seal 0.1.0

data/mpg123/src/libmpg123/synth_mono.h

@@ -0,0 +1,64 @@
+/*
+	monosynth.h: generic mono related synth functions 
+
+	copyright 1995-2008 by the mpg123 project - free software under the terms of the LGPL 2.1
+	see COPYING and AUTHORS files in distribution or http://mpg123.org
+	initially written by Michael Hipp, generalized by Thomas Orgis
+
+	This header is used multiple times to create different variants of these functions.
+	See decode.c and synth.h .
+	Hint: BLOCK, MONO_NAME, MONO2STEREO_NAME, SYNTH_NAME and SAMPLE_T do vary.
+
+	Thomas looked closely at the decode_1to1, decode_2to1 and decode_4to1 contents, seeing that they are too similar to be separate files.
+	This is what resulted...
+
+	Reason to separate this from synth.h:
+	There are decoders that have a special synth_1to1 but still can use these generic derivations for the mono stuff.
+	It generally makes a good deal of sense to set SYNTH_NAME to opt_synth_1to1(fr) (or opt_synth_2to1(fr), etc.).
+*/
+
+/* Mono synth, wrapping over SYNTH_NAME */
+int MONO_NAME(real *bandPtr, mpg123_handle *fr)
+{
+	SAMPLE_T samples_tmp[BLOCK];
+	SAMPLE_T *tmp1 = samples_tmp;
+	int i,ret;
+
+	/* save buffer stuff, trick samples_tmp into there, decode, restore */
+	unsigned char *samples = fr->buffer.data;
+	int pnt = fr->buffer.fill;
+	fr->buffer.data = (unsigned char*) samples_tmp;
+	fr->buffer.fill = 0;
+	ret = SYNTH_NAME(bandPtr, 0, fr, 0); /* decode into samples_tmp */
+	fr->buffer.data = samples; /* restore original value */
+
+	/* now append samples from samples_tmp */
+	samples += pnt; /* just the next mem in frame buffer */
+	for(i=0;i<(BLOCK/2);i++)
+	{
+		*( (SAMPLE_T *)samples) = *tmp1;
+		samples += sizeof(SAMPLE_T);
+		tmp1 += 2;
+	}
+	fr->buffer.fill = pnt + (BLOCK/2)*sizeof(SAMPLE_T);
+
+	return ret;
+}
+
+/* Mono to stereo synth, wrapping over SYNTH_NAME */
+int MONO2STEREO_NAME(real *bandPtr, mpg123_handle *fr)
+{
+	int i,ret;
+	unsigned char *samples = fr->buffer.data;
+
+	ret = SYNTH_NAME(bandPtr,0,fr,1);
+	samples += fr->buffer.fill - BLOCK*sizeof(SAMPLE_T);
+
+	for(i=0;i<(BLOCK/2);i++)
+	{
+		((SAMPLE_T *)samples)[1] = ((SAMPLE_T *)samples)[0];
+		samples+=2*sizeof(SAMPLE_T);
+	}
+
+	return ret;
+}

data/mpg123/src/libmpg123/synth_neon.S

@@ -0,0 +1,123 @@
+/*
+	synth_neon: ARM NEON optimized synth
+
+	copyright 1995-2010 by the mpg123 project - free software under the terms of the LGPL 2.1
+	see COPYING and AUTHORS files in distribution or http://mpg123.org
+	initially written by Taihei Monma
+*/
+
+#include "mangle.h"
+
+#define WINDOW r0
+#define B0 r1
+#define SAMPLES r2
+
+/*
+	int synth_1to1_neon_asm(short *window, short *b0, short *samples, int bo1);
+	return value: number of clipped samples
+*/
+
+	.text
+	.globl ASM_NAME(synth_1to1_neon_asm)
+	ALIGN4
+ASM_NAME(synth_1to1_neon_asm):
+	push		{r4-r5, lr}
+	vpush		{q4-q7}
+
+	add			WINDOW, WINDOW, #32
+	sub			WINDOW, WINDOW, r3, lsl #1
+
+	mov			r3, #4
+	mov			r4, #64
+.Loop_start_1:
+	vld1.16		{d0-d3}, [WINDOW], r4
+	vld1.16		{d4-d7}, [B0, :128]!
