Drop DCTELEM typedef

It does not help as an abstraction and adds dsputil dependencies.

Diego Biurrun committed 13 years ago

Commit 88bd7fd1 parent 2e4bb99163 files · +835 −812Browse files

Unified · Split

163 files

doc/optimization.txt +1 −1
libavcodec/4xm.c +5 −5
libavcodec/alpha/dsputil_alpha.c +6 −6
libavcodec/alpha/dsputil_alpha.h +10 −10
libavcodec/alpha/dsputil_alpha_asm.S +2 −2
libavcodec/alpha/motion_est_alpha.c +2 −2
libavcodec/alpha/mpegvideo_alpha.c +4 −4
libavcodec/alpha/simple_idct_alpha.c +6 −6
libavcodec/arm/dsputil_init_arm.c +9 −9
libavcodec/arm/dsputil_init_armv5te.c +3 −3
libavcodec/arm/dsputil_init_armv6.c +6 −6
libavcodec/arm/dsputil_init_neon.c +8 −8
libavcodec/arm/h264dsp_init_arm.c +8 −8
libavcodec/arm/jrevdct_arm.S +1 −1
libavcodec/arm/mpegvideo_arm.c +2 −2
libavcodec/arm/mpegvideo_armv5te.c +4 −4
libavcodec/arm/rv34dsp_init_arm.c +3 −3
libavcodec/arm/rv34dsp_neon.S +3 −3
libavcodec/arm/simple_idct_armv6.S +3 −3
libavcodec/arm/simple_idct_neon.S +3 −3
libavcodec/arm/vp3dsp_init_arm.c +3 −3
libavcodec/arm/vp8.h +1 −1
libavcodec/arm/vp8dsp_armv6.S +6 −6
libavcodec/arm/vp8dsp_init_armv6.c +6 −6
libavcodec/arm/vp8dsp_init_neon.c +5 −5
libavcodec/asv.h +1 −1
libavcodec/asvdec.c +4 −4
libavcodec/asvenc.c +4 −4
libavcodec/bfin/config_bfin.h +1 −1
libavcodec/bfin/dsputil_bfin.c +3 −3
libavcodec/bfin/dsputil_bfin.h +11 −10
libavcodec/bfin/fdct_bfin.S +4 −4
libavcodec/bfin/idct_bfin.S +2 −2
libavcodec/bfin/mpegvideo_bfin.c +1 −1
libavcodec/bfin/pixels_bfin.S +4 −4
libavcodec/bfin/vp3_bfin.c +2 −2
libavcodec/bfin/vp3_idct_bfin.S +2 −2
libavcodec/bink.c +3 −3
libavcodec/cavs.c +1 −1
libavcodec/cavs.h +1 −1
libavcodec/cavsdec.c +4 −4
libavcodec/cavsdsp.c +2 −2
libavcodec/cavsdsp.h +1 −1
libavcodec/dct-test.c +16 −15
libavcodec/dnxhddec.c +7 −7
libavcodec/dnxhdenc.c +10 −10
libavcodec/dnxhdenc.h +2 −2
libavcodec/dsputil.c +17 −17
libavcodec/dsputil.h +45 −46
libavcodec/dsputil_template.c +7 −7
libavcodec/dv.c +2 −2
libavcodec/dvdata.h +3 −3
libavcodec/dvdec.c +4 −4
libavcodec/eaidct.c +3 −3
libavcodec/eaidct.h +1 −2
libavcodec/eamad.c +3 −3
libavcodec/eatgq.c +3 −3
libavcodec/eatqi.c +3 −3
libavcodec/faandct.c +3 −3
libavcodec/faandct.h +3 −3
libavcodec/faanidct.c +4 −4
libavcodec/faanidct.h +3 −4
libavcodec/h261dec.c +2 −2
libavcodec/h261enc.c +4 −4
libavcodec/h263.c +1 −1
libavcodec/h263.h +4 −4
libavcodec/h264.c +5 −5
libavcodec/h264.h +3 −4
libavcodec/h264_cabac.c +28 −8
libavcodec/h264_cavlc.c +3 −3
libavcodec/h264_mb_template.c +1 −1
libavcodec/h264dsp.h +11 −13
libavcodec/h264idct_template.c +12 −12
libavcodec/h264pred.h +6 −6
libavcodec/h264pred_template.c +10 −10
libavcodec/ituh263dec.c +3 −3
libavcodec/ituh263enc.c +2 −2
libavcodec/jfdctfst.c +9 −9
libavcodec/jfdctint_template.c +14 −14
libavcodec/jrevdct.c +19 −19
libavcodec/mdec.c +4 −4
libavcodec/mimic.c +2 −2
libavcodec/mjpegdec.c +6 −6
libavcodec/mjpegdec.h +2 −2
libavcodec/mjpegenc.c +2 −2
libavcodec/mjpegenc.h +3 −2
libavcodec/mpeg12.c +10 −10
libavcodec/mpeg12.h +1 −1
libavcodec/mpeg12enc.c +4 −4
libavcodec/mpeg4video.h +2 −2
libavcodec/mpeg4videodec.c +4 −4
libavcodec/mpeg4videoenc.c +8 −8
libavcodec/mpegvideo.c +22 −22
libavcodec/mpegvideo.h +25 −25
libavcodec/mpegvideo_enc.c +17 −17
libavcodec/mpegvideo_xvmc.c +1 −1
libavcodec/msmpeg4.c +3 −3
libavcodec/msmpeg4.h +3 −4
libavcodec/msmpeg4enc.c +2 −2
libavcodec/ppc/dsputil_altivec.c +3 −3
libavcodec/ppc/dsputil_ppc.c +7 −7
libavcodec/ppc/h264_altivec.c +9 −9
libavcodec/ppc/mpegvideo_altivec.c +1 −1
libavcodec/ppc/vc1dsp_altivec.c +2 −2
libavcodec/ppc/vp3dsp_altivec.c +2 −2
libavcodec/proresdec.c +6 −5
libavcodec/proresdsp.c +4 −3
libavcodec/proresdsp.h +3 −3
libavcodec/proresenc.c +9 −8
libavcodec/rtjpeg.c +3 −3
libavcodec/rtjpeg.h +1 −1
libavcodec/rv34.c +10 −10
libavcodec/rv34dsp.c +6 −6
libavcodec/rv34dsp.h +2 −2
libavcodec/sh4/dsputil_sh4.c +4 −4
libavcodec/sh4/dsputil_sh4.h +1 −1
libavcodec/sh4/idct_sh4.c +1 −1
libavcodec/simple_idct.c +9 −9
libavcodec/simple_idct.h +11 −12
libavcodec/simple_idct_template.c +7 −7
libavcodec/sparc/dsputil_vis.h +3 −4
libavcodec/sparc/simple_idct_vis.c +3 −3
libavcodec/svq3.c +3 −3
libavcodec/vc1.h +1 −1
libavcodec/vc1dec.c +5 −5
libavcodec/vc1dsp.c +14 −14
libavcodec/vc1dsp.h +10 −10
libavcodec/vp3.c +4 −4
libavcodec/vp3dsp.c +8 −3
libavcodec/vp3dsp.h +3 −4
libavcodec/vp56.h +3 −3
libavcodec/vp8.c +2 −2
libavcodec/vp8.h +2 −2
libavcodec/vp8dsp.c +7 −7
libavcodec/vp8dsp.h +8 −7
libavcodec/wmv2.c +2 −2
libavcodec/wmv2.h +1 −1
libavcodec/wmv2dec.c +2 −2
libavcodec/wmv2dsp.c +3 −2
libavcodec/wmv2dsp.h +2 −3
libavcodec/wmv2enc.c +1 −1
libavcodec/x86/dnxhdenc.c +1 −1
libavcodec/x86/dsputil_mmx.c +8 −8
libavcodec/x86/dsputil_mmx.h +3 −3
libavcodec/x86/dsputilenc.asm +2 −2
libavcodec/x86/dsputilenc_mmx.c +4 −4
libavcodec/x86/h264_idct.asm +13 −13
libavcodec/x86/h264_intrapred_init.c +1 −0
libavcodec/x86/h264dsp_init.c +4 −4
libavcodec/x86/idct_mmx_xvid.c +4 −4
libavcodec/x86/idct_xvid.h +4 −6
libavcodec/x86/mpegvideo.c +8 −8
libavcodec/x86/mpegvideoenc_template.c +1 −1
libavcodec/x86/proresdsp.asm +1 −1
libavcodec/x86/proresdsp_init.c +4 −3
libavcodec/x86/rv34dsp.asm +1 −1
libavcodec/x86/rv34dsp_init.c +3 −3
libavcodec/x86/simple_idct.c +2 −2
libavcodec/x86/vc1dsp_mmx.c +4 −4
libavcodec/x86/vp3dsp_init.c +6 −5
libavcodec/x86/vp8dsp.asm +5 −5
libavcodec/x86/vp8dsp_init.c +9 −9
tools/patcheck +1 −1

@@ -148,7 +148,7 @@ Alignment: · 147 unmodified lines
148	148	Some instructions on some architectures have strict alignment restrictions,
149	149	for example most SSE/SSE2 instructions on x86.
150	150	The minimum guaranteed alignment is written in the .h files, for example:
151		− void (put_pixels_clamped)(const DCTELEM block/align 16/, UINT8 pixels/align 8*/, int line_size);
	151	+ void (put_pixels_clamped)(const int16_t block/align 16/, UINT8 pixels/align 8*/, int line_size);
152	152
153	153
154	154	General Tips:

libavcodec/4xm.c+5−5

@@ -139,7 +139,7 @@ typedef struct FourXContext { · 138 unmodified lines
139	139	int mv[256];
140	140	VLC pre_vlc;
141	141	int last_dc;
142		− DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
	142	+ DECLARE_ALIGNED(16, int16_t, block)[6][64];
143	143	void *bitstream_buffer;
144	144	unsigned int bitstream_buffer_size;
145	145	int version;
@@ -154,7 +154,7 @@ typedef struct FourXContext { · 8 unmodified lines
154	154
155	155	#define MULTIPLY(var, const) (((var) * (const)) >> 16)
156	156
157		−static void idct(DCTELEM block[64])
	157	+static void idct(int16_t block[64])
158	158	{
159	159	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
160	160	int tmp10, tmp11, tmp12, tmp13;
@@ -451,7 +451,7 @@ static int decode_p_frame(FourXContext f, const uint8_t buf, int length) · 290 unmodified lines
451	451	* decode block and dequantize.
452	452	* Note this is almost identical to MJPEG.
453	453	*/
454		−static int decode_i_block(FourXContext f, DCTELEM block)
	454	+static int decode_i_block(FourXContext f, int16_t block)
455	455	{
456	456	int code, i, j, level, val;
457	457
@@ -496,7 +496,7 @@ static int decode_i_block(FourXContext f, DCTELEM block) · 38 unmodified lines
496	496
497	497	static inline void idct_put(FourXContext *f, int x, int y)
498	498	{
499		− DCTELEM (*block)[64] = f->block;
	499	+ int16_t (*block)[64] = f->block;
500	500	int stride = f->current_picture->linesize[0] >> 1;
501	501	int i;
502	502	uint16_t dst = ((uint16_t)f->current_picture->data[0]) + y * stride + x;
@@ -517,7 +517,7 @@ static inline void idct_put(FourXContext *f, int x, int y) · 14 unmodified lines
517	517	* cr = (-1b - 4g + 5r) / 14 */
518	518	for (y = 0; y < 8; y++) {
519	519	for (x = 0; x < 8; x++) {
520		− DCTELEM temp = block[(x >> 2) + 2 (y >> 2)] +
	520	+ int16_t temp = block[(x >> 2) + 2 (y >> 2)] +
521	521	2 * (x & 3) + 2 * 8 * (y & 3); // FIXME optimize
522	522	int cb = block[4][x + 8 * y];
523	523	int cr = block[5][x + 8 * y];

libavcodec/alpha/dsputil_alpha.c+6−6

@@ -23,15 +23,15 @@ · 22 unmodified lines
23	23	#include "dsputil_alpha.h"
24	24	#include "asm.h"
25	25
26		−void (put_pixels_clamped_axp_p)(const DCTELEM block, uint8_t *pixels,
	26	+void (put_pixels_clamped_axp_p)(const int16_t block, uint8_t *pixels,
27	27	int line_size);
28		−void (add_pixels_clamped_axp_p)(const DCTELEM block, uint8_t *pixels,
	28	+void (add_pixels_clamped_axp_p)(const int16_t block, uint8_t *pixels,
29	29	int line_size);
30	30
31	31	#if 0
32	32	/* These functions were the base for the optimized assembler routines,
33	33	and remain here for documentation purposes. */
34		−static void put_pixels_clamped_mvi(const DCTELEM block, uint8_t pixels,
	34	+static void put_pixels_clamped_mvi(const int16_t block, uint8_t pixels,
35	35	int line_size)
36	36	{
37	37	int i = 8;
@@ -55,7 +55,7 @@ static void put_pixels_clamped_mvi(const DCTELEM block, uint8_t pixels, · 17 unmodified lines
55	55	} while (--i);
56	56	}
57	57
58		−void add_pixels_clamped_mvi(const DCTELEM block, uint8_t pixels,
	58	+void add_pixels_clamped_mvi(const int16_t block, uint8_t pixels,
59	59	int line_size)
60	60	{
61	61	int h = 8;
@@ -100,9 +100,9 @@ void add_pixels_clamped_mvi(const DCTELEM block, uint8_t pixels, · 38 unmodified lines
100	100	}
101	101	#endif
102	102
103		−static void clear_blocks_axp(DCTELEM *blocks) {
	103	+static void clear_blocks_axp(int16_t *blocks) {
104	104	uint64_t p = (uint64_t ) blocks;
105		− int n = sizeof(DCTELEM) * 6 * 64;
	105	+ int n = sizeof(int16_t) * 6 * 64;
106	106
107	107	do {
108	108	p[0] = 0;

libavcodec/alpha/dsputil_alpha.h+10−10

@@ -19,26 +19,26 @@ · 18 unmodified lines
19	19	#ifndef AVCODEC_ALPHA_DSPUTIL_ALPHA_H
20	20	#define AVCODEC_ALPHA_DSPUTIL_ALPHA_H
21	21
22		−#include "libavcodec/dsputil.h"
	22	+#include <stdint.h>
23	23
24		−void ff_simple_idct_axp(DCTELEM *block);
25		−void ff_simple_idct_put_axp(uint8_t dest, int line_size, DCTELEM block);
26		−void ff_simple_idct_add_axp(uint8_t dest, int line_size, DCTELEM block);
	24	+void ff_simple_idct_axp(int16_t *block);
	25	+void ff_simple_idct_put_axp(uint8_t dest, int line_size, int16_t block);
	26	+void ff_simple_idct_add_axp(uint8_t dest, int line_size, int16_t block);
27	27
28	28	void put_pixels_axp_asm(uint8_t block, const uint8_t pixels,
29	29	int line_size, int h);
30		−void put_pixels_clamped_mvi_asm(const DCTELEM block, uint8_t pixels,
	30	+void put_pixels_clamped_mvi_asm(const int16_t block, uint8_t pixels,
31	31	int line_size);
32		−void add_pixels_clamped_mvi_asm(const DCTELEM block, uint8_t pixels,
	32	+void add_pixels_clamped_mvi_asm(const int16_t block, uint8_t pixels,
33	33	int line_size);
34		−extern void (put_pixels_clamped_axp_p)(const DCTELEM block, uint8_t *pixels,
	34	+extern void (put_pixels_clamped_axp_p)(const int16_t block, uint8_t *pixels,
35	35	int line_size);
36		−extern void (add_pixels_clamped_axp_p)(const DCTELEM block, uint8_t *pixels,
	36	+extern void (add_pixels_clamped_axp_p)(const int16_t block, uint8_t *pixels,
37	37	int line_size);
38	38
39		−void get_pixels_mvi(DCTELEM *restrict block,
	39	+void get_pixels_mvi(int16_t *restrict block,
40	40	const uint8_t *restrict pixels, int line_size);
41		−void diff_pixels_mvi(DCTELEM block, const uint8_t s1, const uint8_t *s2,
	41	+void diff_pixels_mvi(int16_t block, const uint8_t s1, const uint8_t *s2,
42	42	int stride);
43	43	int pix_abs8x8_mvi(void v, uint8_t pix1, uint8_t *pix2, int line_size, int h);
44	44	int pix_abs16x16_mvi_asm(void v, uint8_t pix1, uint8_t *pix2, int line_size, int h);

