downmix.c

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.

    for (i = 0; i < 256; i++) {
	common = samples[i + 256] + bias;
	surround = samples[i + 768] + samples[i + 1024];
	samples[i] += common - surround;
	samples[i + 256] = samples[i + 512] + common + surround;
    }
}

static void move2to1 (sample_t * src, sample_t * dest, sample_t bias)
{
    int i;

    for (i = 0; i < 256; i++)
	dest[i] = src[i] + src[i + 256] + bias;
}

static void zero (sample_t * samples)
{
    int i;
    for (i = 0; i < 256; i++)
	samples[i] = 0;
}

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)) {

    case CONVERT (A52_CHANNEL, A52_CHANNEL2):
	memcpy (samples, samples + 256, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_CHANNEL, A52_MONO):
    case CONVERT (A52_STEREO, A52_MONO):
    mix_2to1:
	mix2to1 (samples, samples + 256, bias);
	break;

    case CONVERT (A52_2F1R, A52_MONO):
	if (slev == 0)
	    goto mix_2to1;
    case CONVERT (A52_3F, A52_MONO):
    mix_3to1:
	mix3to1 (samples, bias);
	break;

    case CONVERT (A52_3F1R, A52_MONO):
	if (slev == 0)
	    goto mix_3to1;
    case CONVERT (A52_2F2R, A52_MONO):
	if (slev == 0)
	    goto mix_2to1;
	mix4to1 (samples, bias);
	break;

    case CONVERT (A52_3F2R, A52_MONO):
	if (slev == 0)
	    goto mix_3to1;
	mix5to1 (samples, bias);
	break;

    case CONVERT (A52_MONO, A52_DOLBY):
	memcpy (samples + 256, samples, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_3F, A52_STEREO):
    case CONVERT (A52_3F, A52_DOLBY):
    mix_3to2:
	mix3to2 (samples, bias);
	break;

    case CONVERT (A52_2F1R, A52_STEREO):
	if (slev == 0)
	    break;
	mix21to2 (samples, samples + 256, bias);
	break;

    case CONVERT (A52_2F1R, A52_DOLBY):
	mix21toS (samples, bias);
	break;

    case CONVERT (A52_3F1R, A52_STEREO):
	if (slev == 0)
	    goto mix_3to2;
	mix31to2 (samples, bias);
	break;

    case CONVERT (A52_3F1R, A52_DOLBY):
	mix31toS (samples, bias);
	break;

    case CONVERT (A52_2F2R, A52_STEREO):
	if (slev == 0)
	    break;
	mix2to1 (samples, samples + 512, bias);
	mix2to1 (samples + 256, samples + 768, bias);
	break;

    case CONVERT (A52_2F2R, A52_DOLBY):
	mix22toS (samples, bias);
	break;

    case CONVERT (A52_3F2R, A52_STEREO):
	if (slev == 0)
	    goto mix_3to2;
	mix32to2 (samples, bias);
	break;

    case CONVERT (A52_3F2R, A52_DOLBY):
	mix32toS (samples, bias);
	break;

    case CONVERT (A52_3F1R, A52_3F):
	if (slev == 0)
	    break;
	mix21to2 (samples, samples + 512, bias);
	break;

    case CONVERT (A52_3F2R, A52_3F):
	if (slev == 0)
	    break;
	mix2to1 (samples, samples + 768, bias);
	mix2to1 (samples + 512, samples + 1024, bias);
	break;

    case CONVERT (A52_3F1R, A52_2F1R):
	mix3to2 (samples, bias);
	memcpy (samples + 512, samples + 768, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_2F2R, A52_2F1R):
	mix2to1 (samples + 512, samples + 768, bias);
	break;

    case CONVERT (A52_3F2R, A52_2F1R):
	mix3to2 (samples, bias); //FIXME possible bug? (output doesnt seem to be used)
	move2to1 (samples + 768, samples + 512, bias);
	break;

    case CONVERT (A52_3F2R, A52_3F1R):
	mix2to1 (samples + 768, samples + 1024, bias);
	break;