+	vld1.16		{d8-d11}, [WINDOW], r4
+	vswp		d1, d4
+	vld1.16		{d12-d15}, [B0, :128]!
+	vld1.16		{d16-d19}, [WINDOW], r4
+	vld1.16		{d20-d23}, [B0, :128]!
+	vswp		d9, d12
+	vld1.16		{d24-d27}, [WINDOW], r4
+	vld1.16		{d28-d31}, [B0, :128]!
+	vswp		d17, d20
+	vswp		d25, d28
+	vmull.s16	q0, d0, d1
+	vmull.s16	q4, d8, d9
+	vmull.s16	q8, d16, d17
+	vmull.s16	q12, d24, d25
+	vmlal.s16	q0, d4, d5
+	vmlal.s16	q4, d12, d13
+	vmlal.s16	q8, d20, d21
+	vmlal.s16	q12, d28, d29
+	vmlal.s16	q0, d2, d6
+	vmlal.s16	q4, d10, d14
+	vmlal.s16	q8, d18, d22
+	vmlal.s16	q12, d26, d30
+	vmlal.s16	q0, d3, d7
+	vmlal.s16	q4, d11, d15
+	vmlal.s16	q8, d19, d23
+	vmlal.s16	q12, d27, d31
+	vpadd.i32	d0, d0, d1
+	vpadd.i32	d8, d8, d9
+	vpadd.i32	d16, d16, d17
+	vpadd.i32	d24, d24, d25
+	vpadd.i32	d0, d0, d8
+	vpadd.i32	d1, d16, d24
+
+	vld2.16		{d2,d3}, [SAMPLES]
+	vqshrn.s32	d1, q0, #13
+	vst2.16		{d1,d3}, [SAMPLES]!
+
+	subs		r3, r3, #1
+	bne			.Loop_start_1
+
+	mov			r3, #4
+	mov			r5, #-32
+.Loop_start_2:
+	vld1.16		{d0-d3}, [WINDOW], r4
+	vld1.16		{d4-d7}, [B0, :128], r5
+	vld1.16		{d8-d11}, [WINDOW], r4
+	vswp		d1, d4
+	vld1.16		{d12-d15}, [B0, :128], r5
+	vld1.16		{d16-d19}, [WINDOW], r4
+	vld1.16		{d20-d23}, [B0, :128], r5
+	vswp		d9, d12
+	vld1.16		{d24-d27}, [WINDOW], r4
+	vld1.16		{d28-d31}, [B0, :128], r5
+	vswp		d17, d20
+	vswp		d25, d28
+	vmull.s16	q0, d0, d1
+	vmull.s16	q4, d8, d9
+	vmull.s16	q8, d16, d17
+	vmull.s16	q12, d24, d25
+	vmlal.s16	q0, d4, d5
+	vmlal.s16	q4, d12, d13
+	vmlal.s16	q8, d20, d21
+	vmlal.s16	q12, d28, d29
+	vmlal.s16	q0, d2, d6
+	vmlal.s16	q4, d10, d14
+	vmlal.s16	q8, d18, d22
+	vmlal.s16	q12, d26, d30
+	vmlal.s16	q0, d3, d7
+	vmlal.s16	q4, d11, d15
+	vmlal.s16	q8, d19, d23
+	vmlal.s16	q12, d27, d31
+	vpadd.i32	d0, d0, d1
+	vpadd.i32	d8, d8, d9
+	vpadd.i32	d16, d16, d17
+	vpadd.i32	d24, d24, d25
+	vpadd.i32	d0, d0, d8
+	vpadd.i32	d1, d16, d24
+
+	vld2.16		{d2,d3}, [SAMPLES]
+	vqshrn.s32	d1, q0, #13
+	vst2.16		{d1,d3}, [SAMPLES]!