libavcodec/alpha/dsputil_alpha_asm.S+2−2

@@ -135,7 +135,7 @@ $aligned: · 134 unmodified lines
135	135	.end put_pixels_axp_asm
136	136
137	137	/************************************************************************
138		− * void put_pixels_clamped_mvi_asm(const DCTELEM block, uint8_t pixels,
	138	+ * void put_pixels_clamped_mvi_asm(const int16_t block, uint8_t pixels,
139	139	* int line_size)
140	140	*/
141	141	.align 6
@@ -185,7 +185,7 @@ put_pixels_clamped_mvi_asm: · 43 unmodified lines
185	185	.end put_pixels_clamped_mvi_asm
186	186
187	187	/************************************************************************
188		− * void add_pixels_clamped_mvi_asm(const DCTELEM block, uint8_t pixels,
	188	+ * void add_pixels_clamped_mvi_asm(const int16_t block, uint8_t pixels,
189	189	* int line_size)
190	190	*/
191	191	.align 6

libavcodec/alpha/motion_est_alpha.c+2−2

@@ -23,7 +23,7 @@ · 22 unmodified lines
23	23	#include "dsputil_alpha.h"
24	24	#include "asm.h"
25	25
26		−void get_pixels_mvi(DCTELEM *restrict block,
	26	+void get_pixels_mvi(int16_t *restrict block,
27	27	const uint8_t *restrict pixels, int line_size)
28	28	{
29	29	int h = 8;
@@ -40,7 +40,7 @@ void get_pixels_mvi(DCTELEM *restrict block, · 10 unmodified lines
40	40	} while (--h);
41	41	}
42	42
43		−void diff_pixels_mvi(DCTELEM block, const uint8_t s1, const uint8_t *s2,
	43	+void diff_pixels_mvi(int16_t block, const uint8_t s1, const uint8_t *s2,
44	44	int stride) {
45	45	int h = 8;
46	46	uint64_t mask = 0x4040;

libavcodec/alpha/mpegvideo_alpha.c+4−4

@@ -23,7 +23,7 @@ · 22 unmodified lines
23	23	#include "libavcodec/mpegvideo.h"
24	24	#include "asm.h"
25	25
26		−static void dct_unquantize_h263_axp(DCTELEM *block, int n_coeffs,
	26	+static void dct_unquantize_h263_axp(int16_t *block, int n_coeffs,
27	27	uint64_t qscale, uint64_t qadd)
28	28	{
29	29	uint64_t qmul = qscale << 1;
@@ -69,12 +69,12 @@ static void dct_unquantize_h263_axp(DCTELEM *block, int n_coeffs, · 39 unmodified lines
69	69	}
70	70	}
71	71
72		−static void dct_unquantize_h263_intra_axp(MpegEncContext s, DCTELEM block,
	72	+static void dct_unquantize_h263_intra_axp(MpegEncContext s, int16_t block,
73	73	int n, int qscale)
74	74	{
75	75	int n_coeffs;
76	76	uint64_t qadd;
77		− DCTELEM block0 = block[0];
	77	+ int16_t block0 = block[0];
78	78
79	79	if (!s->h263_aic) {
80	80	if (n < 4)
@@ -96,7 +96,7 @@ static void dct_unquantize_h263_intra_axp(MpegEncContext s, DCTELEM block, · 15 unmodified lines
96	96	block[0] = block0;
97	97	}
98	98
99		−static void dct_unquantize_h263_inter_axp(MpegEncContext s, DCTELEM block,
	99	+static void dct_unquantize_h263_inter_axp(MpegEncContext s, int16_t block,
100	100	int n, int qscale)
101	101	{
102	102	int n_coeffs = s->inter_scantable.raster_end[s->block_last_index[n]];

libavcodec/alpha/simple_idct_alpha.c+6−6

@@ -44,7 +44,7 @@ · 43 unmodified lines
44	44	#define COL_SHIFT 20
45	45
46	46	/* 0: all entries 0, 1: only first entry nonzero, 2: otherwise */
47		−static inline int idct_row(DCTELEM *row)
	47	+static inline int idct_row(int16_t *row)
48	48	{
49	49	int a0, a1, a2, a3, b0, b1, b2, b3, t;
50	50	uint64_t l, r, t2;
@@ -152,7 +152,7 @@ static inline int idct_row(DCTELEM *row) · 101 unmodified lines
152	152	return 2;
153	153	}
154	154
155		−static inline void idct_col(DCTELEM *col)
	155	+static inline void idct_col(int16_t *col)
156	156	{
157	157	int a0, a1, a2, a3, b0, b1, b2, b3;
158	158
@@ -229,7 +229,7 @@ static inline void idct_col(DCTELEM *col) · 70 unmodified lines
229	229
230	230	/* If all rows but the first one are zero after row transformation,
231	231	all rows will be identical after column transformation. */
232		−static inline void idct_col2(DCTELEM *col)
	232	+static inline void idct_col2(int16_t *col)
233	233	{
234	234	int i;
235	235	uint64_t l, r;
@@ -251,7 +251,7 @@ static inline void idct_col2(DCTELEM *col) · 15 unmodified lines
251	251	stq(l, col + 14 * 4); stq(r, col + 15 * 4);
252	252	}
253	253
254		−void ff_simple_idct_axp(DCTELEM *block)
	254	+void ff_simple_idct_axp(int16_t *block)
255	255	{
256	256
257	257	int i;
@@ -291,13 +291,13 @@ void ff_simple_idct_axp(DCTELEM *block) · 33 unmodified lines
291	291	}
292	292	}
293	293
294		−void ff_simple_idct_put_axp(uint8_t dest, int line_size, DCTELEM block)
	294	+void ff_simple_idct_put_axp(uint8_t dest, int line_size, int16_t block)
295	295	{
296	296	ff_simple_idct_axp(block);
297	297	put_pixels_clamped_axp_p(block, dest, line_size);
298	298	}
299	299
300		−void ff_simple_idct_add_axp(uint8_t dest, int line_size, DCTELEM block)
	300	+void ff_simple_idct_add_axp(uint8_t dest, int line_size, int16_t block)
301	301	{
302	302	ff_simple_idct_axp(block);
303	303	add_pixels_clamped_axp_p(block, dest, line_size);

libavcodec/arm/dsputil_init_arm.c+9−9

@@ -23,12 +23,12 @@ · 22 unmodified lines
23	23	#include "libavcodec/dsputil.h"
24	24	#include "dsputil_arm.h"
25	25
26		−void ff_j_rev_dct_arm(DCTELEM *data);
27		−void ff_simple_idct_arm(DCTELEM *data);
	26	+void ff_j_rev_dct_arm(int16_t *data);
	27	+void ff_simple_idct_arm(int16_t *data);
28	28
29	29	/* XXX: local hack */
30		−static void (ff_put_pixels_clamped)(const DCTELEM block, uint8_t *pixels, int line_size);
31		−static void (ff_add_pixels_clamped)(const DCTELEM block, uint8_t *pixels, int line_size);
	30	+static void (ff_put_pixels_clamped)(const int16_t block, uint8_t *pixels, int line_size);
	31	+static void (ff_add_pixels_clamped)(const int16_t block, uint8_t *pixels, int line_size);
32	32
33	33	void ff_put_pixels8_arm(uint8_t block, const uint8_t pixels, int line_size, int h);
34	34	void ff_put_pixels8_x2_arm(uint8_t block, const uint8_t pixels, int line_size, int h);
@@ -48,27 +48,27 @@ CALL_2X_PIXELS(ff_put_no_rnd_pixels16_x2_arm, ff_put_no_rnd_pixels8_x2_arm, 8) · 13 unmodified lines
48	48	CALL_2X_PIXELS(ff_put_no_rnd_pixels16_y2_arm, ff_put_no_rnd_pixels8_y2_arm, 8)
49	49	CALL_2X_PIXELS(ff_put_no_rnd_pixels16_xy2_arm, ff_put_no_rnd_pixels8_xy2_arm,8)
50	50
51		−void ff_add_pixels_clamped_arm(const DCTELEM block, uint8_t dest,
	51	+void ff_add_pixels_clamped_arm(const int16_t block, uint8_t dest,
52	52	int line_size);
53	53
54	54	/* XXX: those functions should be suppressed ASAP when all IDCTs are
55	55	converted */
56		−static void j_rev_dct_arm_put(uint8_t dest, int line_size, DCTELEM block)
	56	+static void j_rev_dct_arm_put(uint8_t dest, int line_size, int16_t block)
57	57	{
58	58	ff_j_rev_dct_arm (block);
59	59	ff_put_pixels_clamped(block, dest, line_size);
60	60	}
61		−static void j_rev_dct_arm_add(uint8_t dest, int line_size, DCTELEM block)
	61	+static void j_rev_dct_arm_add(uint8_t dest, int line_size, int16_t block)
62	62	{
63	63	ff_j_rev_dct_arm (block);
64	64	ff_add_pixels_clamped(block, dest, line_size);
65	65	}
66		−static void simple_idct_arm_put(uint8_t dest, int line_size, DCTELEM block)
	66	+static void simple_idct_arm_put(uint8_t dest, int line_size, int16_t block)
67	67	{
68	68	ff_simple_idct_arm (block);
69	69	ff_put_pixels_clamped(block, dest, line_size);
70	70	}
71		−static void simple_idct_arm_add(uint8_t dest, int line_size, DCTELEM block)
	71	+static void simple_idct_arm_add(uint8_t dest, int line_size, int16_t block)
72	72	{
73	73	ff_simple_idct_arm (block);
74	74	ff_add_pixels_clamped(block, dest, line_size);

libavcodec/arm/dsputil_init_armv5te.c+3−3

@@ -21,9 +21,9 @@ · 20 unmodified lines
21	21	#include "libavcodec/dsputil.h"
22	22	#include "dsputil_arm.h"
23	23
24		−void ff_simple_idct_armv5te(DCTELEM *data);
25		−void ff_simple_idct_put_armv5te(uint8_t dest, int line_size, DCTELEM data);
26		−void ff_simple_idct_add_armv5te(uint8_t dest, int line_size, DCTELEM data);
	24	+void ff_simple_idct_armv5te(int16_t *data);
	25	+void ff_simple_idct_put_armv5te(uint8_t dest, int line_size, int16_t data);
	26	+void ff_simple_idct_add_armv5te(uint8_t dest, int line_size, int16_t data);
27	27
28	28	av_cold void ff_dsputil_init_armv5te(DSPContext c, AVCodecContext avctx)
29	29	{

libavcodec/arm/dsputil_init_armv6.c+6−6

@@ -24,9 +24,9 @@ · 23 unmodified lines
24	24	#include "libavcodec/dsputil.h"
25	25	#include "dsputil_arm.h"
26	26
27		−void ff_simple_idct_armv6(DCTELEM *data);
28		−void ff_simple_idct_put_armv6(uint8_t dest, int line_size, DCTELEM data);
29		−void ff_simple_idct_add_armv6(uint8_t dest, int line_size, DCTELEM data);
	27	+void ff_simple_idct_armv6(int16_t *data);
	28	+void ff_simple_idct_put_armv6(uint8_t dest, int line_size, int16_t data);
	29	+void ff_simple_idct_add_armv6(uint8_t dest, int line_size, int16_t data);
30	30
31	31	void ff_put_pixels16_armv6(uint8_t , const uint8_t , int, int);
32	32	void ff_put_pixels16_x2_armv6(uint8_t , const uint8_t , int, int);
@@ -46,12 +46,12 @@ void ff_put_pixels8_y2_no_rnd_armv6(uint8_t , const uint8_t , int, int); · 13 unmodified lines
46	46
47	47	void ff_avg_pixels8_armv6(uint8_t , const uint8_t , int, int);
48	48
49		−void ff_add_pixels_clamped_armv6(const DCTELEM *block,
	49	+void ff_add_pixels_clamped_armv6(const int16_t *block,
50	50	uint8_t *restrict pixels,
51	51	int line_size);
52	52
53		−void ff_get_pixels_armv6(DCTELEM block, const uint8_t pixels, int stride);
54		−void ff_diff_pixels_armv6(DCTELEM block, const uint8_t s1,
	53	+void ff_get_pixels_armv6(int16_t block, const uint8_t pixels, int stride);
	54	+void ff_diff_pixels_armv6(int16_t block, const uint8_t s1,
55	55	const uint8_t *s2, int stride);
56	56
57	57	int ff_pix_abs16_armv6(void s, uint8_t blk1, uint8_t *blk2,

libavcodec/arm/dsputil_init_neon.c+8−8

@@ -25,12 +25,12 @@ · 24 unmodified lines
25	25	#include "libavcodec/dsputil.h"
26	26	#include "dsputil_arm.h"
27	27
28		−void ff_simple_idct_neon(DCTELEM *data);
29		−void ff_simple_idct_put_neon(uint8_t dest, int line_size, DCTELEM data);
30		−void ff_simple_idct_add_neon(uint8_t dest, int line_size, DCTELEM data);
	28	+void ff_simple_idct_neon(int16_t *data);
	29	+void ff_simple_idct_put_neon(uint8_t dest, int line_size, int16_t data);
	30	+void ff_simple_idct_add_neon(uint8_t dest, int line_size, int16_t data);
31	31
32		−void ff_clear_block_neon(DCTELEM *block);
33		−void ff_clear_blocks_neon(DCTELEM *blocks);
	32	+void ff_clear_block_neon(int16_t *block);
	33	+void ff_clear_blocks_neon(int16_t *blocks);
34	34
35	35	void ff_put_pixels16_neon(uint8_t , const uint8_t , int, int);
36	36	void ff_put_pixels16_x2_neon(uint8_t , const uint8_t , int, int);
@@ -62,9 +62,9 @@ void ff_avg_pixels8_x2_no_rnd_neon(uint8_t , const uint8_t , int, int); · 25 unmodified lines
62	62	void ff_avg_pixels8_y2_no_rnd_neon(uint8_t , const uint8_t , int, int);
63	63	void ff_avg_pixels8_xy2_no_rnd_neon(uint8_t , const uint8_t , int, int);
64	64
65		−void ff_add_pixels_clamped_neon(const DCTELEM , uint8_t , int);
66		−void ff_put_pixels_clamped_neon(const DCTELEM , uint8_t , int);
67		−void ff_put_signed_pixels_clamped_neon(const DCTELEM , uint8_t , int);
	65	+void ff_add_pixels_clamped_neon(const int16_t , uint8_t , int);
	66	+void ff_put_pixels_clamped_neon(const int16_t , uint8_t , int);
	67	+void ff_put_signed_pixels_clamped_neon(const int16_t , uint8_t , int);
68	68
69	69	void ff_put_h264_qpel16_mc00_neon(uint8_t , uint8_t , int);
70	70	void ff_put_h264_qpel16_mc10_neon(uint8_t , uint8_t , int);

libavcodec/arm/h264dsp_init_arm.c+8−8

@@ -50,22 +50,22 @@ void ff_biweight_h264_pixels_4_neon(uint8_t dst, uint8_t src, int stride, · 49 unmodified lines
50	50	int height, int log2_den, int weightd,
51	51	int weights, int offset);
52	52
53		−void ff_h264_idct_add_neon(uint8_t dst, DCTELEM block, int stride);
54		−void ff_h264_idct_dc_add_neon(uint8_t dst, DCTELEM block, int stride);
	53	+void ff_h264_idct_add_neon(uint8_t dst, int16_t block, int stride);
	54	+void ff_h264_idct_dc_add_neon(uint8_t dst, int16_t block, int stride);
55	55	void ff_h264_idct_add16_neon(uint8_t dst, const int block_offset,
56		− DCTELEM *block, int stride,
	56	+ int16_t *block, int stride,
57	57	const uint8_t nnzc[6*8]);
58	58	void ff_h264_idct_add16intra_neon(uint8_t dst, const int block_offset,
59		− DCTELEM *block, int stride,
	59	+ int16_t *block, int stride,
60	60	const uint8_t nnzc[6*8]);
61	61	void ff_h264_idct_add8_neon(uint8_t *dest, const int block_offset,
62		− DCTELEM *block, int stride,
	62	+ int16_t *block, int stride,
63	63	const uint8_t nnzc[6*8]);
64	64
65		−void ff_h264_idct8_add_neon(uint8_t dst, DCTELEM block, int stride);
66		−void ff_h264_idct8_dc_add_neon(uint8_t dst, DCTELEM block, int stride);
	65	+void ff_h264_idct8_add_neon(uint8_t dst, int16_t block, int stride);
	66	+void ff_h264_idct8_dc_add_neon(uint8_t dst, int16_t block, int stride);
67	67	void ff_h264_idct8_add4_neon(uint8_t dst, const int block_offset,
68		− DCTELEM *block, int stride,
	68	+ int16_t *block, int stride,
69	69	const uint8_t nnzc[6*8]);
70	70
71	71	static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)

