liba52_changes.diff

自己移植的linux下的流媒体播放器原代码,支持mms协议,支持ftp和http协议.

--- include/a52.h	2005-03-22 19:58:53.000000000 +0100
+++ a52.h	2004-03-19 01:15:49.000000000 +0100
@@ -19,6 +25,9 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 
+#ifndef A52_H
+#define A52_H
+
 #ifndef LIBA52_DOUBLE
 typedef float sample_t;
 #else
@@ -113,3 +122,10 @@
 void a52_dynrng (a52_state_t * state,
 		 sample_t (* call) (sample_t, void *), void * data);
 int a52_block (a52_state_t * state, sample_t * samples);
+
+void* a52_resample_init(uint32_t mm_accel,int flags,int chans);
+extern int (* a52_resample) (float * _f, int16_t * s16);
+
+uint16_t crc16_block(uint8_t *data,uint32_t num_bytes);
+
+#endif /* A52_H */
--- liba52/a52_internal.h	2005-03-22 19:59:35.000000000 +0100
+++ a52_internal.h	2004-03-19 01:15:49.000000000 +0100
@@ -41,11 +43,12 @@
 
 int downmix_init (int input, int flags, sample_t * level,
 		  sample_t clev, sample_t slev);
+void downmix_accel_init(uint32_t mm_accel);
 int downmix_coeff (sample_t * coeff, int acmod, int output, sample_t level,
 		   sample_t clev, sample_t slev);
-void downmix (sample_t * samples, int acmod, int output, sample_t bias,
+extern void (*downmix) (sample_t * samples, int acmod, int output, sample_t bias,
 	      sample_t clev, sample_t slev);
-void upmix (sample_t * samples, int acmod, int output);
+extern void (*upmix) (sample_t * samples, int acmod, int output);
 
 void imdct_init (uint32_t mm_accel);
 extern void (* imdct_256) (sample_t * data, sample_t * delay, sample_t bias);
--- liba52/bitstream.c	2005-03-22 19:59:35.000000000 +0100
+++ bitstream.c	2004-03-19 01:15:49.000000000 +0100
@@ -29,7 +35,12 @@
 
 #define BUFFER_SIZE 4096
 
+#ifdef ALT_BITSTREAM_READER
+int indx=0;
+uint32_t * buffer_start;
+#else
 static uint32_t * buffer_start;
+#endif
 
 uint32_t bits_left;
 uint32_t current_word;
@@ -41,6 +52,9 @@
     align = (int)buf & 3;
     buffer_start = (uint32_t *) (buf - align);
     bits_left = 0;
+#ifdef ALT_BITSTREAM_READER
+    indx=0;
+#endif
     bitstream_get (align * 8);
 }
 
--- liba52/bitstream.h	2005-03-22 19:59:35.000000000 +0100
+++ bitstream.h	2004-03-19 01:15:49.000000000 +0100
@@ -19,6 +25,48 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 
+/* code from ffmpeg/libavcodec */
+#if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC_ == 3 && __GNUC_MINOR__ > 0)
+#    define always_inline __attribute__((always_inline)) inline
+#else
+#    define always_inline inline
+#endif
+
+#if defined(__sparc__) || defined(hpux)
+/*
+ * the alt bitstream reader performs unaligned memory accesses; that doesn't work
+ * on sparc/hpux.  For now, disable ALT_BITSTREAM_READER.
+ */
+#undef	ALT_BITSTREAM_READER
+#else
+// alternative (faster) bitstram reader (reades upto 3 bytes over the end of the input)
+#define ALT_BITSTREAM_READER
+
+/* used to avoid missaligned exceptions on some archs (alpha, ...) */
+#if defined (ARCH_X86) || defined(ARCH_ARMV4L)
+#    define unaligned32(a) (*(uint32_t*)(a))
+#else
+#    ifdef __GNUC__
+static always_inline uint32_t unaligned32(const void *v) {
+    struct Unaligned {
+	uint32_t i;
+    } __attribute__((packed));
+
+    return ((const struct Unaligned *) v)->i;
+}
+#    elif defined(__DECC)
+static inline uint32_t unaligned32(const void *v) {
+    return *(const __unaligned uint32_t *) v;
+}
+#    else
+static inline uint32_t unaligned32(const void *v) {
+    return *(const uint32_t *) v;
+}
+#    endif
+#endif //!ARCH_X86
+
+#endif
+ 
 /* (stolen from the kernel) */
 #ifdef WORDS_BIGENDIAN
 