+
+	subs		r3, r3, #1
+	bne			.Loop_start_2
+
+	mov			r0, #0
+
+	vpop		{q4-q7}
+	pop			{r4-r5, pc}

data/mpg123/src/libmpg123/synth_neon_accurate.S

@@ -0,0 +1,173 @@
+/*
+	synth_neon_accurate: ARM NEON optimized synth (MPEG compliant 16-bit output version)
+
+	copyright 1995-2010 by the mpg123 project - free software under the terms of the LGPL 2.1
+	see COPYING and AUTHORS files in distribution or http://mpg123.org
+	initially written by Taihei Monma
+*/
+
+#include "mangle.h"
+
+#define WINDOW r0
+#define B0 r1
+#define SAMPLES r2
+
+/*
+	int synth_1to1_real_neon_accurate_asm(real *window, real *b0, real *samples, int bo1);
+	return value: number of clipped samples (0)
+*/
+
+	.text
+	.globl ASM_NAME(synth_1to1_neon_accurate_asm)
+ASM_NAME(synth_1to1_neon_accurate_asm):
+	push		{r4-r6, lr}
+	vpush		{q4-q7}
+	mov			r6, sp
+	sub			sp, sp, #16
+	bic			sp, #0xff
+	
+	add			WINDOW, WINDOW, #64
+	sub			WINDOW, WINDOW, r3, lsl #2
+
+	mov			r3, #4
+	mov			r4, #128
+	mov			r5, #64
+.Loop_start_1:
+	vld1.32		{q0,q1}, [WINDOW], r4
+	vld1.32		{q2,q3}, [WINDOW], r4
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW]
+	sub			WINDOW, WINDOW, #352
+	vld1.32		{q8,q9}, [B0, :128], r5
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128]
+	vswp		q1, q4
+	vswp		q3, q6
+	sub			B0, B0, #160
+	vmul.f32	q0, q0, q8
+	vmul.f32	q2, q2, q10
+	vmul.f32	q1, q1, q12
+	vmul.f32	q3, q3, q14
+	vmla.f32	q0, q4, q9
+	vmla.f32	q2, q6, q11
+	vmla.f32	q1, q5, q13
+	vmla.f32	q3, q7, q15
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW], r4
+	vld1.32		{q8,q9}, [WINDOW], r4
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128], r5
+	vswp		q5, q6
+	vswp		q11, q12
+	vmla.f32	q0, q4, q10
+	vmla.f32	q2, q5, q11
+	vmla.f32	q1, q8, q14
+	vld1.32		{q4,q5}, [WINDOW]
+	vld1.32		{q10,q11}, [B0, :128]!
+	add			WINDOW, WINDOW, #96
+	vmla.f32	q3, q4, q10
+	vmla.f32	q0, q6, q12
+	vmla.f32	q2, q7, q13
+	vmla.f32	q1, q9, q15
+	vmla.f32	q3, q5, q11
+	vmov.i32	q4, #0x4b000000
+	vmvn.i32	q5, #0xb9000000
+	vorr.i32	q4, #0x00400000
+	vpadd.f32	d0, d0, d1
+	vpadd.f32	d4, d4, d5
+	vpadd.f32	d2, d2, d3
+	vpadd.f32	d6, d6, d7
+	vld1.32		{q6}, [sp, :128]
+	vpadd.f32	d0, d0, d4
+	vpadd.f32	d1, d2, d6
+
+	vadd.f32	q3, q0, q4
+	vacgt.f32	q5, q0, q5
+	vld2.16		{d4,d5}, [SAMPLES]
+	vshl.i32	q3, q3, #10
+	vqshrn.s32	d3, q3, #10
+	vshr.u32	q5, q5, #31
+	vst2.16		{d3,d5}, [SAMPLES]!