libavcodec/arm/jrevdct_arm.S+1−1

@@ -66,7 +66,7 @@ row_loop: · 65 unmodified lines
66	66	ldrsh r2, [lr, # 2] @ r2 = 'd2'
67	67
68	68	@ Optimization for row that have all items except the first set to 0
69		− @ (this works as the DCTELEMS are always 4-byte aligned)
	69	+ @ (this works as the int16_t are always 4-byte aligned)
70	70	ldr r5, [lr, # 0]
71	71	ldr r6, [lr, # 4]
72	72	ldr r3, [lr, # 8]

libavcodec/arm/mpegvideo_arm.c+2−2

@@ -34,9 +34,9 @@ CHK_OFFS(MpegEncContext, inter_scantable.raster_end, INTER_SCANTAB_RASTER_END); · 33 unmodified lines
34	34	CHK_OFFS(MpegEncContext, h263_aic, H263_AIC);
35	35	#endif
36	36
37		−void ff_dct_unquantize_h263_inter_neon(MpegEncContext s, DCTELEM block,
	37	+void ff_dct_unquantize_h263_inter_neon(MpegEncContext s, int16_t block,
38	38	int n, int qscale);
39		−void ff_dct_unquantize_h263_intra_neon(MpegEncContext s, DCTELEM block,
	39	+void ff_dct_unquantize_h263_intra_neon(MpegEncContext s, int16_t block,
40	40	int n, int qscale);
41	41
42	42	void ff_MPV_common_init_arm(MpegEncContext *s)

libavcodec/arm/mpegvideo_armv5te.c+4−4

@@ -24,7 +24,7 @@ · 23 unmodified lines
24	24	#include "libavcodec/mpegvideo.h"
25	25	#include "mpegvideo_arm.h"
26	26
27		−void ff_dct_unquantize_h263_armv5te(DCTELEM *block, int qmul, int qadd, int count);
	27	+void ff_dct_unquantize_h263_armv5te(int16_t *block, int qmul, int qadd, int count);
28	28
29	29	#ifdef ENABLE_ARM_TESTS
30	30	/**
@@ -32,7 +32,7 @@ void ff_dct_unquantize_h263_armv5te(DCTELEM *block, int qmul, int qadd, int coun · 1 unmodified line
32	32	* have optimized implementations for each architecture. Is also used as a reference
33	33	* implementation in regression tests
34	34	*/
35		−static inline void dct_unquantize_h263_helper_c(DCTELEM *block, int qmul, int qadd, int count)
	35	+static inline void dct_unquantize_h263_helper_c(int16_t *block, int qmul, int qadd, int count)
36	36	{
37	37	int i, level;
38	38	for (i = 0; i < count; i++) {
@@ -50,7 +50,7 @@ static inline void dct_unquantize_h263_helper_c(DCTELEM *block, int qmul, int qa · 11 unmodified lines
50	50	#endif
51	51
52	52	static void dct_unquantize_h263_intra_armv5te(MpegEncContext *s,
53		− DCTELEM *block, int n, int qscale)
	53	+ int16_t *block, int n, int qscale)
54	54	{
55	55	int level, qmul, qadd;
56	56	int nCoeffs;
@@ -79,7 +79,7 @@ static void dct_unquantize_h263_intra_armv5te(MpegEncContext *s, · 22 unmodified lines
79	79	}
80	80
81	81	static void dct_unquantize_h263_inter_armv5te(MpegEncContext *s,
82		− DCTELEM *block, int n, int qscale)
	82	+ int16_t *block, int n, int qscale)
83	83	{
84	84	int qmul, qadd;
85	85	int nCoeffs;

libavcodec/arm/rv34dsp_init_arm.c+3−3

@@ -24,11 +24,11 @@ · 23 unmodified lines
24	24	#include "libavcodec/rv34dsp.h"
25	25	#include "libavutil/arm/cpu.h"
26	26
27		−void ff_rv34_inv_transform_noround_neon(DCTELEM *block);
	27	+void ff_rv34_inv_transform_noround_neon(int16_t *block);
28	28
29		−void ff_rv34_inv_transform_noround_dc_neon(DCTELEM *block);
	29	+void ff_rv34_inv_transform_noround_dc_neon(int16_t *block);
30	30
31		−void ff_rv34_idct_add_neon(uint8_t dst, ptrdiff_t stride, DCTELEM block);
	31	+void ff_rv34_idct_add_neon(uint8_t dst, ptrdiff_t stride, int16_t block);
32	32	void ff_rv34_idct_dc_add_neon(uint8_t *dst, ptrdiff_t stride, int dc);
33	33
34	34	void ff_rv34dsp_init_arm(RV34DSPContext c, DSPContext dsp)

libavcodec/arm/rv34dsp_neon.S+3−3

@@ -67,7 +67,7 @@ · 66 unmodified lines
67	67	vsub.s32 q15, q14, q9 @ z0 - z3
68	68	.endm
69	69
70		−/* void rv34_idct_add_c(uint8_t dst, int stride, DCTELEM block) */
	70	+/* void rv34_idct_add_c(uint8_t dst, int stride, int16_t block) */
71	71	function ff_rv34_idct_add_neon, export=1
72	72	mov r3, r0
73	73	rv34_inv_transform r2
@@ -97,7 +97,7 @@ function ff_rv34_idct_add_neon, export=1 · 23 unmodified lines
97	97	bx lr
98	98	endfunc
99	99
100		−/* void rv34_inv_transform_noround_neon(DCTELEM block); /
	100	+/* void rv34_inv_transform_noround_neon(int16_t block); /
101	101	function ff_rv34_inv_transform_noround_neon, export=1
102	102	rv34_inv_transform r0
103	103	vshl.s32 q11, q2, #1
@@ -142,7 +142,7 @@ function ff_rv34_idct_dc_add_neon, export=1 · 38 unmodified lines
142	142	bx lr
143	143	endfunc
144	144
145		−/* void rv34_inv_transform_dc_noround_c(DCTELEM block) /
	145	+/* void rv34_inv_transform_dc_noround_c(int16_t block) /
146	146	function ff_rv34_inv_transform_noround_dc_neon, export=1
147	147	vld1.16 {d28[]}, [r0,:16] @ block[0]
148	148	vmov.i16 d4, #251

libavcodec/arm/simple_idct_armv6.S+3−3

@@ -375,7 +375,7 @@ endfunc · 374 unmodified lines
375	375	sub r0, r0, #(16*7)
376	376	.endm
377	377
378		−/* void ff_simple_idct_armv6(DCTELEM data); /
	378	+/* void ff_simple_idct_armv6(int16_t data); /
379	379	function ff_simple_idct_armv6, export=1
380	380	push {r4-r11, lr}
381	381	sub sp, sp, #128
@@ -390,7 +390,7 @@ function ff_simple_idct_armv6, export=1 · 8 unmodified lines
390	390	pop {r4-r11, pc}
391	391	endfunc
392	392
393		−/* ff_simple_idct_add_armv6(uint8_t dest, int line_size, DCTELEM data); */
	393	+/* ff_simple_idct_add_armv6(uint8_t dest, int line_size, int16_t data); */
394	394	function ff_simple_idct_add_armv6, export=1
395	395	push {r0, r1, r4-r11, lr}
396	396	sub sp, sp, #128
@@ -407,7 +407,7 @@ function ff_simple_idct_add_armv6, export=1 · 10 unmodified lines
407	407	pop {r4-r11, pc}
408	408	endfunc
409	409
410		−/* ff_simple_idct_put_armv6(uint8_t dest, int line_size, DCTELEM data); */
	410	+/* ff_simple_idct_put_armv6(uint8_t dest, int line_size, int16_t data); */
411	411	function ff_simple_idct_put_armv6, export=1
412	412	push {r0, r1, r4-r11, lr}
413	413	sub sp, sp, #128

libavcodec/arm/simple_idct_neon.S+3−3

@@ -261,7 +261,7 @@ endconst · 260 unmodified lines
261	261	pop {r4-r7, pc}
262	262	.endm
263	263
264		−/* void ff_simple_idct_put_neon(uint8_t dst, int line_size, DCTELEM data); */
	264	+/* void ff_simple_idct_put_neon(uint8_t dst, int line_size, int16_t data); */
265	265	function ff_simple_idct_put_neon, export=1
266	266	idct_start r2
267	267
@@ -316,7 +316,7 @@ function idct_col4_add8_neon · 48 unmodified lines
316	316	bx lr
317	317	endfunc
318	318
319		−/* void ff_simple_idct_add_neon(uint8_t dst, int line_size, DCTELEM data); */
	319	+/* void ff_simple_idct_add_neon(uint8_t dst, int line_size, int16_t data); */
320	320	function ff_simple_idct_add_neon, export=1
321	321	idct_start r2
322	322
@@ -355,7 +355,7 @@ function idct_col4_st16_neon · 32 unmodified lines
355	355	bx lr
356	356	endfunc
357	357
358		−/* void ff_simple_idct_neon(DCTELEM data); /
	358	+/* void ff_simple_idct_neon(int16_t data); /
359	359	function ff_simple_idct_neon, export=1
360	360	idct_start r0
361	361

libavcodec/arm/vp3dsp_init_arm.c+3−3

@@ -24,9 +24,9 @@ · 23 unmodified lines
24	24	#include "libavcodec/dsputil.h"
25	25	#include "libavcodec/vp3dsp.h"
26	26
27		−void ff_vp3_idct_put_neon(uint8_t dest, int line_size, DCTELEM data);
28		−void ff_vp3_idct_add_neon(uint8_t dest, int line_size, DCTELEM data);
29		−void ff_vp3_idct_dc_add_neon(uint8_t dest, int line_size, const DCTELEM data);
	27	+void ff_vp3_idct_put_neon(uint8_t dest, int line_size, int16_t data);
	28	+void ff_vp3_idct_add_neon(uint8_t dest, int line_size, int16_t data);
	29	+void ff_vp3_idct_dc_add_neon(uint8_t dest, int line_size, const int16_t data);
30	30
31	31	void ff_vp3_v_loop_filter_neon(uint8_t , int, int );
32	32	void ff_vp3_h_loop_filter_neon(uint8_t , int, int );

libavcodec/arm/vp8.h+1−1

@@ -23,7 +23,7 @@ · 22 unmodified lines
23	23
24	24	#if HAVE_ARMV6_EXTERNAL
25	25	#define decode_block_coeffs_internal ff_decode_block_coeffs_armv6
26		−int ff_decode_block_coeffs_armv6(VP56RangeCoder *rc, DCTELEM block[16],
	26	+int ff_decode_block_coeffs_armv6(VP56RangeCoder *rc, int16_t block[16],
27	27	uint8_t probs[8][3][NUM_DCT_TOKENS-1],
28	28	int i, uint8_t *token_prob, int16_t qmul[2]);
29	29	#endif

libavcodec/arm/vp8dsp_armv6.S+6−6

@@ -56,7 +56,7 @@ · 55 unmodified lines
56	56
57	57	@ idct
58	58
59		−@ void vp8_luma_dc_wht(DCTELEM block[4][4][16], DCTELEM dc[16])
	59	+@ void vp8_luma_dc_wht(int16_t block[4][4][16], int16_t dc[16])
60	60	function ff_vp8_luma_dc_wht_armv6, export=1
61	61	push {r4-r10, lr}
62	62
@@ -179,7 +179,7 @@ function ff_vp8_luma_dc_wht_armv6, export=1 · 116 unmodified lines
179	179	pop {r4-r10, pc}
180	180	endfunc
181	181
182		−@ void vp8_luma_dc_wht_dc(DCTELEM block[4][4][16], DCTELEM dc[16])
	182	+@ void vp8_luma_dc_wht_dc(int16_t block[4][4][16], int16_t dc[16])
183	183	function ff_vp8_luma_dc_wht_dc_armv6, export=1
184	184	ldrsh r2, [r1]
185	185	mov r3, #0
@@ -192,7 +192,7 @@ function ff_vp8_luma_dc_wht_dc_armv6, export=1 · 6 unmodified lines
192	192	bx lr
193	193	endfunc
194	194
195		−@ void vp8_idct_add(uint8_t *dst, DCTELEM block[16], int stride)
	195	+@ void vp8_idct_add(uint8_t *dst, int16_t block[16], int stride)
196	196	function ff_vp8_idct_add_armv6, export=1
197	197	push {r4-r12, lr}
198	198	sub sp, sp, #32
@@ -314,7 +314,7 @@ function ff_vp8_idct_add_armv6, export=1 · 115 unmodified lines
314	314	pop {r4-r12, pc}
315	315	endfunc
316	316
317		−@ void vp8_idct_dc_add(uint8_t *dst, DCTELEM block[16], int stride)
	317	+@ void vp8_idct_dc_add(uint8_t *dst, int16_t block[16], int stride)
318	318	function ff_vp8_idct_dc_add_armv6, export=1
319	319	push {r4-r6, lr}
320	320	add r6, r0, r2, lsl #1
@@ -355,7 +355,7 @@ function ff_vp8_idct_dc_add_armv6, export=1 · 34 unmodified lines
355	355	pop {r4-r6, pc}
356	356	endfunc
357	357
358		−@ void vp8_idct_dc_add4uv(uint8_t *dst, DCTELEM block[4][16], int stride)
	358	+@ void vp8_idct_dc_add4uv(uint8_t *dst, int16_t block[4][16], int stride)
359	359	function ff_vp8_idct_dc_add4uv_armv6, export=1
360	360	push {r4, lr}
361	361
@@ -371,7 +371,7 @@ function ff_vp8_idct_dc_add4uv_armv6, export=1 · 9 unmodified lines
371	371	pop {r4, pc}
372	372	endfunc
373	373
374		−@ void vp8_idct_dc_add4y(uint8_t *dst, DCTELEM block[4][16], int stride)
	374	+@ void vp8_idct_dc_add4y(uint8_t *dst, int16_t block[4][16], int stride)
375	375	function ff_vp8_idct_dc_add4y_armv6, export=1
376	376	push {r4, lr}
377	377

libavcodec/arm/vp8dsp_init_armv6.c+6−6

@@ -22,13 +22,13 @@ · 21 unmodified lines
22	22	#include "libavcodec/vp8dsp.h"
23	23	#include "vp8dsp.h"
24	24
25		−void ff_vp8_luma_dc_wht_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);
26		−void ff_vp8_luma_dc_wht_dc_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);
	25	+void ff_vp8_luma_dc_wht_armv6(int16_t block[4][4][16], int16_t dc[16]);
	26	+void ff_vp8_luma_dc_wht_dc_armv6(int16_t block[4][4][16], int16_t dc[16]);
27	27
28		−void ff_vp8_idct_add_armv6(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
29		−void ff_vp8_idct_dc_add_armv6(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
30		−void ff_vp8_idct_dc_add4y_armv6(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
31		−void ff_vp8_idct_dc_add4uv_armv6(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
	28	+void ff_vp8_idct_add_armv6(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
	29	+void ff_vp8_idct_dc_add_armv6(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
	30	+void ff_vp8_idct_dc_add4y_armv6(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
	31	+void ff_vp8_idct_dc_add4uv_armv6(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
32	32
33	33	VP8_LF(armv6);
34	34

libavcodec/arm/vp8dsp_init_neon.c+5−5

@@ -22,12 +22,12 @@ · 21 unmodified lines
22	22	#include "libavcodec/vp8dsp.h"
23	23	#include "vp8dsp.h"
24	24
25		−void ff_vp8_luma_dc_wht_neon(DCTELEM block[4][4][16], DCTELEM dc[16]);
	25	+void ff_vp8_luma_dc_wht_neon(int16_t block[4][4][16], int16_t dc[16]);
26	26
27		−void ff_vp8_idct_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
28		−void ff_vp8_idct_dc_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
29		−void ff_vp8_idct_dc_add4y_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
30		−void ff_vp8_idct_dc_add4uv_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
	27	+void ff_vp8_idct_add_neon(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
	28	+void ff_vp8_idct_dc_add_neon(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
	29	+void ff_vp8_idct_dc_add4y_neon(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
	30	+void ff_vp8_idct_dc_add4uv_neon(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
31	31
32	32	VP8_LF(neon);
33	33