@@ -29,7 +77,7 @@
 #	if defined (__i386__)
 
 #	define swab32(x) __i386_swab32(x)
-	static inline const uint32_t __i386_swab32(uint32_t x)
+	static always_inline const uint32_t __i386_swab32(uint32_t x)
 	{
 		__asm__("bswap %0" : "=r" (x) : "0" (x));
 		return x;
@@ -37,25 +85,42 @@
 
 #	else
 
-#	define swab32(x)\
-((((uint8_t*)&x)[0] << 24) | (((uint8_t*)&x)[1] << 16) |  \
- (((uint8_t*)&x)[2] << 8)  | (((uint8_t*)&x)[3]))
-
+#	define swab32(x) __generic_swab32(x)
+	static always_inline const uint32_t __generic_swab32(uint32_t x)
+	{
+		return ((((uint8_t*)&x)[0] << 24) | (((uint8_t*)&x)[1] << 16) |
+		 (((uint8_t*)&x)[2] << 8)  | (((uint8_t*)&x)[3]));
+	}
 #	endif
 #endif
 
+#ifdef ALT_BITSTREAM_READER
+extern uint32_t *buffer_start; 
+extern int indx;
+#else
 extern uint32_t bits_left;
 extern uint32_t current_word;
+#endif
 
 void bitstream_set_ptr (uint8_t * buf);
 uint32_t bitstream_get_bh(uint32_t num_bits);
 int32_t bitstream_get_bh_2(uint32_t num_bits);
 
+
 static inline uint32_t 
-bitstream_get(uint32_t num_bits)
+bitstream_get(uint32_t num_bits) // note num_bits is practically a constant due to inlineing
 {
+#ifdef ALT_BITSTREAM_READER
+    uint32_t result= swab32( unaligned32(((uint8_t *)buffer_start)+(indx>>3)) );
+
+    result<<= (indx&0x07);
+    result>>= 32 - num_bits;
+    indx+= num_bits;
+    
+    return result;
+#else
     uint32_t result;
-	
+    
     if(num_bits < bits_left) {
 	result = (current_word << (32 - bits_left)) >> (32 - num_bits);
 	bits_left -= num_bits;
@@ -63,11 +128,30 @@
     }
 
     return bitstream_get_bh(num_bits);
+#endif
+}
+
+static inline void bitstream_skip(int num_bits)
+{
+#ifdef ALT_BITSTREAM_READER
+	indx+= num_bits;
+#else
+	bitstream_get(num_bits);
+#endif
 }
 
 static inline int32_t 
 bitstream_get_2(uint32_t num_bits)
 {
+#ifdef ALT_BITSTREAM_READER
+    int32_t result= swab32( unaligned32(((uint8_t *)buffer_start)+(indx>>3)) );
+
+    result<<= (indx&0x07);
+    result>>= 32 - num_bits;
+    indx+= num_bits;
+        
+    return result;
+#else
     int32_t result;
 	
     if(num_bits < bits_left) {
@@ -77,4 +161,5 @@
     }
 
     return bitstream_get_bh_2(num_bits);
+#endif
 }
--- liba52/downmix.c	2005-03-22 19:59:35.000000000 +0100
+++ downmix.c	2004-04-12 18:42:14.000000000 +0200
@@ -17,18 +23,46 @@
  * You should have received a copy of the GNU General Public License
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ *
+ * SSE optimizations from Michael Niedermayer ([e-mail])
  */
 
 #include "config.h"
 
-#include <inttypes.h>
 #include <string.h>
+#include <inttypes.h>
 
 #include "a52.h"
 #include "a52_internal.h"
+#include "mm_accel.h"
 
 #define CONVERT(acmod,output) (((output) << 3) + (acmod))
 