+	vadd.i32	q5, q5, q6
+	vst1.32		{q5}, [sp, :128]
+
+	subs		r3, r3, #1
+	bne			.Loop_start_1
+
+	mov			r3, #4
+	mov			r5, #-64
+.Loop_start_2:
+	vld1.32		{q0,q1}, [WINDOW], r4
+	vld1.32		{q2,q3}, [WINDOW], r4
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW]
+	sub			WINDOW, WINDOW, #352
+	vld1.32		{q8,q9}, [B0, :128], r5
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128]
+	vswp		q1, q4
+	vswp		q3, q6
+	add			B0, B0, #224
+	vmul.f32	q0, q0, q8
+	vmul.f32	q2, q2, q10
+	vmul.f32	q1, q1, q12
+	vmul.f32	q3, q3, q14
+	vmla.f32	q0, q4, q9
+	vmla.f32	q2, q6, q11
+	vmla.f32	q1, q5, q13
+	vmla.f32	q3, q7, q15
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW], r4
+	vld1.32		{q8,q9}, [WINDOW], r4
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128], r5
+	vswp		q5, q6
+	vswp		q11, q12
+	vmla.f32	q0, q4, q10
+	vmla.f32	q2, q5, q11
+	vmla.f32	q1, q8, q14
+	vld1.32		{q4,q5}, [WINDOW]
+	vld1.32		{q10,q11}, [B0, :128]
+	add			WINDOW, WINDOW, #96
+	sub			B0, B0, #96
+	vmla.f32	q3, q4, q10
+	vmla.f32	q0, q6, q12
+	vmla.f32	q2, q7, q13
+	vmla.f32	q1, q9, q15
+	vmla.f32	q3, q5, q11
+	vmov.i32	q4, #0x4b000000
+	vmvn.i32	q5, #0xb9000000
+	vorr.i32	q4, #0x00400000
+	vpadd.f32	d0, d0, d1
+	vpadd.f32	d4, d4, d5
+	vpadd.f32	d2, d2, d3
+	vpadd.f32	d6, d6, d7
+	vld1.32		{q6}, [sp, :128]
+	vpadd.f32	d0, d0, d4
+	vpadd.f32	d1, d2, d6
+
+	vadd.f32	q3, q0, q4
+	vacgt.f32	q5, q0, q5
+	vld2.16		{d4,d5}, [SAMPLES]
+	vshl.i32	q3, q3, #10
+	vqshrn.s32	d3, q3, #10
+	vshr.u32	q5, q5, #31
+	vst2.16		{d3,d5}, [SAMPLES]!
+	vadd.i32	q5, q5, q6
+	vst1.32		{q5}, [sp, :128]
+
+	subs		r3, r3, #1
+	bne			.Loop_start_2
+
+	vld1.32		{q0}, [sp, :128]
+	vpadd.i32	d0, d0, d1
+	vpadd.i32	d0, d0, d0
+	vmov.32		r0, d0[0]
+
+	mov			sp, r6
+	vpop		{q4-q7}
+	pop			{r4-r6, pc}

data/mpg123/src/libmpg123/synth_neon_float.S

@@ -0,0 +1,149 @@
+/*
+	synth_neon_float: ARM NEON optimized synth (float output version)
+
+	copyright 1995-2010 by the mpg123 project - free software under the terms of the LGPL 2.1
+	see COPYING and AUTHORS files in distribution or http://mpg123.org
+	initially written by Taihei Monma
+*/
+
+#include "mangle.h"
+
+#define WINDOW r0
+#define B0 r1
+#define SAMPLES r2
+
+/*
+	int synth_1to1_real_neon_asm(real *window, real *b0, real *samples, int bo1);
+	return value: number of clipped samples (0)
+*/
+
+	.text
+	.globl ASM_NAME(synth_1to1_real_neon_asm)
+	ALIGN4
+ASM_NAME(synth_1to1_real_neon_asm):
+	push		{r4-r5, lr}
+	vpush		{q4-q7}
+
+	add			WINDOW, WINDOW, #64
+	sub			WINDOW, WINDOW, r3, lsl #2
+
+	mov			r3, #4
+	mov			r4, #128
+	mov			r5, #64
+.Loop_start_1:
+	vld1.32		{q0,q1}, [WINDOW], r4
+	vld1.32		{q2,q3}, [WINDOW], r4
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW]
+	sub			WINDOW, WINDOW, #352
+	vld1.32		{q8,q9}, [B0, :128], r5
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128]
+	vswp		q1, q4
+	vswp		q3, q6
+	sub			B0, B0, #160
+	vmul.f32	q0, q0, q8
+	vmul.f32	q2, q2, q10
+	vmul.f32	q1, q1, q12
+	vmul.f32	q3, q3, q14
+	vmla.f32	q0, q4, q9
+	vmla.f32	q2, q6, q11
+	vmla.f32	q1, q5, q13
+	vmla.f32	q3, q7, q15
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW], r4
+	vld1.32		{q8,q9}, [WINDOW], r4
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128], r5
+	vswp		q5, q6
+	vswp		q11, q12
+	vmla.f32	q0, q4, q10
+	vmla.f32	q2, q5, q11
+	vmla.f32	q1, q8, q14
+	vld1.32		{q4,q5}, [WINDOW]
+	vld1.32		{q10,q11}, [B0, :128]!