libavcodec/asv.h+1−1

@@ -48,7 +48,7 @@ typedef struct ASV1Context{ · 47 unmodified lines
48	48	int mb_height;
49	49	int mb_width2;
50	50	int mb_height2;
51		− DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
	51	+ DECLARE_ALIGNED(16, int16_t, block)[6][64];
52	52	uint16_t intra_matrix[64];
53	53	int q_intra_matrix[64];
54	54	uint8_t *bitstream_buffer;

libavcodec/asvdec.c+4−4

@@ -97,7 +97,7 @@ static inline int asv2_get_level(GetBitContext *gb) · 96 unmodified lines
97	97	return code - 31;
98	98	}
99	99
100		−static inline int asv1_decode_block(ASV1Context *a, DCTELEM block[64])
	100	+static inline int asv1_decode_block(ASV1Context *a, int16_t block[64])
101	101	{
102	102	int i;
103	103
@@ -128,7 +128,7 @@ static inline int asv1_decode_block(ASV1Context *a, DCTELEM block[64]) · 24 unmodified lines
128	128	return 0;
129	129	}
130	130
131		−static inline int asv2_decode_block(ASV1Context *a, DCTELEM block[64])
	131	+static inline int asv2_decode_block(ASV1Context *a, int16_t block[64])
132	132	{
133	133	int i, count, ccp;
134	134
@@ -164,7 +164,7 @@ static inline int asv2_decode_block(ASV1Context *a, DCTELEM block[64]) · 29 unmodified lines
164	164	return 0;
165	165	}
166	166
167		−static inline int decode_mb(ASV1Context *a, DCTELEM block[6][64])
	167	+static inline int decode_mb(ASV1Context *a, int16_t block[6][64])
168	168	{
169	169	int i;
170	170
@@ -186,7 +186,7 @@ static inline int decode_mb(ASV1Context *a, DCTELEM block[6][64]) · 15 unmodified lines
186	186
187	187	static inline void idct_put(ASV1Context *a, int mb_x, int mb_y)
188	188	{
189		− DCTELEM (*block)[64] = a->block;
	189	+ int16_t (*block)[64] = a->block;
190	190	int linesize = a->picture.linesize[0];
191	191
192	192	uint8_t dest_y = a->picture.data[0] + (mb_y 16* linesize ) + mb_x * 16;

libavcodec/asvenc.c+4−4

@@ -55,7 +55,7 @@ static inline void asv2_put_level(PutBitContext *pb, int level){ · 54 unmodified lines
55	55	}
56	56	}
57	57
58		−static inline void asv1_encode_block(ASV1Context *a, DCTELEM block[64]){
	58	+static inline void asv1_encode_block(ASV1Context *a, int16_t block[64]){
59	59	int i;
60	60	int nc_count=0;
61	61
@@ -88,7 +88,7 @@ static inline void asv1_encode_block(ASV1Context *a, DCTELEM block[64]){ · 26 unmodified lines
88	88	put_bits(&a->pb, ff_asv_ccp_tab[16][1], ff_asv_ccp_tab[16][0]);
89	89	}
90	90
91		−static inline void asv2_encode_block(ASV1Context *a, DCTELEM block[64]){
	91	+static inline void asv2_encode_block(ASV1Context *a, int16_t block[64]){
92	92	int i;
93	93	int count=0;
94	94
@@ -129,7 +129,7 @@ static inline void asv2_encode_block(ASV1Context *a, DCTELEM block[64]){ · 34 unmodified lines
129	129
130	130	#define MAX_MB_SIZE (301616*3/2/8)
131	131
132		−static inline int encode_mb(ASV1Context *a, DCTELEM block[6][64]){
	132	+static inline int encode_mb(ASV1Context *a, int16_t block[6][64]){
133	133	int i;
134	134
135	135	if (a->pb.buf_end - a->pb.buf - (put_bits_count(&a->pb)>>3) < MAX_MB_SIZE) {
@@ -148,7 +148,7 @@ static inline int encode_mb(ASV1Context *a, DCTELEM block[6][64]){ · 12 unmodified lines
148	148	}
149	149
150	150	static inline void dct_get(ASV1Context *a, int mb_x, int mb_y){
151		− DCTELEM (*block)[64]= a->block;
	151	+ int16_t (*block)[64]= a->block;
152	152	int linesize= a->picture.linesize[0];
153	153	int i;
154	154

libavcodec/bfin/config_bfin.h+1−1

@@ -21,7 +21,7 @@ · 20 unmodified lines
21	21	low level assembler interface wrapper
22	22
23	23	DEFUN(put_pixels_clamped,mL1,
24		− (DCTELEM block, uint8_t dest, int line_size)):
	24	+ (int16_t block, uint8_t dest, int line_size)):
25	25
26	26	body
27	27

libavcodec/bfin/dsputil_bfin.c+3−3

@@ -27,20 +27,20 @@ · 26 unmodified lines
27	27
28	28	int off;
29	29
30		−static void bfin_idct_add (uint8_t dest, int line_size, DCTELEM block)
	30	+static void bfin_idct_add (uint8_t dest, int line_size, int16_t block)
31	31	{
32	32	ff_bfin_idct (block);
33	33	ff_bfin_add_pixels_clamped (block, dest, line_size);
34	34	}
35	35
36		−static void bfin_idct_put (uint8_t dest, int line_size, DCTELEM block)
	36	+static void bfin_idct_put (uint8_t dest, int line_size, int16_t block)
37	37	{
38	38	ff_bfin_idct (block);
39	39	ff_bfin_put_pixels_clamped (block, dest, line_size);
40	40	}
41	41
42	42
43		−static void bfin_clear_blocks (DCTELEM *blocks)
	43	+static void bfin_clear_blocks (int16_t *blocks)
44	44	{
45	45	// This is just a simple memset.
46	46	//

libavcodec/bfin/dsputil_bfin.h+11−10

@@ -24,8 +24,9 @@ · 23 unmodified lines
24	24	#ifndef AVCODEC_BFIN_DSPUTIL_BFIN_H
25	25	#define AVCODEC_BFIN_DSPUTIL_BFIN_H
26	26
	27	+#include <stdint.h>
	28	+
27	29	#include "config.h"
28		−#include "libavcodec/dsputil.h"
29	30
30	31	#if defined(__FDPIC__) && CONFIG_SRAM
31	32	#define attribute_l1_text __attribute__ ((l1_text))
@@ -35,15 +36,15 @@ · 3 unmodified lines
35	36	#define attribute_l1_data_b
36	37	#endif
37	38
38		−void ff_bfin_idct (DCTELEM *block) attribute_l1_text;
39		−void ff_bfin_fdct (DCTELEM *block) attribute_l1_text;
40		−void ff_bfin_vp3_idct (DCTELEM *block);
41		−void ff_bfin_vp3_idct_put (uint8_t dest, int line_size, DCTELEM block);
42		−void ff_bfin_vp3_idct_add (uint8_t dest, int line_size, DCTELEM block);
43		−void ff_bfin_add_pixels_clamped (const DCTELEM block, uint8_t dest, int line_size) attribute_l1_text;
44		−void ff_bfin_put_pixels_clamped (const DCTELEM block, uint8_t dest, int line_size) attribute_l1_text;
45		−void ff_bfin_diff_pixels (DCTELEM block, const uint8_t s1, const uint8_t *s2, int stride) attribute_l1_text;
46		−void ff_bfin_get_pixels (DCTELEM restrict block, const uint8_t pixels, int line_size) attribute_l1_text;
	39	+void ff_bfin_idct (int16_t *block) attribute_l1_text;
	40	+void ff_bfin_fdct (int16_t *block) attribute_l1_text;
	41	+void ff_bfin_vp3_idct (int16_t *block);
	42	+void ff_bfin_vp3_idct_put (uint8_t dest, int line_size, int16_t block);
	43	+void ff_bfin_vp3_idct_add (uint8_t dest, int line_size, int16_t block);
	44	+void ff_bfin_add_pixels_clamped (const int16_t block, uint8_t dest, int line_size) attribute_l1_text;
	45	+void ff_bfin_put_pixels_clamped (const int16_t block, uint8_t dest, int line_size) attribute_l1_text;
	46	+void ff_bfin_diff_pixels (int16_t block, const uint8_t s1, const uint8_t *s2, int stride) attribute_l1_text;
	47	+void ff_bfin_get_pixels (int16_t restrict block, const uint8_t pixels, int line_size) attribute_l1_text;
47	48	int ff_bfin_pix_norm1 (uint8_t * pix, int line_size) attribute_l1_text;
48	49	int ff_bfin_z_sad8x8 (uint8_t blk1, uint8_t blk2, int dsz, int line_size, int h) attribute_l1_text;
49	50	int ff_bfin_z_sad16x16 (uint8_t blk1, uint8_t blk2, int dsz, int line_size, int h) attribute_l1_text;

libavcodec/bfin/fdct_bfin.S+4−4

@@ -20,7 +20,7 @@ · 19 unmodified lines
20	20	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21	21	*/
22	22	/*
23		− void ff_bfin_fdct (DCTELEM *buf);
	23	+ void ff_bfin_fdct (int16_t *buf);
24	24
25	25	This implementation works only for 8x8 input. The range of input
26	26	must be -256 to 255 i.e. 8bit input represented in a 16bit data
@@ -61,9 +61,9 @@ Notation · 34 unmodified lines
61	61	Other registers used:
62	62	I0, I1, I2, I3, B0, B2, B3, M0, M1, L3 registers and LC0.
63	63
64		− Input - r0 - pointer to start of DCTELEM *block
	64	+ Input - r0 - pointer to start of int16_t *block
65	65
66		− Output - The DCT output coefficients in the DCTELEM *block
	66	+ Output - The DCT output coefficients in the int16_t *block
67	67
68	68	Register constraint:
69	69	This code is called from jpeg_encode.
@@ -145,7 +145,7 @@ vtmp: .space 128 · 75 unmodified lines
145	145
146	146	.text
147	147	DEFUN(fdct,mL1,
148		− (DCTELEM *block)):
	148	+ (int16_t *block)):
149	149	[--SP] = (R7:4, P5:3); // Push the registers onto the stack.
150	150
151	151	b0 = r0;

libavcodec/bfin/idct_bfin.S+2−2

@@ -22,7 +22,7 @@ · 21 unmodified lines
22	22	/*
23	23	This blackfin DSP code implements an 8x8 inverse type II DCT.
24	24
25		−Prototype : void ff_bfin_idct(DCTELEM *in)
	25	+Prototype : void ff_bfin_idct(int16_t *in)
26	26
27	27	Registers Used : A0, A1, R0-R7, I0-I3, B0, B2, B3, M0-M2, L0-L3, P0-P5, LC0.
28	28
@@ -90,7 +90,7 @@ vtmp: .space 256 · 61 unmodified lines
90	90
91	91	.text
92	92	DEFUN(idct,mL1,
93		− (DCTELEM *block)):
	93	+ (int16_t *block)):
94	94
95	95	/******************** Function Prologue *******************************/
96	96	link 16;

libavcodec/bfin/mpegvideo_bfin.c+1−1

@@ -26,7 +26,7 @@ · 25 unmodified lines
26	26	#include "dsputil_bfin.h"
27	27
28	28	static int dct_quantize_bfin (MpegEncContext *s,
29		− DCTELEM *block, int n,
	29	+ int16_t *block, int n,
30	30	int qscale, int *overflow)
31	31	{
32	32	int last_non_zero, q, start_i;

libavcodec/bfin/pixels_bfin.S+4−4

@@ -21,7 +21,7 @@ · 20 unmodified lines
21	21	#include "config_bfin.h"
22	22
23	23	DEFUN(put_pixels_clamped,mL1,
24		− (DCTELEM block, uint8_t dest, int line_size)):
	24	+ (int16_t block, uint8_t dest, int line_size)):
25	25	[--SP] = (R7:4);
26	26	R4 = 0;
27	27	R5.l = 0x00ff;
@@ -51,7 +51,7 @@ ppc$1: R2 = Max(R0, R4) (V) \|\| [I1++M1] = R6; · 23 unmodified lines
51	51	DEFUN_END(put_pixels_clamped)
52	52
53	53	DEFUN(add_pixels_clamped,mL1,
54		− (DCTELEM block, uint8_t dest, int line_size)):
	54	+ (int16_t block, uint8_t dest, int line_size)):
55	55	[-- SP] = (R7:4);
56	56	R4 = 0;
57	57	I0 = 0;
@@ -442,7 +442,7 @@ LE$8OT: DISALGNEXCPT \|\| R2 =[I1++] \|\| [I3++M2] = R5; · 384 unmodified lines
442	442	rts;
443	443
444	444	DEFUN(diff_pixels,mL1,
445		− (DCTELEM block, uint8_t s1, uint8_t *s2, int stride)):
	445	+ (int16_t block, uint8_t s1, uint8_t *s2, int stride)):
446	446	link 0;
447	447	[--sp] = (r7:4);
448	448	p0=8;
@@ -518,7 +518,7 @@ DEFUN_END(pix_sum) · 69 unmodified lines
518	518
519	519
520	520	DEFUN(get_pixels,mL1,
521		− (DCTELEM restrict block, const uint8_t pixels, int line_size)):
	521	+ (int16_t restrict block, const uint8_t pixels, int line_size)):
522	522	[--sp] = (r7:4);
523	523	i3=r0; // dest
524	524	i0=r1; // src0

libavcodec/bfin/vp3_bfin.c+2−2

@@ -23,7 +23,7 @@ · 22 unmodified lines
23	23	#include "dsputil_bfin.h"
24	24
25	25	/* Intra iDCT offset 128 */
26		−void ff_bfin_vp3_idct_put (uint8_t dest, int line_size, DCTELEM block)
	26	+void ff_bfin_vp3_idct_put (uint8_t dest, int line_size, int16_t block)
27	27	{
28	28	uint8_t *cm = ff_cropTbl + MAX_NEG_CROP + 128;
29	29	int i,j;
@@ -36,7 +36,7 @@ void ff_bfin_vp3_idct_put (uint8_t dest, int line_size, DCTELEM block) · 6 unmodified lines
36	36	}
37	37
38	38	/* Inter iDCT */
39		−void ff_bfin_vp3_idct_add (uint8_t dest, int line_size, DCTELEM block)
	39	+void ff_bfin_vp3_idct_add (uint8_t dest, int line_size, int16_t block)
40	40	{
41	41	ff_bfin_vp3_idct (block);
42	42	ff_bfin_add_pixels_clamped (block, dest, line_size);

libavcodec/bfin/vp3_idct_bfin.S+2−2

@@ -22,7 +22,7 @@ · 21 unmodified lines
22	22	/*
23	23	This blackfin DSP code implements an 8x8 inverse type II DCT.
24	24
25		−Prototype : void ff_bfin_vp3_idct(DCTELEM *in)
	25	+Prototype : void ff_bfin_vp3_idct(int16_t *in)
26	26
27	27	Registers Used : A0, A1, R0-R7, I0-I3, B0, B2, B3, M0-M2, L0-L3, P0-P5, LC0.
28	28
@@ -63,7 +63,7 @@ vtmp: .space 256 · 34 unmodified lines
63	63
64	64	.text
65	65	DEFUN(vp3_idct,mL1,
66		− (DCTELEM *block)):
	66	+ (int16_t *block)):
67	67
68	68	/******************** Function Prologue *******************************/
69	69	link 16;

libavcodec/bink.c+3−3

@@ -696,7 +696,7 @@ static int read_dct_coeffs(GetBitContext gb, int32_t block[64], const uint8_t · 695 unmodified lines
696	696	* @param masks_count number of masks to decode
697	697	* @return 0 on success, negative value in other cases
698	698	*/
699		−static int read_residue(GetBitContext *gb, DCTELEM block[64], int masks_count)
	699	+static int read_residue(GetBitContext *gb, int16_t block[64], int masks_count)
700	700	{
701	701	int coef_list[128];
702	702	int mode_list[128];
@@ -800,7 +800,7 @@ static int binkb_decode_plane(BinkContext c, GetBitContext gb, int plane_idx, · 97 unmodified lines
800	800	int v, col[2];
801	801	const uint8_t *scan;
802	802	int xoff, yoff;
803		− LOCAL_ALIGNED_16(DCTELEM, block, [64]);
	803	+ LOCAL_ALIGNED_16(int16_t, block, [64]);
804	804	LOCAL_ALIGNED_16(int32_t, dctblock, [64]);
805	805	int coordmap[64];
806	806	int ybias = is_key ? -15 : 0;
@@ -946,7 +946,7 @@ static int bink_decode_plane(BinkContext c, GetBitContext gb, int plane_idx, · 139 unmodified lines
946	946	int v, col[2];
947	947	const uint8_t *scan;
948	948	int xoff, yoff;
949		− LOCAL_ALIGNED_16(DCTELEM, block, [64]);
	949	+ LOCAL_ALIGNED_16(int16_t, block, [64]);
950	950	LOCAL_ALIGNED_16(uint8_t, ublock, [64]);
951	951	LOCAL_ALIGNED_16(int32_t, dctblock, [64]);
952	952	int coordmap[64];