+
+void (*downmix)(sample_t * samples, int acmod, int output, sample_t bias,
+	      sample_t clev, sample_t slev)= NULL;
+void (*upmix)(sample_t * samples, int acmod, int output)= NULL;
+
+static void downmix_SSE (sample_t * samples, int acmod, int output, sample_t bias,
+	      sample_t clev, sample_t slev);
+static void downmix_3dnow (sample_t * samples, int acmod, int output, sample_t bias,
+	      sample_t clev, sample_t slev);
+static void downmix_C (sample_t * samples, int acmod, int output, sample_t bias,
+	      sample_t clev, sample_t slev);
+static void upmix_MMX (sample_t * samples, int acmod, int output);
+static void upmix_C (sample_t * samples, int acmod, int output);
+
+void downmix_accel_init(uint32_t mm_accel)
+{
+    upmix= upmix_C;
+    downmix= downmix_C;
+#ifdef ARCH_X86    
+    if(mm_accel & MM_ACCEL_X86_MMX) upmix= upmix_MMX;
+    if(mm_accel & MM_ACCEL_X86_SSE) downmix= downmix_SSE;
+    if(mm_accel & MM_ACCEL_X86_3DNOW) downmix= downmix_3dnow;
+#endif
+}
+   
 int downmix_init (int input, int flags, sample_t * level,
 		  sample_t clev, sample_t slev)
 {
@@ -61,7 +95,7 @@
     output = flags & A52_CHANNEL_MASK;
     if (output > A52_DOLBY)
 	return -1;
-
+    
     output = table[output][input & 7];
 
     if ((output == A52_STEREO) &&
@@ -145,7 +179,6 @@
 	    *level *= 1 / (1 + 3 * LEVEL_3DB);
 	    break;
 	}
-
     return output;
 }
 
@@ -440,12 +473,11 @@
 static void zero (sample_t * samples)
 {
     int i;
-
     for (i = 0; i < 256; i++)
 	samples[i] = 0;
 }
 
-void downmix (sample_t * samples, int acmod, int output, sample_t bias,
+static void downmix_C (sample_t * samples, int acmod, int output, sample_t bias,
 	      sample_t clev, sample_t slev)
 {
     switch (CONVERT (acmod, output & A52_CHANNEL_MASK)) {
@@ -557,7 +589,7 @@
 	break;
 
     case CONVERT (A52_3F2R, A52_2F1R):
-	mix3to2 (samples, bias);
+	mix3to2 (samples, bias); //FIXME possible bug? (output doesnt seem to be used)
 	move2to1 (samples + 768, samples + 512, bias);
 	break;
 
@@ -581,12 +613,12 @@
 	break;
 
     case CONVERT (A52_3F1R, A52_3F2R):
-	memcpy (samples + 1027, samples + 768, 256 * sizeof (sample_t));
+	memcpy (samples + 1024, samples + 768, 256 * sizeof (sample_t));
 	break;
     }
 }
 
-void upmix (sample_t * samples, int acmod, int output)
+static void upmix_C (sample_t * samples, int acmod, int output)
 {
     switch (CONVERT (acmod, output & A52_CHANNEL_MASK)) {
 