+	add			WINDOW, WINDOW, #96
+	vmla.f32	q3, q4, q10
+	vmla.f32	q0, q6, q12
+	vmla.f32	q2, q7, q13
+	vmla.f32	q1, q9, q15
+	vmla.f32	q3, q5, q11
+	vld2.32		{q4,q5}, [SAMPLES]
+	vpadd.f32	d0, d0, d1
+	vpadd.f32	d4, d4, d5
+	vpadd.f32	d2, d2, d3
+	vpadd.f32	d6, d6, d7
+	vpadd.f32	d0, d0, d4
+	vpadd.f32	d1, d2, d6
+
+	vmov.i32	q1, #0x38000000
+	vmul.f32	q4, q0, q1
+	vst2.32		{q4,q5}, [SAMPLES]!
+
+	subs		r3, r3, #1
+	bne			.Loop_start_1
+
+	mov			r3, #4
+	mov			r5, #-64
+.Loop_start_2:
+	vld1.32		{q0,q1}, [WINDOW], r4
+	vld1.32		{q2,q3}, [WINDOW], r4
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW]
+	sub			WINDOW, WINDOW, #352
+	vld1.32		{q8,q9}, [B0, :128], r5
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128]
+	vswp		q1, q4
+	vswp		q3, q6
+	add			B0, B0, #224
+	vmul.f32	q0, q0, q8
+	vmul.f32	q2, q2, q10
+	vmul.f32	q1, q1, q12
+	vmul.f32	q3, q3, q14
+	vmla.f32	q0, q4, q9
+	vmla.f32	q2, q6, q11
+	vmla.f32	q1, q5, q13
+	vmla.f32	q3, q7, q15
+	vld1.32		{q4,q5}, [WINDOW], r4
+	vld1.32		{q6,q7}, [WINDOW], r4
+	vld1.32		{q8,q9}, [WINDOW], r4
+	vld1.32		{q10,q11}, [B0, :128], r5
+	vld1.32		{q12,q13}, [B0, :128], r5
+	vld1.32		{q14,q15}, [B0, :128], r5
+	vswp		q5, q6
+	vswp		q11, q12
+	vmla.f32	q0, q4, q10
+	vmla.f32	q2, q5, q11
+	vmla.f32	q1, q8, q14
+	vld1.32		{q4,q5}, [WINDOW]
+	vld1.32		{q10,q11}, [B0, :128]
+	add			WINDOW, WINDOW, #96
+	sub			B0, B0, #96
+	vmla.f32	q3, q4, q10
+	vmla.f32	q0, q6, q12
+	vmla.f32	q2, q7, q13
+	vmla.f32	q1, q9, q15
+	vmla.f32	q3, q5, q11
+	vld2.32		{q4,q5}, [SAMPLES]
+	vpadd.f32	d0, d0, d1
+	vpadd.f32	d4, d4, d5
+	vpadd.f32	d2, d2, d3
+	vpadd.f32	d6, d6, d7
+	vpadd.f32	d0, d0, d4
+	vpadd.f32	d1, d2, d6
+
+	vmov.i32	q1, #0x38000000
+	vmul.f32	q4, q0, q1
+	vst2.32		{q4,q5}, [SAMPLES]!
+
+	subs		r3, r3, #1
+	bne			.Loop_start_2
+
+	mov			r0, #0
+
+	vpop		{q4-q7}
+	pop			{r4-r5, pc}