libavcodec/cavs.c+1−1

@@ -713,7 +713,7 @@ void ff_cavs_init_top_lines(AVSContext *h) { · 712 unmodified lines
713	713	/* alloc space for co-located MVs and types */
714	714	h->col_mv = av_malloc( h->mb_widthh->mb_height4*sizeof(cavs_vector));
715	715	h->col_type_base = av_malloc(h->mb_width*h->mb_height);
716		− h->block = av_mallocz(64*sizeof(DCTELEM));
	716	+ h->block = av_mallocz(64*sizeof(int16_t));
717	717	}
718	718
719	719	av_cold int ff_cavs_init(AVCodecContext *avctx) {

libavcodec/cavs.h+1−1

@@ -234,7 +234,7 @@ typedef struct AVSContext { · 233 unmodified lines
234	234	uint8_t *edge_emu_buffer;
235	235
236	236	int got_keyframe;
237		− DCTELEM *block;
	237	+ int16_t *block;
238	238	} AVSContext;
239	239
240	240	extern const uint8_t ff_cavs_partition_flags[30];

libavcodec/cavsdec.c+4−4

@@ -516,8 +516,8 @@ static inline int get_ue_code(GetBitContext *gb, int order) · 515 unmodified lines
516	516	return get_ue_golomb(gb);
517	517	}
518	518
519		−static inline int dequant(AVSContext h, DCTELEM level_buf, uint8_t *run_buf,
520		− DCTELEM *dst, int mul, int shift, int coeff_num)
	519	+static inline int dequant(AVSContext h, int16_t level_buf, uint8_t *run_buf,
	520	+ int16_t *dst, int mul, int shift, int coeff_num)
521	521	{
522	522	int round = 1 << (shift - 1);
523	523	int pos = -1;
@@ -551,9 +551,9 @@ static int decode_residual_block(AVSContext h, GetBitContext gb, · 27 unmodified lines
551	551	int qp, uint8_t *dst, int stride)
552	552	{
553	553	int i, level_code, esc_code, level, run, mask;
554		− DCTELEM level_buf[65];
	554	+ int16_t level_buf[65];
555	555	uint8_t run_buf[65];
556		− DCTELEM *block = h->block;
	556	+ int16_t *block = h->block;
557	557
558	558	for (i = 0;i < 65; i++) {
559	559	level_code = get_ue_code(gb, r->golomb_order);

libavcodec/cavsdsp.c+2−2

@@ -183,9 +183,9 @@ static void cavs_filter_ch_c(uint8_t *d, int stride, int alpha, int beta, int tc · 182 unmodified lines
183	183	*
184	184	****************************************************************************/
185	185
186		−static void cavs_idct8_add_c(uint8_t dst, DCTELEM block, int stride) {
	186	+static void cavs_idct8_add_c(uint8_t dst, int16_t block, int stride) {
187	187	int i;
188		− DCTELEM (src)[8] = (DCTELEM()[8])block;
	188	+ int16_t (src)[8] = (int16_t()[8])block;
189	189	uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
190	190
191	191	src[0][0] += 8;

libavcodec/cavsdsp.h+1−1

@@ -32,7 +32,7 @@ typedef struct CAVSDSPContext { · 31 unmodified lines
32	32	void (cavs_filter_lh)(uint8_t pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
33	33	void (cavs_filter_cv)(uint8_t pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
34	34	void (cavs_filter_ch)(uint8_t pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
35		− void (cavs_idct8_add)(uint8_t dst, DCTELEM *block, int stride);
	35	+ void (cavs_idct8_add)(uint8_t dst, int16_t *block, int stride);
36	36	int idct_perm;
37	37	} CAVSDSPContext;
38	38

libavcodec/dct-test.c+16−15

@@ -39,6 +39,7 @@ · 38 unmodified lines
39	39	#include "libavutil/lfg.h"
40	40	#include "libavutil/time.h"
41	41
	42	+#include "dsputil.h"
42	43	#include "simple_idct.h"
43	44	#include "aandcttab.h"
44	45	#include "faandct.h"
@@ -49,24 +50,24 @@ · 4 unmodified lines
49	50	#undef printf
50	51
51	52	// BFIN
52		−void ff_bfin_idct(DCTELEM *block);
53		−void ff_bfin_fdct(DCTELEM *block);
	53	+void ff_bfin_idct(int16_t *block);
	54	+void ff_bfin_fdct(int16_t *block);
54	55
55	56	// ALTIVEC
56		−void ff_fdct_altivec(DCTELEM *block);
	57	+void ff_fdct_altivec(int16_t *block);
57	58
58	59	// ARM
59		−void ff_j_rev_dct_arm(DCTELEM *data);
60		−void ff_simple_idct_arm(DCTELEM *data);
61		−void ff_simple_idct_armv5te(DCTELEM *data);
62		−void ff_simple_idct_armv6(DCTELEM *data);
63		−void ff_simple_idct_neon(DCTELEM *data);
	60	+void ff_j_rev_dct_arm(int16_t *data);
	61	+void ff_simple_idct_arm(int16_t *data);
	62	+void ff_simple_idct_armv5te(int16_t *data);
	63	+void ff_simple_idct_armv6(int16_t *data);
	64	+void ff_simple_idct_neon(int16_t *data);
64	65
65		−void ff_simple_idct_axp(DCTELEM *data);
	66	+void ff_simple_idct_axp(int16_t *data);
66	67
67	68	struct algo {
68	69	const char *name;
69		− void (func)(DCTELEM block);
	70	+ void (func)(int16_t block);
70	71	enum formattag { NO_PERM, MMX_PERM, MMX_SIMPLE_PERM, SCALE_PERM,
71	72	SSE2_PERM, PARTTRANS_PERM } format;
72	73	int mm_support;
@@ -166,10 +167,10 @@ static void idct_mmx_init(void) · 93 unmodified lines
166	167	}
167	168	}
168	169
169		−DECLARE_ALIGNED(16, static DCTELEM, block)[64];
170		−DECLARE_ALIGNED(8, static DCTELEM, block1)[64];
	170	+DECLARE_ALIGNED(16, static int16_t, block)[64];
	171	+DECLARE_ALIGNED(8, static int16_t, block1)[64];
171	172
172		−static void init_block(DCTELEM block[64], int test, int is_idct, AVLFG *prng)
	173	+static void init_block(int16_t block[64], int test, int is_idct, AVLFG *prng)
173	174	{
174	175	int i, j;
175	176
@@ -197,7 +198,7 @@ static void init_block(DCTELEM block[64], int test, int is_idct, AVLFG *prng) · 21 unmodified lines
197	198	}
198	199	}
199	200
200		−static void permute(DCTELEM dst[64], const DCTELEM src[64], int perm)
	201	+static void permute(int16_t dst[64], const int16_t src[64], int perm)
201	202	{
202	203	int i;
203	204
@@ -221,7 +222,7 @@ static void permute(DCTELEM dst[64], const DCTELEM src[64], int perm) · 17 unmodified lines
221	222
222	223	static int dct_error(const struct algo *dct, int test, int is_idct, int speed)
223	224	{
224		− void (ref)(DCTELEM block) = is_idct ? ff_ref_idct : ff_ref_fdct;
	225	+ void (ref)(int16_t block) = is_idct ? ff_ref_idct : ff_ref_fdct;
225	226	int it, i, scale;
226	227	int err_inf, v;
227	228	int64_t err2, ti, ti1, it1, err_sum = 0;

libavcodec/dnxhddec.c+7−7

@@ -44,19 +44,19 @@ typedef struct DNXHDContext { · 43 unmodified lines
44	44	VLC ac_vlc, dc_vlc, run_vlc;
45	45	int last_dc[3];
46	46	DSPContext dsp;
47		− DECLARE_ALIGNED(16, DCTELEM, blocks)[8][64];
	47	+ DECLARE_ALIGNED(16, int16_t, blocks)[8][64];
48	48	ScanTable scantable;
49	49	const CIDEntry *cid_table;
50	50	int bit_depth; // 8, 10 or 0 if not initialized at all.
51		− void (decode_dct_block)(struct DNXHDContext ctx, DCTELEM *block,
	51	+ void (decode_dct_block)(struct DNXHDContext ctx, int16_t *block,
52	52	int n, int qscale);
53	53	} DNXHDContext;
54	54
55	55	#define DNXHD_VLC_BITS 9
56	56	#define DNXHD_DC_VLC_BITS 7
57	57
58		−static void dnxhd_decode_dct_block_8(DNXHDContext ctx, DCTELEM block, int n, int qscale);
59		−static void dnxhd_decode_dct_block_10(DNXHDContext ctx, DCTELEM block, int n, int qscale);
	58	+static void dnxhd_decode_dct_block_8(DNXHDContext ctx, int16_t block, int n, int qscale);
	59	+static void dnxhd_decode_dct_block_10(DNXHDContext ctx, int16_t block, int n, int qscale);
60	60
61	61	static av_cold int dnxhd_decode_init(AVCodecContext *avctx)
62	62	{
@@ -180,7 +180,7 @@ static int dnxhd_decode_header(DNXHDContext ctx, const uint8_t buf, int buf_si · 117 unmodified lines
180	180	}
181	181
182	182	static av_always_inline void dnxhd_decode_dct_block(DNXHDContext *ctx,
183		− DCTELEM *block, int n,
	183	+ int16_t *block, int n,
184	184	int qscale,
185	185	int index_bits,
186	186	int level_bias,
@@ -250,13 +250,13 @@ static av_always_inline void dnxhd_decode_dct_block(DNXHDContext *ctx, · 63 unmodified lines
250	250	CLOSE_READER(bs, &ctx->gb);
251	251	}
252	252
253		−static void dnxhd_decode_dct_block_8(DNXHDContext ctx, DCTELEM block,
	253	+static void dnxhd_decode_dct_block_8(DNXHDContext ctx, int16_t block,
254	254	int n, int qscale)
255	255	{
256	256	dnxhd_decode_dct_block(ctx, block, n, qscale, 4, 32, 6);
257	257	}
258	258
259		−static void dnxhd_decode_dct_block_10(DNXHDContext ctx, DCTELEM block,
	259	+static void dnxhd_decode_dct_block_10(DNXHDContext ctx, int16_t block,
260	260	int n, int qscale)
261	261	{
262	262	dnxhd_decode_dct_block(ctx, block, n, qscale, 6, 8, 4);

libavcodec/dnxhdenc.c+10−10

@@ -44,7 +44,7 @@ static const AVClass class = { "dnxhd", av_default_item_name, options, LIBAVUTIL · 43 unmodified lines
44	44
45	45	#define LAMBDA_FRAC_BITS 10
46	46
47		−static void dnxhd_8bit_get_pixels_8x4_sym(DCTELEM restrict block, const uint8_t pixels, int line_size)
	47	+static void dnxhd_8bit_get_pixels_8x4_sym(int16_t restrict block, const uint8_t pixels, int line_size)
48	48	{
49	49	int i;
50	50	for (i = 0; i < 4; i++) {
@@ -61,7 +61,7 @@ static void dnxhd_8bit_get_pixels_8x4_sym(DCTELEM *restrict block, const uint8_t · 10 unmodified lines
61	61	memcpy(block + 24, block - 32, sizeof(block) 8);
62	62	}
63	63
64		−static av_always_inline void dnxhd_10bit_get_pixels_8x4_sym(DCTELEM restrict block, const uint8_t pixels, int line_size)
	64	+static av_always_inline void dnxhd_10bit_get_pixels_8x4_sym(int16_t restrict block, const uint8_t pixels, int line_size)
65	65	{
66	66	int i;
67	67
@@ -73,7 +73,7 @@ static av_always_inline void dnxhd_10bit_get_pixels_8x4_sym(DCTELEM *restrict bl · 5 unmodified lines
73	73	}
74	74	}
75	75
76		−static int dnxhd_10bit_dct_quantize(MpegEncContext ctx, DCTELEM block,
	76	+static int dnxhd_10bit_dct_quantize(MpegEncContext ctx, int16_t block,
77	77	int n, int qscale, int *overflow)
78	78	{
79	79	const uint8_t *scantable= ctx->intra_scantable.scantable;
@@ -370,7 +370,7 @@ static av_always_inline void dnxhd_encode_dc(DNXHDEncContext *ctx, int diff) · 290 unmodified lines
370	370	(ctx->cid_table->dc_codes[nbits]<<nbits) + (diff & ((1 << nbits) - 1)));
371	371	}
372	372
373		−static av_always_inline void dnxhd_encode_block(DNXHDEncContext ctx, DCTELEM block, int last_index, int n)
	373	+static av_always_inline void dnxhd_encode_block(DNXHDEncContext ctx, int16_t block, int last_index, int n)
374	374	{
375	375	int last_non_zero = 0;
376	376	int slevel, i, j;
@@ -393,7 +393,7 @@ static av_always_inline void dnxhd_encode_block(DNXHDEncContext ctx, DCTELEM b · 16 unmodified lines
393	393	put_bits(&ctx->m.pb, ctx->vlc_bits[0], ctx->vlc_codes[0]); // EOB
394	394	}
395	395
396		−static av_always_inline void dnxhd_unquantize_c(DNXHDEncContext ctx, DCTELEM block, int n, int qscale, int last_index)
	396	+static av_always_inline void dnxhd_unquantize_c(DNXHDEncContext ctx, int16_t block, int n, int qscale, int last_index)
397	397	{
398	398	const uint8_t *weight_matrix;
399	399	int level;
@@ -434,7 +434,7 @@ static av_always_inline void dnxhd_unquantize_c(DNXHDEncContext ctx, DCTELEM b · 34 unmodified lines
434	434	}
435	435	}
436	436
437		−static av_always_inline int dnxhd_ssd_block(DCTELEM qblock, DCTELEM block)
	437	+static av_always_inline int dnxhd_ssd_block(int16_t qblock, int16_t block)
438	438	{
439	439	int score = 0;
440	440	int i;
@@ -443,7 +443,7 @@ static av_always_inline int dnxhd_ssd_block(DCTELEM qblock, DCTELEM block) · 2 unmodified lines
443	443	return score;
444	444	}
445	445
446		−static av_always_inline int dnxhd_calc_ac_bits(DNXHDEncContext ctx, DCTELEM block, int last_index)
	446	+static av_always_inline int dnxhd_calc_ac_bits(DNXHDEncContext ctx, int16_t block, int last_index)
447	447	{
448	448	int last_non_zero = 0;
449	449	int bits = 0;
@@ -512,7 +512,7 @@ static int dnxhd_calc_bits_thread(AVCodecContext avctx, void arg, int jobnr, i · 62 unmodified lines
512	512	DNXHDEncContext *ctx = avctx->priv_data;
513	513	int mb_y = jobnr, mb_x;
514	514	int qscale = ctx->qscale;
515		− LOCAL_ALIGNED_16(DCTELEM, block, [64]);
	515	+ LOCAL_ALIGNED_16(int16_t, block, [64]);
516	516	ctx = ctx->thread[threadnr];
517	517
518	518	ctx->m.last_dc[0] =
@@ -529,7 +529,7 @@ static int dnxhd_calc_bits_thread(AVCodecContext avctx, void arg, int jobnr, i · 10 unmodified lines
529	529	dnxhd_get_blocks(ctx, mb_x, mb_y);
530	530
531	531	for (i = 0; i < 8; i++) {
532		− DCTELEM *src_block = ctx->blocks[i];
	532	+ int16_t *src_block = ctx->blocks[i];
533	533	int overflow, nbits, diff, last_index;
534	534	int n = dnxhd_switch_matrix(ctx, i);
535	535
@@ -578,7 +578,7 @@ static int dnxhd_encode_thread(AVCodecContext avctx, void arg, int jobnr, int · 42 unmodified lines
578	578	dnxhd_get_blocks(ctx, mb_x, mb_y);
579	579
580	580	for (i = 0; i < 8; i++) {
581		− DCTELEM *block = ctx->blocks[i];
	581	+ int16_t *block = ctx->blocks[i];
582	582	int overflow, n = dnxhd_switch_matrix(ctx, i);
583	583	int last_index = ctx->m.dct_quantize(&ctx->m, block, i,
584	584	qscale, &overflow);