@@ -651,3 +683,1137 @@
 	goto mix_31to21;
     }
 }
+
+#ifdef ARCH_X86
+static void mix2to1_SSE (sample_t * dest, sample_t * src, sample_t bias)
+{
+	asm volatile(
+	"movlps %2, %%xmm7		\n\t"
+	"shufps $0x00, %%xmm7, %%xmm7	\n\t"
+	"movl $-1024, %%esi		\n\t"
+	".balign 16\n\t"
+	"1:				\n\t"
+	"movaps (%0, %%esi), %%xmm0	\n\t" 
+	"movaps 16(%0, %%esi), %%xmm1	\n\t" 
+	"addps (%1, %%esi), %%xmm0	\n\t" 
+	"addps 16(%1, %%esi), %%xmm1	\n\t" 
+	"addps %%xmm7, %%xmm0		\n\t"
+	"addps %%xmm7, %%xmm1		\n\t"
+	"movaps %%xmm0, (%1, %%esi)	\n\t"
+	"movaps %%xmm1, 16(%1, %%esi)	\n\t"
+	"addl $32, %%esi		\n\t"
+	" jnz 1b			\n\t"
+	:: "r" (src+256), "r" (dest+256), "m" (bias)
+	: "%esi"
+	);
+}
+
+static void mix3to1_SSE (sample_t * samples, sample_t bias)
+{
+	asm volatile(
+	"movlps %1, %%xmm7		\n\t"
+	"shufps $0x00, %%xmm7, %%xmm7	\n\t"
+	"movl $-1024, %%esi		\n\t"
+	".balign 16\n\t"
+	"1:				\n\t"
+	"movaps (%0, %%esi), %%xmm0	\n\t" 
+	"movaps 1024(%0, %%esi), %%xmm1	\n\t" 
+	"addps 2048(%0, %%esi), %%xmm0	\n\t" 
+	"addps %%xmm7, %%xmm1		\n\t"
+	"addps %%xmm1, %%xmm0		\n\t"
+	"movaps %%xmm0, (%0, %%esi)	\n\t"
+	"addl $16, %%esi		\n\t"
+	" jnz 1b			\n\t"
+	:: "r" (samples+256), "m" (bias)
+	: "%esi"
+	);
+}
+
+static void mix4to1_SSE (sample_t * samples, sample_t bias)
+{
+	asm volatile(
+	"movlps %1, %%xmm7		\n\t"
+	"shufps $0x00, %%xmm7, %%xmm7	\n\t"
+	"movl $-1024, %%esi		\n\t"
+	".balign 16\n\t"
+	"1:				\n\t"
+	"movaps (%0, %%esi), %%xmm0	\n\t" 
+	"movaps 1024(%0, %%esi), %%xmm1	\n\t" 
+	"addps 2048(%0, %%esi), %%xmm0	\n\t" 
+	"addps 3072(%0, %%esi), %%xmm1	\n\t" 
+	"addps %%xmm7, %%xmm0		\n\t"
+	"addps %%xmm1, %%xmm0		\n\t"
+	"movaps %%xmm0, (%0, %%esi)	\n\t"
+	"addl $16, %%esi		\n\t"
+	" jnz 1b			\n\t"
+	:: "r" (samples+256), "m" (bias)
+	: "%esi"
+	);
+}
+
+static void mix5to1_SSE (sample_t * samples, sample_t bias)
+{
+	asm volatile(
+	"movlps %1, %%xmm7		\n\t"
+	"shufps $0x00, %%xmm7, %%xmm7	\n\t"
+	"movl $-1024, %%esi		\n\t"
+	".balign 16\n\t"
+	"1:				\n\t"
+	"movaps (%0, %%esi), %%xmm0	\n\t" 
+	"movaps 1024(%0, %%esi), %%xmm1	\n\t" 
+	"addps 2048(%0, %%esi), %%xmm0	\n\t" 
+	"addps 3072(%0, %%esi), %%xmm1	\n\t" 
+	"addps %%xmm7, %%xmm0		\n\t"
+	"addps 4096(%0, %%esi), %%xmm1	\n\t" 
+	"addps %%xmm1, %%xmm0		\n\t"
+	"movaps %%xmm0, (%0, %%esi)	\n\t"
+	"addl $16, %%esi		\n\t"
+	" jnz 1b			\n\t"
+	:: "r" (samples+256), "m" (bias)
+	: "%esi"
+	);
+}
+
+static void mix3to2_SSE (sample_t * samples, sample_t bias)
+{
+	asm volatile(
+	"movlps %1, %%xmm7		\n\t"
+	"shufps $0x00, %%xmm7, %%xmm7	\n\t"
+	"movl $-1024, %%esi		\n\t"
+	".balign 16\n\t"
+	"1:				\n\t"
+	"movaps 1024(%0, %%esi), %%xmm0	\n\t" 
+	"addps %%xmm7, %%xmm0		\n\t" //common
+	"movaps (%0, %%esi), %%xmm1	\n\t" 
+	"movaps 2048(%0, %%esi), %%xmm2	\n\t"
+	"addps %%xmm0, %%xmm1		\n\t"
+	"addps %%xmm0, %%xmm2		\n\t"
+	"movaps %%xmm1, (%0, %%esi)	\n\t"
+	"movaps %%xmm2, 1024(%0, %%esi)	\n\t"
+	"addl $16, %%esi		\n\t"
+	" jnz 1b			\n\t"
+	:: "r" (samples+256), "m" (bias)
+	: "%esi"
+	);
+}
+
+static void mix21to2_SSE (sample_t * left, sample_t * right, sample_t bias)
+{
+	asm volatile(
+		"movlps %2, %%xmm7		\n\t"
+		"shufps $0x00, %%xmm7, %%xmm7	\n\t"
+		"movl $-1024, %%esi		\n\t"
+		".balign 16\n\t"
+		"1:				\n\t"
+		"movaps 1024(%1, %%esi), %%xmm0	\n\t" 
+		"addps %%xmm7, %%xmm0		\n\t" //common
+		"movaps (%0, %%esi), %%xmm1	\n\t" 
+		"movaps (%1, %%esi), %%xmm2	\n\t"
+		"addps %%xmm0, %%xmm1		\n\t"
+		"addps %%xmm0, %%xmm2		\n\t"
+		"movaps %%xmm1, (%0, %%esi)	\n\t"
+		"movaps %%xmm2, (%1, %%esi)	\n\t"
+		"addl $16, %%esi		\n\t"