    case CONVERT (A52_2F1R, A52_2F2R):
	memcpy (samples + 768, samples + 512, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_3F1R, A52_2F2R):
	mix3to2 (samples, bias);
	memcpy (samples + 512, samples + 768, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_3F2R, A52_2F2R):
	mix3to2 (samples, bias);
	memcpy (samples + 512, samples + 768, 256 * sizeof (sample_t));
	memcpy (samples + 768, samples + 1024, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_3F1R, A52_3F2R):
	memcpy (samples + 1024, samples + 768, 256 * sizeof (sample_t));
	break;
    }
}

static void upmix_C (sample_t * samples, int acmod, int output)
{
    switch (CONVERT (acmod, output & A52_CHANNEL_MASK)) {

    case CONVERT (A52_CHANNEL, A52_CHANNEL2):
	memcpy (samples + 256, samples, 256 * sizeof (sample_t));
	break;

    case CONVERT (A52_3F2R, A52_MONO):
	zero (samples + 1024);
    case CONVERT (A52_3F1R, A52_MONO):
    case CONVERT (A52_2F2R, A52_MONO):
	zero (samples + 768);
    case CONVERT (A52_3F, A52_MONO):
    case CONVERT (A52_2F1R, A52_MONO):
	zero (samples + 512);
    case CONVERT (A52_CHANNEL, A52_MONO):
    case CONVERT (A52_STEREO, A52_MONO):
	zero (samples + 256);
	break;

    case CONVERT (A52_3F2R, A52_STEREO):
    case CONVERT (A52_3F2R, A52_DOLBY):
	zero (samples + 1024);
    case CONVERT (A52_3F1R, A52_STEREO):
    case CONVERT (A52_3F1R, A52_DOLBY):
	zero (samples + 768);
    case CONVERT (A52_3F, A52_STEREO):
    case CONVERT (A52_3F, A52_DOLBY):
    mix_3to2:
	memcpy (samples + 512, samples + 256, 256 * sizeof (sample_t));
	zero (samples + 256);
	break;

    case CONVERT (A52_2F2R, A52_STEREO):
    case CONVERT (A52_2F2R, A52_DOLBY):
	zero (samples + 768);
    case CONVERT (A52_2F1R, A52_STEREO):
    case CONVERT (A52_2F1R, A52_DOLBY):
	zero (samples + 512);
	break;

    case CONVERT (A52_3F2R, A52_3F):
	zero (samples + 1024);
    case CONVERT (A52_3F1R, A52_3F):
    case CONVERT (A52_2F2R, A52_2F1R):
	zero (samples + 768);
	break;

    case CONVERT (A52_3F2R, A52_3F1R):
	zero (samples + 1024);
	break;

    case CONVERT (A52_3F2R, A52_2F1R):
	zero (samples + 1024);
    case CONVERT (A52_3F1R, A52_2F1R):
    mix_31to21:
	memcpy (samples + 768, samples + 512, 256 * sizeof (sample_t));
	goto mix_3to2;

    case CONVERT (A52_3F2R, A52_2F2R):
	memcpy (samples + 1024, samples + 768, 256 * sizeof (sample_t));
	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"
		" jnz 1b			\n\t"
	:: "r" (left+256), "r" (right+256), "m" (bias)
	: "%esi"
	);
}

static void mix21toS_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 2048(%0, %%esi), %%xmm0	\n\t"  // surround
		"movaps (%0, %%esi), %%xmm1	\n\t" 
		"movaps 1024(%0, %%esi), %%xmm2	\n\t"
		"addps %%xmm7, %%xmm1		\n\t"
		"addps %%xmm7, %%xmm2		\n\t"
		"subps %%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 mix31to2_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 3072(%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 mix31toS_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"  
		"movaps 3072(%0, %%esi), %%xmm3	\n\t" // surround
		"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"
		"subps %%xmm3, %%xmm1		\n\t"
		"addps %%xmm3, %%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 mix22toS_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 2048(%0, %%esi), %%xmm0	\n\t"  
		"addps 3072(%0, %%esi), %%xmm0	\n\t" // surround
		"movaps (%0, %%esi), %%xmm1	\n\t" 
		"movaps 1024(%0, %%esi), %%xmm2	\n\t"
		"addps %%xmm7, %%xmm1		\n\t"
		"addps %%xmm7, %%xmm2		\n\t"