libavcodec/dnxhdenc.h+2−2

@@ -64,7 +64,7 @@ typedef struct DNXHDEncContext { · 63 unmodified lines
64	64	int nitris_compat;
65	65	unsigned min_padding;
66	66
67		− DECLARE_ALIGNED(16, DCTELEM, blocks)[8][64];
	67	+ DECLARE_ALIGNED(16, int16_t, blocks)[8][64];
68	68
69	69	int (*qmatrix_c) [64];
70	70	int (*qmatrix_l) [64];
@@ -92,7 +92,7 @@ typedef struct DNXHDEncContext { · 21 unmodified lines
92	92	RCCMPEntry *mb_cmp;
93	93	RCEntry (*mb_rc)[8160];
94	94
95		− void (get_pixels_8x4_sym)(DCTELEM /align 16/, const uint8_t *, int);
	95	+ void (get_pixels_8x4_sym)(int16_t /align 16/, const uint8_t *, int);
96	96	} DNXHDEncContext;
97	97
98	98	void ff_dnxhdenc_init_x86(DNXHDEncContext *ctx);

libavcodec/dsputil.c+17−17

@@ -336,7 +336,7 @@ static int sse16_c(void v, uint8_t pix1, uint8_t *pix2, int line_size, int h) · 335 unmodified lines
336	336	return s;
337	337	}
338	338
339		−static void diff_pixels_c(DCTELEM restrict block, const uint8_t s1,
	339	+static void diff_pixels_c(int16_t restrict block, const uint8_t s1,
340	340	const uint8_t *s2, int stride){
341	341	int i;
342	342
@@ -357,7 +357,7 @@ static void diff_pixels_c(DCTELEM restrict block, const uint8_t s1, · 14 unmodified lines
357	357	}
358	358
359	359
360		−static void put_pixels_clamped_c(const DCTELEM block, uint8_t restrict pixels,
	360	+static void put_pixels_clamped_c(const int16_t block, uint8_t restrict pixels,
361	361	int line_size)
362	362	{
363	363	int i;
@@ -378,7 +378,7 @@ static void put_pixels_clamped_c(const DCTELEM block, uint8_t restrict pixels, · 14 unmodified lines
378	378	}
379	379	}
380	380
381		−static void put_signed_pixels_clamped_c(const DCTELEM *block,
	381	+static void put_signed_pixels_clamped_c(const int16_t *block,
382	382	uint8_t *restrict pixels,
383	383	int line_size)
384	384	{
@@ -399,7 +399,7 @@ static void put_signed_pixels_clamped_c(const DCTELEM *block, · 14 unmodified lines
399	399	}
400	400	}
401	401
402		−static void add_pixels_clamped_c(const DCTELEM block, uint8_t restrict pixels,
	402	+static void add_pixels_clamped_c(const int16_t block, uint8_t restrict pixels,
403	403	int line_size)
404	404	{
405	405	int i;
@@ -419,7 +419,7 @@ static void add_pixels_clamped_c(const DCTELEM block, uint8_t restrict pixels, · 13 unmodified lines
419	419	}
420	420	}
421	421
422		−static int sum_abs_dctelem_c(DCTELEM *block)
	422	+static int sum_abs_dctelem_c(int16_t *block)
423	423	{
424	424	int sum=0, i;
425	425	for(i=0; i<64; i++)
@@ -1712,10 +1712,10 @@ static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){ · 1286 unmodified lines
1712	1712	* @param scantable the used scantable, this is only used to speed the permutation up, the block is not
1713	1713	* (inverse) permutated to scantable order!
1714	1714	*/
1715		−void ff_block_permute(DCTELEM block, uint8_t permutation, const uint8_t *scantable, int last)
	1715	+void ff_block_permute(int16_t block, uint8_t permutation, const uint8_t *scantable, int last)
1716	1716	{
1717	1717	int i;
1718		− DCTELEM temp[64];
	1718	+ int16_t temp[64];
1719	1719
1720	1720	if(last<=0) return;
1721	1721	//if(permutation[1]==1) return; //FIXME it is ok but not clean and might fail for some permutations
@@ -2030,7 +2030,7 @@ static int hadamard8_intra8x8_c(/MpegEncContext/ void s, uint8_t src, uint8_ · 308 unmodified lines
2030	2030
2031	2031	static int dct_sad8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int stride, int h){
2032	2032	MpegEncContext * const s= (MpegEncContext *)c;
2033		− LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
	2033	+ LOCAL_ALIGNED_16(int16_t, temp, [64]);
2034	2034
2035	2035	assert(h==8);
2036	2036
@@ -2069,7 +2069,7 @@ static int dct_sad8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2 · 32 unmodified lines
2069	2069
2070	2070	static int dct264_sad8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int stride, int h){
2071	2071	MpegEncContext * const s= (MpegEncContext *)c;
2072		− DCTELEM dct[8][8];
	2072	+ int16_t dct[8][8];
2073	2073	int i;
2074	2074	int sum=0;
2075	2075
@@ -2094,7 +2094,7 @@ static int dct264_sad8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *s · 18 unmodified lines
2094	2094
2095	2095	static int dct_max8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int stride, int h){
2096	2096	MpegEncContext * const s= (MpegEncContext *)c;
2097		− LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
	2097	+ LOCAL_ALIGNED_16(int16_t, temp, [64]);
2098	2098	int sum=0, i;
2099	2099
2100	2100	assert(h==8);
@@ -2110,8 +2110,8 @@ static int dct_max8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2 · 9 unmodified lines
2110	2110
2111	2111	static int quant_psnr8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int stride, int h){
2112	2112	MpegEncContext * const s= (MpegEncContext *)c;
2113		− LOCAL_ALIGNED_16(DCTELEM, temp, [64*2]);
2114		− DCTELEM * const bak = temp+64;
	2113	+ LOCAL_ALIGNED_16(int16_t, temp, [64*2]);
	2114	+ int16_t * const bak = temp+64;
2115	2115	int sum=0, i;
2116	2116
2117	2117	assert(h==8);
@@ -2119,7 +2119,7 @@ static int quant_psnr8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *s · 1 unmodified line
2119	2119
2120	2120	s->dsp.diff_pixels(temp, src1, src2, stride);
2121	2121
2122		− memcpy(bak, temp, 64*sizeof(DCTELEM));
	2122	+ memcpy(bak, temp, 64*sizeof(int16_t));
2123	2123
2124	2124	s->block_last_index[0/FIXME/]= s->fast_dct_quantize(s, temp, 0/FIXME/, s->qscale, &i);
2125	2125	s->dct_unquantize_inter(s, temp, 0, s->qscale);
@@ -2134,7 +2134,7 @@ static int quant_psnr8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *s · 8 unmodified lines
2134	2134	static int rd8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int stride, int h){
2135	2135	MpegEncContext * const s= (MpegEncContext *)c;
2136	2136	const uint8_t *scantable= s->intra_scantable.permutated;
2137		− LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
	2137	+ LOCAL_ALIGNED_16(int16_t, temp, [64]);
2138	2138	LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]);
2139	2139	LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]);
2140	2140	int i, last, run, bits, level, distortion, start_i;
@@ -2210,7 +2210,7 @@ static int rd8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int · 69 unmodified lines
2210	2210	static int bit8x8_c(/MpegEncContext/ void c, uint8_t src1, uint8_t *src2, int stride, int h){
2211	2211	MpegEncContext * const s= (MpegEncContext *)c;
2212	2212	const uint8_t *scantable= s->intra_scantable.permutated;
2213		− LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
	2213	+ LOCAL_ALIGNED_16(int16_t, temp, [64]);
2214	2214	int i, last, run, bits, level, start_i;
2215	2215	const int esc_length= s->ac_esc_length;
2216	2216	uint8_t * length;
@@ -2447,12 +2447,12 @@ static void vector_clip_int32_c(int32_t dst, const int32_t src, int32_t min, · 230 unmodified lines
2447	2447	} while (len > 0);
2448	2448	}
2449	2449
2450		−static void ff_jref_idct_put(uint8_t dest, int line_size, DCTELEM block)
	2450	+static void ff_jref_idct_put(uint8_t dest, int line_size, int16_t block)
2451	2451	{
2452	2452	ff_j_rev_dct (block);
2453	2453	put_pixels_clamped_c(block, dest, line_size);
2454	2454	}
2455		−static void ff_jref_idct_add(uint8_t dest, int line_size, DCTELEM block)
	2455	+static void ff_jref_idct_add(uint8_t dest, int line_size, int16_t block)
2456	2456	{
2457	2457	ff_j_rev_dct (block);
2458	2458	add_pixels_clamped_c(block, dest, line_size);

libavcodec/dsputil.h+45−46

@@ -36,41 +36,40 @@ · 35 unmodified lines
36	36
37	37	//#define DEBUG
38	38	/* dct code */
39		−typedef short DCTELEM;
40	39
41		−void ff_fdct_ifast (DCTELEM *data);
42		−void ff_fdct_ifast248 (DCTELEM *data);
43		−void ff_jpeg_fdct_islow_8(DCTELEM *data);
44		−void ff_jpeg_fdct_islow_10(DCTELEM *data);
45		−void ff_fdct248_islow_8(DCTELEM *data);
46		−void ff_fdct248_islow_10(DCTELEM *data);
	40	+void ff_fdct_ifast(int16_t *data);
	41	+void ff_fdct_ifast248(int16_t *data);
	42	+void ff_jpeg_fdct_islow_8(int16_t *data);
	43	+void ff_jpeg_fdct_islow_10(int16_t *data);
	44	+void ff_fdct248_islow_8(int16_t *data);
	45	+void ff_fdct248_islow_10(int16_t *data);
47	46
48		−void ff_j_rev_dct (DCTELEM *data);
	47	+void ff_j_rev_dct(int16_t *data);
49	48
50		−void ff_fdct_mmx(DCTELEM *block);
51		−void ff_fdct_mmxext(DCTELEM *block);
52		−void ff_fdct_sse2(DCTELEM *block);
	49	+void ff_fdct_mmx(int16_t *block);
	50	+void ff_fdct_mmxext(int16_t *block);
	51	+void ff_fdct_sse2(int16_t *block);
53	52
54	53	#define H264_IDCT(depth) \
55		−void ff_h264_idct8_add_ ## depth ## _c(uint8_t dst, DCTELEM block, int stride);\
56		−void ff_h264_idct_add_ ## depth ## _c(uint8_t dst, DCTELEM block, int stride);\
57		−void ff_h264_idct8_dc_add_ ## depth ## _c(uint8_t dst, DCTELEM block, int stride);\
58		−void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t dst, DCTELEM block, int stride);\
59		−void ff_h264_idct_add16_ ## depth ## _c(uint8_t dst, const int blockoffset, DCTELEM block, int stride, const uint8_t nnzc[68]);\
60		−void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t dst, const int blockoffset, DCTELEM block, int stride, const uint8_t nnzc[68]);\
61		−void ff_h264_idct8_add4_ ## depth ## _c(uint8_t dst, const int blockoffset, DCTELEM block, int stride, const uint8_t nnzc[68]);\
62		−void ff_h264_idct_add8_422_ ## depth ## _c(uint8_t *dest, const int blockoffset, DCTELEM block, int stride, const uint8_t nnzc[68]);\
63		−void ff_h264_idct_add8_ ## depth ## _c(uint8_t *dest, const int blockoffset, DCTELEM block, int stride, const uint8_t nnzc[68]);\
64		−void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(DCTELEM output, DCTELEM input, int qmul);\
65		−void ff_h264_chroma422_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);\
66		−void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);
	54	+void ff_h264_idct8_add_ ## depth ## _c(uint8_t dst, int16_t block, int stride);\
	55	+void ff_h264_idct_add_ ## depth ## _c(uint8_t dst, int16_t block, int stride);\
	56	+void ff_h264_idct8_dc_add_ ## depth ## _c(uint8_t dst, int16_t block, int stride);\
	57	+void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t dst, int16_t block, int stride);\
	58	+void ff_h264_idct_add16_ ## depth ## _c(uint8_t dst, const int blockoffset, int16_t block, int stride, const uint8_t nnzc[68]);\
	59	+void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t dst, const int blockoffset, int16_t block, int stride, const uint8_t nnzc[68]);\
	60	+void ff_h264_idct8_add4_ ## depth ## _c(uint8_t dst, const int blockoffset, int16_t block, int stride, const uint8_t nnzc[68]);\
	61	+void ff_h264_idct_add8_422_ ## depth ## _c(uint8_t *dest, const int blockoffset, int16_t block, int stride, const uint8_t nnzc[68]);\
	62	+void ff_h264_idct_add8_ ## depth ## _c(uint8_t *dest, const int blockoffset, int16_t block, int stride, const uint8_t nnzc[68]);\
	63	+void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(int16_t output, int16_t input, int qmul);\
	64	+void ff_h264_chroma422_dc_dequant_idct_ ## depth ## _c(int16_t *block, int qmul);\
	65	+void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(int16_t *block, int qmul);
67	66
68	67	H264_IDCT( 8)
69	68	H264_IDCT( 9)
70	69	H264_IDCT(10)
71	70
72		−void ff_svq3_luma_dc_dequant_idct_c(DCTELEM output, DCTELEM input, int qp);
73		−void ff_svq3_add_idct_c(uint8_t dst, DCTELEM block, int stride, int qp, int dc);
	71	+void ff_svq3_luma_dc_dequant_idct_c(int16_t output, int16_t input, int qp);
	72	+void ff_svq3_add_idct_c(uint8_t dst, int16_t block, int stride, int qp, int dc);
74	73
75	74	/* encoding scans */
76	75	extern const uint8_t ff_alternate_horizontal_scan[64];
@@ -128,11 +127,11 @@ could be reached easily ... · 51 unmodified lines
128	127	*/
129	128
130	129	/*
131		−void get_pixels_c(DCTELEM block, const uint8_t pixels, int line_size);
132		−void diff_pixels_c(DCTELEM block, const uint8_t s1, const uint8_t *s2, int stride);
133		−void put_pixels_clamped_c(const DCTELEM block, uint8_t pixels, int line_size);
134		−void add_pixels_clamped_c(const DCTELEM block, uint8_t pixels, int line_size);
135		−void clear_blocks_c(DCTELEM *blocks);
	130	+void get_pixels_c(int16_t block, const uint8_t pixels, int line_size);
	131	+void diff_pixels_c(int16_t block, const uint8_t s1, const uint8_t *s2, int stride);
	132	+void put_pixels_clamped_c(const int16_t block, uint8_t pixels, int line_size);
	133	+void add_pixels_clamped_c(const int16_t block, uint8_t pixels, int line_size);
	134	+void clear_blocks_c(int16_t *blocks);
136	135	*/
137	136
138	137	/* add and put pixel (decoding) */
@@ -197,14 +196,14 @@ typedef struct DSPContext { · 58 unmodified lines
197	196	int dct_bits;
198	197
199	198	/* pixel ops : interface with DCT */
200		− void (get_pixels)(DCTELEM block/align 16/, const uint8_t pixels/align 8*/, int line_size);
201		− void (diff_pixels)(DCTELEM block/align 16/, const uint8_t s1/align 8/, const uint8_t s2/align 8/, int stride);
202		− void (put_pixels_clamped)(const DCTELEM block/align 16/, uint8_t pixels/align 8*/, int line_size);
203		− void (put_signed_pixels_clamped)(const DCTELEM block/align 16/, uint8_t pixels/align 8*/, int line_size);
204		− void (add_pixels_clamped)(const DCTELEM block/align 16/, uint8_t pixels/align 8*/, int line_size);
205		− void (add_pixels8)(uint8_t pixels, DCTELEM *block, int line_size);
206		− void (add_pixels4)(uint8_t pixels, DCTELEM *block, int line_size);
207		− int (sum_abs_dctelem)(DCTELEM block/align 16/);
	199	+ void (get_pixels)(int16_t block/align 16/, const uint8_t pixels/align 8*/, int line_size);
	200	+ void (diff_pixels)(int16_t block/align 16/, const uint8_t s1/align 8/, const uint8_t s2/align 8/, int stride);
	201	+ void (put_pixels_clamped)(const int16_t block/align 16/, uint8_t pixels/align 8*/, int line_size);
	202	+ void (put_signed_pixels_clamped)(const int16_t block/align 16/, uint8_t pixels/align 8*/, int line_size);
	203	+ void (add_pixels_clamped)(const int16_t block/align 16/, uint8_t pixels/align 8*/, int line_size);
	204	+ void (add_pixels8)(uint8_t pixels, int16_t *block, int line_size);
	205	+ void (add_pixels4)(uint8_t pixels, int16_t *block, int line_size);
	206	+ int (sum_abs_dctelem)(int16_t block/align 16/);
208	207	/**
209	208	* translational global motion compensation.
210	209	*/
@@ -214,8 +213,8 @@ typedef struct DSPContext { · 3 unmodified lines
214	213	*/
215	214	void (gmc )(uint8_t dst/align 8/, uint8_t src/align 1*/, int stride, int h, int ox, int oy,
216	215	int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
217		− void (clear_block)(DCTELEM block/align 16/);
218		− void (clear_blocks)(DCTELEM blocks/align 16/);
	216	+ void (clear_block)(int16_t block/align 16/);
	217	+ void (clear_blocks)(int16_t blocks/align 16/);
219	218	int (pix_sum)(uint8_t pix, int line_size);
220	219	int (pix_norm1)(uint8_t pix, int line_size);
221	220	// 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4
@@ -344,24 +343,24 @@ typedef struct DSPContext { · 122 unmodified lines
344	343	void (vector_clipf)(float dst /* align 16 /, const float src /* align 16 /, float min, float max, int len / align 16 */);
345	344
346	345	/* (I)DCT */
347		− void (fdct)(DCTELEM block/* align 16*/);
348		− void (fdct248)(DCTELEM block/* align 16*/);
	346	+ void (fdct)(int16_t block/* align 16*/);
	347	+ void (fdct248)(int16_t block/* align 16*/);
349	348
350	349	/* IDCT really*/
351		− void (idct)(DCTELEM block/* align 16*/);
	350	+ void (idct)(int16_t block/* align 16*/);
352	351
353	352	/**
354	353	* block -> idct -> clip to unsigned 8 bit -> dest.
355	354	* (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...)
356	355	* @param line_size size in bytes of a horizontal line of dest
357	356	*/
358		− void (idct_put)(uint8_t dest/align 8/, int line_size, DCTELEM block/align 16*/);
	357	+ void (idct_put)(uint8_t dest/align 8/, int line_size, int16_t block/align 16*/);
359	358
360	359	/**
361	360	* block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
362	361	* @param line_size size in bytes of a horizontal line of dest
363	362	*/
364		− void (idct_add)(uint8_t dest/align 8/, int line_size, DCTELEM block/align 16*/);
	363	+ void (idct_add)(uint8_t dest/align 8/, int line_size, int16_t block/align 16*/);
365	364
366	365	/**
367	366	* idct input permutation.
@@ -451,7 +450,7 @@ int ff_check_alignment(void); · 83 unmodified lines
451	450	* permute block according to permuatation.
452	451	* @param last last non zero element in scantable order
453	452	*/
454		−void ff_block_permute(DCTELEM block, uint8_t permutation, const uint8_t *scantable, int last);
	453	+void ff_block_permute(int16_t block, uint8_t permutation, const uint8_t *scantable, int last);
455	454
456	455	void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
457	456

