPC_PERF_START_COUNT(altivec_dct_unquantize_h263_num, 1);
     }
 #else /* ALTIVEC_USE_REFERENCE_C_CODE */
     {
-      register const vector short vczero = (const vector short)vec_splat_s16(0);
+      register const_vector signed short vczero = (const_vector signed short)vec_splat_s16(0);
       short __attribute__ ((aligned(16))) qmul8[] =
           {
             qmul, qmul, qmul, qmul,
@@ -578,7 +579,7 @@ POWERPC_PERF_START_COUNT(altivec_dct_unquantize_h263_num, 1);
             -qadd, -qadd, -qadd, -qadd,
             -qadd, -qadd, -qadd, -qadd
           };
-      register vector short blockv, qmulv, qaddv, nqaddv, temp1;
+      register vector signed short blockv, qmulv, qaddv, nqaddv, temp1;
       register vector bool short blockv_null, blockv_neg;
       register short backup_0 = block[0];
       register int j = 0;
-- 
cgit v1.2.3