libavcodec/dsputil_template.c+7−7

@@ -114,7 +114,7 @@ static void FUNCC(draw_edges)(uint8_t *_buf, int _wrap, int width, int height, i · 113 unmodified lines
114	114	}
115	115
116	116	#define DCTELEM_FUNCS(dctcoef, suffix) \
117		−static void FUNCC(get_pixels ## suffix)(DCTELEM *restrict _block, \
	117	+static void FUNCC(get_pixels ## suffix)(int16_t *restrict _block, \
118	118	const uint8_t *_pixels, \
119	119	int line_size) \
120	120	{ \
@@ -138,7 +138,7 @@ static void FUNCC(get_pixels ## suffix)(DCTELEM *restrict _block, \ · 17 unmodified lines
138	138	} \
139	139	\
140	140	static void FUNCC(add_pixels8 ## suffix)(uint8_t *restrict _pixels, \
141		− DCTELEM *_block, \
	141	+ int16_t *_block, \
142	142	int line_size) \
143	143	{ \
144	144	int i; \
@@ -161,7 +161,7 @@ static void FUNCC(add_pixels8 ## suffix)(uint8_t *restrict _pixels, \ · 16 unmodified lines
161	161	} \
162	162	\
163	163	static void FUNCC(add_pixels4 ## suffix)(uint8_t *restrict _pixels, \
164		− DCTELEM *_block, \
	164	+ int16_t *_block, \
165	165	int line_size) \
166	166	{ \
167	167	int i; \
@@ -179,20 +179,20 @@ static void FUNCC(add_pixels4 ## suffix)(uint8_t *restrict _pixels, \ · 11 unmodified lines
179	179	} \
180	180	} \
181	181	\
182		−static void FUNCC(clear_block ## suffix)(DCTELEM *block) \
	182	+static void FUNCC(clear_block ## suffix)(int16_t *block) \
183	183	{ \
184	184	memset(block, 0, sizeof(dctcoef)*64); \
185	185	} \
186	186	\
187	187	/** \
188		− * memset(blocks, 0, sizeof(DCTELEM)664) \
	188	+ * memset(blocks, 0, sizeof(int16_t)664) \
189	189	*/ \
190		−static void FUNCC(clear_blocks ## suffix)(DCTELEM *blocks) \
	190	+static void FUNCC(clear_blocks ## suffix)(int16_t *blocks) \
191	191	{ \
192	192	memset(blocks, 0, sizeof(dctcoef)664); \
193	193	}
194	194
195		−DCTELEM_FUNCS(DCTELEM, _16)
	195	+DCTELEM_FUNCS(int16_t, _16)
196	196	#if BIT_DEPTH > 8
197	197	DCTELEM_FUNCS(dctcoef, _32)
198	198	#endif

libavcodec/dv.c+2−2

@@ -406,7 +406,7 @@ typedef struct EncBlockInfo { · 405 unmodified lines
406	406	int cur_ac;
407	407	int cno;
408	408	int dct_mode;
409		− DCTELEM mb[64];
	409	+ int16_t mb[64];
410	410	uint8_t next[64];
411	411	uint8_t sign[64];
412	412	uint8_t partial_bit_count;
@@ -495,7 +495,7 @@ static av_always_inline int dv_init_enc_block(EncBlockInfo* bi, uint8_t *data, i · 82 unmodified lines
495	495	{
496	496	const int *weight;
497	497	const uint8_t* zigzag_scan;
498		− LOCAL_ALIGNED_16(DCTELEM, blk, [64]);
	498	+ LOCAL_ALIGNED_16(int16_t, blk, [64]);
499	499	int i, area;
500	500	/* We offer two different methods for class number assignment: the
501	501	method suggested in SMPTE 314M Table 22, and an improved

libavcodec/dvdata.h+3−3

@@ -40,9 +40,9 @@ typedef struct DVVideoContext { · 39 unmodified lines
40	40
41	41	uint8_t dv_zigzag[2][64];
42	42
43		− void (get_pixels)(DCTELEM block, const uint8_t *pixels, int line_size);
44		− void (fdct[2])(DCTELEM block);
45		− void (idct_put[2])(uint8_t dest, int line_size, DCTELEM *block);
	43	+ void (get_pixels)(int16_t block, const uint8_t *pixels, int line_size);
	44	+ void (fdct[2])(int16_t block);
	45	+ void (idct_put[2])(uint8_t dest, int line_size, int16_t *block);
46	46	me_cmp_func ildct_cmp;
47	47	} DVVideoContext;
48	48

libavcodec/dvdec.c+4−4

@@ -48,7 +48,7 @@ typedef struct BlockInfo { · 47 unmodified lines
48	48	const uint32_t *factor_table;
49	49	const uint8_t *scan_table;
50	50	uint8_t pos; /* position in block */
51		− void (idct_put)(uint8_t dest, int line_size, DCTELEM *block);
	51	+ void (idct_put)(uint8_t dest, int line_size, int16_t *block);
52	52	uint8_t partial_bit_count;
53	53	uint32_t partial_bit_buffer;
54	54	int shift_offset;
@@ -57,7 +57,7 @@ typedef struct BlockInfo { · 2 unmodified lines
57	57	static const int dv_iweight_bits = 14;
58	58
59	59	/* decode AC coefficients */
60		−static void dv_decode_ac(GetBitContext gb, BlockInfo mb, DCTELEM *block)
	60	+static void dv_decode_ac(GetBitContext gb, BlockInfo mb, int16_t *block)
61	61	{
62	62	int last_index = gb->size_in_bits;
63	63	const uint8_t *scan_table = mb->scan_table;
@@ -135,14 +135,14 @@ static int dv_decode_video_segment(AVCodecContext avctx, void arg) · 71 unmodified lines
135	135	int quant, dc, dct_mode, class1, j;
136	136	int mb_index, mb_x, mb_y, last_index;
137	137	int y_stride, linesize;
138		− DCTELEM block, block1;
	138	+ int16_t block, block1;
139	139	int c_offset;
140	140	uint8_t *y_ptr;
141	141	const uint8_t *buf_ptr;
142	142	PutBitContext pb, vs_pb;
143	143	GetBitContext gb;
144	144	BlockInfo mb_data[5 * DV_MAX_BPM], mb, mb1;
145		− LOCAL_ALIGNED_16(DCTELEM, sblock, [5*DV_MAX_BPM], [64]);
	145	+ LOCAL_ALIGNED_16(int16_t, sblock, [5*DV_MAX_BPM], [64]);
146	146	LOCAL_ALIGNED_16(uint8_t, mb_bit_buffer, [ 80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
147	147	LOCAL_ALIGNED_16(uint8_t, vs_bit_buffer, [580 + FF_INPUT_BUFFER_PADDING_SIZE]); / allow some slack */
148	148	const int log2_blocksize = 3;

libavcodec/eaidct.c+3−3

@@ -64,7 +64,7 @@ · 63 unmodified lines
64	64	#define MUNGE_8BIT(x) av_clip_uint8((x)>>4)
65	65	#define IDCT_ROW(dest,src) IDCT_TRANSFORM(dest,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,MUNGE_8BIT,src)
66	66
67		−static inline void ea_idct_col(DCTELEM dest, const DCTELEM src) {
	67	+static inline void ea_idct_col(int16_t dest, const int16_t src) {
68	68	if ((src[8]\|src[16]\|src[24]\|src[32]\|src[40]\|src[48]\|src[56])==0) {
69	69	dest[0] =
70	70	dest[8] =
@@ -78,9 +78,9 @@ static inline void ea_idct_col(DCTELEM dest, const DCTELEM src) { · 7 unmodified lines
78	78	IDCT_COL(dest, src);
79	79	}
80	80
81		−void ff_ea_idct_put_c(uint8_t dest, int linesize, DCTELEM block) {
	81	+void ff_ea_idct_put_c(uint8_t dest, int linesize, int16_t block) {
82	82	int i;
83		− DCTELEM temp[64];
	83	+ int16_t temp[64];
84	84	block[0] += 4;
85	85	for (i=0; i<8; i++)
86	86	ea_idct_col(&temp[i], &block[i]);

libavcodec/eaidct.h+1−2

@@ -20,8 +20,7 @@ · 19 unmodified lines
20	20	#define AVCODEC_EAIDCT_H
21	21
22	22	#include <stdint.h>
23		−#include "dsputil.h"
24	23
25		−void ff_ea_idct_put_c(uint8_t dest, int linesize, DCTELEM block);
	24	+void ff_ea_idct_put_c(uint8_t dest, int linesize, int16_t block);
26	25
27	26	#endif /* AVCODEC_EAIDCT_H */

libavcodec/eamad.c+3−3

@@ -51,7 +51,7 @@ typedef struct MadContext { · 50 unmodified lines
51	51	GetBitContext gb;
52	52	void *bitstream_buf;
53	53	unsigned int bitstream_buf_size;
54		− DECLARE_ALIGNED(16, DCTELEM, block)[64];
	54	+ DECLARE_ALIGNED(16, int16_t, block)[64];
55	55	ScanTable scantable;
56	56	uint16_t quant_matrix[64];
57	57	int mb_x;
@@ -96,7 +96,7 @@ static inline void comp_block(MadContext *t, int mb_x, int mb_y, · 38 unmodified lines
96	96	}
97	97	}
98	98
99		−static inline void idct_put(MadContext t, DCTELEM block, int mb_x, int mb_y, int j)
	99	+static inline void idct_put(MadContext t, int16_t block, int mb_x, int mb_y, int j)
100	100	{
101	101	if (j < 4) {
102	102	ff_ea_idct_put_c(
@@ -110,7 +110,7 @@ static inline void idct_put(MadContext t, DCTELEM block, int mb_x, int mb_y, i · 7 unmodified lines
110	110	}
111	111	}
112	112
113		−static inline void decode_block_intra(MadContext s, DCTELEM block)
	113	+static inline void decode_block_intra(MadContext s, int16_t block)
114	114	{
115	115	int level, i, j, run;
116	116	RLTable *rl = &ff_rl_mpeg1;

libavcodec/eatgq.c+3−3

@@ -43,7 +43,7 @@ typedef struct TgqContext { · 42 unmodified lines
43	43	int width,height;
44	44	ScanTable scantable;
45	45	int qtable[64];
46		− DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
	46	+ DECLARE_ALIGNED(16, int16_t, block)[6][64];
47	47	GetByteContext gb;
48	48	} TgqContext;
49	49
@@ -58,7 +58,7 @@ static av_cold int tgq_decode_init(AVCodecContext *avctx){ · 8 unmodified lines
58	58	return 0;
59	59	}
60	60
61		−static void tgq_decode_block(TgqContext s, DCTELEM block[64], GetBitContext gb){
	61	+static void tgq_decode_block(TgqContext s, int16_t block[64], GetBitContext gb){
62	62	uint8_t *perm = s->scantable.permutated;
63	63	int i,j,value;
64	64	block[0] = get_sbits(gb,8) * s->qtable[0];
@@ -103,7 +103,7 @@ static void tgq_decode_block(TgqContext s, DCTELEM block[64], GetBitContext gb · 38 unmodified lines
103	103	block[0] += 128<<4;
104	104	}
105	105
106		−static void tgq_idct_put_mb(TgqContext s, DCTELEM (block)[64], int mb_x, int mb_y){
	106	+static void tgq_idct_put_mb(TgqContext s, int16_t (block)[64], int mb_x, int mb_y){
107	107	int linesize= s->frame.linesize[0];
108	108	uint8_t dest_y = s->frame.data[0] + (mb_y 16* linesize ) + mb_x * 16;
109	109	uint8_t dest_cb = s->frame.data[1] + (mb_y 8 * s->frame.linesize[1]) + mb_x * 8;

libavcodec/eatqi.c+3−3

@@ -40,7 +40,7 @@ typedef struct TqiContext { · 39 unmodified lines
40	40	AVFrame frame;
41	41	void *bitstream_buf;
42	42	unsigned int bitstream_buf_size;
43		− DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
	43	+ DECLARE_ALIGNED(16, int16_t, block)[6][64];
44	44	} TqiContext;
45	45
46	46	static av_cold int tqi_decode_init(AVCodecContext *avctx)
@@ -58,7 +58,7 @@ static av_cold int tqi_decode_init(AVCodecContext *avctx) · 11 unmodified lines
58	58	return 0;
59	59	}
60	60
61		−static int tqi_decode_mb(MpegEncContext s, DCTELEM (block)[64])
	61	+static int tqi_decode_mb(MpegEncContext s, int16_t (block)[64])
62	62	{
63	63	int n;
64	64	s->dsp.clear_blocks(block[0]);
@@ -69,7 +69,7 @@ static int tqi_decode_mb(MpegEncContext s, DCTELEM (block)[64]) · 4 unmodified lines
69	69	return 0;
70	70	}
71	71
72		−static inline void tqi_idct_put(TqiContext t, DCTELEM (block)[64])
	72	+static inline void tqi_idct_put(TqiContext t, int16_t (block)[64])
73	73	{
74	74	MpegEncContext *s = &t->s;
75	75	int linesize= t->frame.linesize[0];

libavcodec/faandct.c+3−3

@@ -64,7 +64,7 @@ B6B0, B6B1, B6B2, B6B3, B6B4, B6B5, B6B6, B6B7, · 63 unmodified lines
64	64	B7B0, B7B1, B7B2, B7B3, B7B4, B7B5, B7B6, B7B7,
65	65	};
66	66
67		−static av_always_inline void row_fdct(FLOAT temp[64], DCTELEM * data)
	67	+static av_always_inline void row_fdct(FLOAT temp[64], int16_t *data)
68	68	{
69	69	FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
70	70	FLOAT tmp10, tmp11, tmp12, tmp13;
@@ -119,7 +119,7 @@ static av_always_inline void row_fdct(FLOAT temp[64], DCTELEM * data) · 48 unmodified lines
119	119	}
120	120	}
121	121
122		−void ff_faandct(DCTELEM * data)
	122	+void ff_faandct(int16_t *data)
123	123	{
124	124	FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
125	125	FLOAT tmp10, tmp11, tmp12, tmp13;
@@ -179,7 +179,7 @@ void ff_faandct(DCTELEM * data) · 53 unmodified lines
179	179	}
180	180	}
181	181
182		−void ff_faandct248(DCTELEM * data)
	182	+void ff_faandct248(int16_t *data)
183	183	{
184	184	FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
185	185	FLOAT tmp10, tmp11, tmp12, tmp13;

libavcodec/faandct.h+3−3

@@ -29,9 +29,9 @@ · 28 unmodified lines
29	29	#ifndef AVCODEC_FAANDCT_H
30	30	#define AVCODEC_FAANDCT_H
31	31
32		−#include "dsputil.h"
	32	+#include <stdint.h>
33	33
34		−void ff_faandct(DCTELEM * data);
35		−void ff_faandct248(DCTELEM * data);
	34	+void ff_faandct(int16_t *data);
	35	+void ff_faandct248(int16_t *data);
36	36
37	37	#endif /* AVCODEC_FAANDCT_H */

libavcodec/faanidct.c+4−4

@@ -47,7 +47,7 @@ B6B0/8, B6B1/8, B6B2/8, B6B3/8, B6B4/8, B6B5/8, B6B6/8, B6B7/8, · 46 unmodified lines
47	47	B7B0/8, B7B1/8, B7B2/8, B7B3/8, B7B4/8, B7B5/8, B7B6/8, B7B7/8,
48	48	};
49	49
50		−static inline void p8idct(DCTELEM data[64], FLOAT temp[64], uint8_t *dest, int stride, int x, int y, int type){
	50	+static inline void p8idct(int16_t data[64], FLOAT temp[64], uint8_t *dest, int stride, int x, int y, int type){
51	51	int i;
52	52	FLOAT av_unused tmp0;
53	53	FLOAT s04, d04, s17, d17, s26, d26, s53, d53;
@@ -129,7 +129,7 @@ static inline void p8idct(DCTELEM data[64], FLOAT temp[64], uint8_t *dest, int s · 75 unmodified lines
129	129	}
130	130	}
131	131
132		−void ff_faanidct(DCTELEM block[64]){
	132	+void ff_faanidct(int16_t block[64]){
133	133	FLOAT temp[64];
134	134	int i;
135	135
@@ -142,7 +142,7 @@ void ff_faanidct(DCTELEM block[64]){ · 6 unmodified lines
142	142	p8idct(block, temp, NULL, 0, 8, 1, 1);
143	143	}
144	144
145		−void ff_faanidct_add(uint8_t *dest, int line_size, DCTELEM block[64]){
	145	+void ff_faanidct_add(uint8_t *dest, int line_size, int16_t block[64]){
146	146	FLOAT temp[64];
147	147	int i;
148	148
@@ -155,7 +155,7 @@ void ff_faanidct_add(uint8_t *dest, int line_size, DCTELEM block[64]){ · 6 unmodified lines
155	155	p8idct(NULL , temp, dest, line_size, 8, 1, 2);
156	156	}
157	157
158		−void ff_faanidct_put(uint8_t *dest, int line_size, DCTELEM block[64]){
	158	+void ff_faanidct_put(uint8_t *dest, int line_size, int16_t block[64]){
159	159	FLOAT temp[64];
160	160	int i;
161	161

libavcodec/faanidct.h+3−4

@@ -23,10 +23,9 @@ · 22 unmodified lines
23	23	#define AVCODEC_FAANIDCT_H
24	24
25	25	#include <stdint.h>
26		−#include "dsputil.h"
27	26
28		−void ff_faanidct(DCTELEM block[64]);
29		−void ff_faanidct_add(uint8_t *dest, int line_size, DCTELEM block[64]);
30		−void ff_faanidct_put(uint8_t *dest, int line_size, DCTELEM block[64]);
	27	+void ff_faanidct(int16_t block[64]);
	28	+void ff_faanidct_add(uint8_t *dest, int line_size, int16_t block[64]);
	29	+void ff_faanidct_put(uint8_t *dest, int line_size, int16_t block[64]);
31	30
32	31	#endif /* AVCODEC_FAANIDCT_H */

libavcodec/h261dec.c+2−2

@@ -47,7 +47,7 @@ static VLC h261_mtype_vlc; · 46 unmodified lines
47	47	static VLC h261_mv_vlc;
48	48	static VLC h261_cbp_vlc;
49	49
50		−static int h261_decode_block(H261Context * h, DCTELEM * block, int n, int coded);
	50	+static int h261_decode_block(H261Context * h, int16_t * block, int n, int coded);
51	51
52	52	static av_cold void h261_decode_init_vlc(H261Context *h){
53	53	static int done = 0;
@@ -358,7 +358,7 @@ intra: · 304 unmodified lines
358	358	* Decode a macroblock.
359	359	* @return <0 if an error occurred
360	360	*/
361		−static int h261_decode_block(H261Context * h, DCTELEM * block,
	361	+static int h261_decode_block(H261Context * h, int16_t * block,
362	362	int n, int coded)
363	363	{
364	364	MpegEncContext * const s = &h->s;

libavcodec/h261enc.c+4−4

@@ -34,7 +34,7 @@ · 33 unmodified lines
34	34
35	35	extern uint8_t ff_h261_rl_table_store[2][2*MAX_RUN + MAX_LEVEL + 3];
36	36
37		−static void h261_encode_block(H261Context * h, DCTELEM * block,
	37	+static void h261_encode_block(H261Context * h, int16_t * block,
38	38	int n);
39	39
40	40	int ff_h261_get_picture_format(int width, int height){
@@ -143,7 +143,7 @@ static void h261_encode_motion(H261Context * h, int val){ · 102 unmodified lines
143	143	}
144	144
145	145	static inline int get_cbp(MpegEncContext * s,
146		− DCTELEM block[6][64])
	146	+ int16_t block[6][64])
147	147	{
148	148	int i, cbp;
149	149	cbp= 0;
@@ -154,7 +154,7 @@ static inline int get_cbp(MpegEncContext * s, · 4 unmodified lines
154	154	return cbp;
155	155	}
156	156	void ff_h261_encode_mb(MpegEncContext * s,
157		− DCTELEM block[6][64],
	157	+ int16_t block[6][64],
158	158	int motion_x, int motion_y)
159	159	{
160	160	H261Context * h = (H261Context *)s;
@@ -255,7 +255,7 @@ void ff_h261_encode_init(MpegEncContext *s){ · 94 unmodified lines
255	255	* @param block the 8x8 block
256	256	* @param n block index (0-3 are luma, 4-5 are chroma)
257	257	*/
258		−static void h261_encode_block(H261Context * h, DCTELEM * block, int n){
	258	+static void h261_encode_block(H261Context * h, int16_t * block, int n){
259	259	MpegEncContext * const s = &h->s;
260	260	int level, run, i, j, last_index, last_non_zero, sign, slevel, code;
261	261	RLTable *rl;

libavcodec/h263.c+1−1

@@ -226,7 +226,7 @@ void ff_h263_loop_filter(MpegEncContext * s){ · 225 unmodified lines
226	226	}
227	227	}
228	228
229		−void ff_h263_pred_acdc(MpegEncContext * s, DCTELEM *block, int n)
	229	+void ff_h263_pred_acdc(MpegEncContext * s, int16_t *block, int n)
230	230	{
231	231	int x, y, wrap, a, c, pred_dc, scale, i;
232	232	int16_t dc_val, ac_val, *ac_val1;

libavcodec/h263.h+4−4

@@ -73,7 +73,7 @@ int ff_h263_decode_frame(AVCodecContext *avctx, · 72 unmodified lines
73	73	AVPacket *avpkt);
74	74	int ff_h263_decode_end(AVCodecContext *avctx);
75	75	void ff_h263_encode_mb(MpegEncContext *s,
76		− DCTELEM block[6][64],
	76	+ int16_t block[6][64],
77	77	int motion_x, int motion_y);
78	78	void ff_h263_encode_picture_header(MpegEncContext *s, int picture_number);
79	79	void ff_h263_encode_gob_header(MpegEncContext * s, int mb_line);
@@ -89,7 +89,7 @@ int ff_h263_decode_mba(MpegEncContext *s); · 9 unmodified lines
89	89	void ff_h263_encode_mba(MpegEncContext *s);
90	90	void ff_init_qscale_tab(MpegEncContext *s);
91	91	int ff_h263_pred_dc(MpegEncContext * s, int n, int16_t **dc_val_ptr);
92		−void ff_h263_pred_acdc(MpegEncContext * s, DCTELEM *block, int n);
	92	+void ff_h263_pred_acdc(MpegEncContext * s, int16_t *block, int n);
93	93
94	94
95	95	/**
@@ -99,7 +99,7 @@ void ff_h263_show_pict_info(MpegEncContext *s); · 3 unmodified lines
99	99
100	100	int ff_intel_h263_decode_picture_header(MpegEncContext *s);
101	101	int ff_h263_decode_mb(MpegEncContext *s,
102		− DCTELEM block[6][64]);
	102	+ int16_t block[6][64]);
103	103
104	104	/**
105	105	* Return the value of the 3bit "source format" syntax element.
@@ -144,7 +144,7 @@ static inline void ff_h263_encode_motion_vector(MpegEncContext * s, int x, int y · 38 unmodified lines
144	144	}
145	145
146	146	static inline int get_p_cbp(MpegEncContext * s,
147		− DCTELEM block[6][64],
	147	+ int16_t block[6][64],
148	148	int motion_x, int motion_y){
149	149	int cbp, i;
150	150

libavcodec/h264.c+5−5

@@ -1750,7 +1750,7 @@ static av_always_inline void xchg_mb_border(H264Context h, uint8_t src_y, · 1749 unmodified lines
1750	1750	}
1751	1751	}
1752	1752
1753		−static av_always_inline int dctcoef_get(DCTELEM *mb, int high_bit_depth,
	1753	+static av_always_inline int dctcoef_get(int16_t *mb, int high_bit_depth,
1754	1754	int index)
1755	1755	{
1756	1756	if (high_bit_depth) {
@@ -1759,7 +1759,7 @@ static av_always_inline int dctcoef_get(DCTELEM *mb, int high_bit_depth, · 2 unmodified lines
1759	1759	return AV_RN16A(mb + index);
1760	1760	}
1761	1761
1762		−static av_always_inline void dctcoef_set(DCTELEM *mb, int high_bit_depth,
	1762	+static av_always_inline void dctcoef_set(int16_t *mb, int high_bit_depth,
1763	1763	int index, int value)
1764	1764	{
1765	1765	if (high_bit_depth) {
@@ -1778,8 +1778,8 @@ static av_always_inline void hl_decode_mb_predict_luma(H264Context *h, · 12 unmodified lines
1778	1778	uint8_t *dest_y, int p)
1779	1779	{
1780	1780	MpegEncContext *const s = &h->s;
1781		− void (idct_add)(uint8_t dst, DCTELEM *block, int stride);
1782		− void (idct_dc_add)(uint8_t dst, DCTELEM *block, int stride);
	1781	+ void (idct_add)(uint8_t dst, int16_t *block, int stride);
	1782	+ void (idct_dc_add)(uint8_t dst, int16_t *block, int stride);
1783	1783	int i;
1784	1784	int qscale = p == 0 ? s->qscale : h->chroma_qp[p - 1];
1785	1785	block_offset += 16 * p;
@@ -1895,7 +1895,7 @@ static av_always_inline void hl_decode_mb_idct_luma(H264Context *h, int mb_type, · 109 unmodified lines
1895	1895	uint8_t *dest_y, int p)
1896	1896	{
1897	1897	MpegEncContext *const s = &h->s;
1898		− void (idct_add)(uint8_t dst, DCTELEM *block, int stride);
	1898	+ void (idct_add)(uint8_t dst, int16_t *block, int stride);
1899	1899	int i;
1900	1900	block_offset += 16 * p;
1901	1901	if (!IS_INTRA4x4(mb_type)) {

libavcodec/h264.h+3−4

@@ -29,7 +29,6 @@ · 28 unmodified lines
29	29	#define AVCODEC_H264_H
30	30
31	31	#include "libavutil/intreadwrite.h"
32		−#include "dsputil.h"
33	32	#include "cabac.h"
34	33	#include "mpegvideo.h"
35	34	#include "h264dsp.h"
@@ -386,9 +385,9 @@ typedef struct H264Context { · 350 unmodified lines
386	385	GetBitContext *intra_gb_ptr;
387	386	GetBitContext *inter_gb_ptr;
388	387
389		− DECLARE_ALIGNED(16, DCTELEM, mb)[16 * 48 * 2]; ///< as a dct coeffecient is int32_t in high depth, we need to reserve twice the space.
390		− DECLARE_ALIGNED(16, DCTELEM, mb_luma_dc)[3][16 * 2];
391		− DCTELEM mb_padding[256 * 2]; ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not too large or ensure that there is some unused stuff after mb
	388	+ DECLARE_ALIGNED(16, int16_t, mb)[16 * 48 * 2]; ///< as a dct coeffecient is int32_t in high depth, we need to reserve twice the space.
	389	+ DECLARE_ALIGNED(16, int16_t, mb_luma_dc)[3][16 * 2];
	390	+ int16_t mb_padding[256 * 2]; ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not too large or ensure that there is some unused stuff after mb
392	391
393	392	/**
394	393	* Cabac

libavcodec/h264_cabac.c+28−8

@@ -1561,7 +1561,7 @@ static av_always_inline int get_cabac_cbf_ctx( H264Context *h, int cat, int idx, · 1560 unmodified lines
1561	1561	}
1562	1562
1563	1563	static av_always_inline void
1564		−decode_cabac_residual_internal(H264Context h, DCTELEM block,
	1564	+decode_cabac_residual_internal(H264Context h, int16_t block,
1565	1565	int cat, int n, const uint8_t *scantable,
1566	1566	const uint32_t *qmul, int max_coeff,
1567	1567	int is_dc, int chroma422)
@@ -1745,18 +1745,27 @@ decode_cabac_residual_internal(H264Context h, DCTELEM block, · 177 unmodified lines
1745	1745
1746	1746	}
1747	1747
1748		−static void decode_cabac_residual_dc_internal( H264Context h, DCTELEM block, int cat, int n, const uint8_t *scantable, int max_coeff ) {
	1748	+static void decode_cabac_residual_dc_internal(H264Context h, int16_t block,
	1749	+ int cat, int n,
	1750	+ const uint8_t *scantable,
	1751	+ int max_coeff)
	1752	+{
1749	1753	decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1, 0);
1750	1754	}
1751	1755
1752		−static void decode_cabac_residual_dc_internal_422(H264Context h, DCTELEM block,
	1756	+static void decode_cabac_residual_dc_internal_422(H264Context h, int16_t block,
1753	1757	int cat, int n, const uint8_t *scantable,
1754	1758	int max_coeff)
1755	1759	{
1756	1760	decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1, 1);
1757	1761	}
1758	1762
1759		−static void decode_cabac_residual_nondc_internal( H264Context h, DCTELEM block, int cat, int n, const uint8_t scantable, const uint32_t qmul, int max_coeff ) {
	1763	+static void decode_cabac_residual_nondc_internal(H264Context h, int16_t block,
	1764	+ int cat, int n,
	1765	+ const uint8_t *scantable,
	1766	+ const uint32_t *qmul,
	1767	+ int max_coeff)
	1768	+{
1760	1769	decode_cabac_residual_internal(h, block, cat, n, scantable, qmul, max_coeff, 0, 0);
1761	1770	}
1762	1771
@@ -1772,7 +1781,12 @@ static void decode_cabac_residual_nondc_internal( H264Context h, DCTELEM block · 9 unmodified lines
1772	1781	* because it allows improved constant propagation into get_cabac_cbf_ctx,
1773	1782	* as well as because most blocks have zero CBFs. */
1774	1783
1775		−static av_always_inline void decode_cabac_residual_dc( H264Context h, DCTELEM block, int cat, int n, const uint8_t *scantable, int max_coeff ) {
	1784	+static av_always_inline void decode_cabac_residual_dc(H264Context *h,
	1785	+ int16_t *block,
	1786	+ int cat, int n,
	1787	+ const uint8_t *scantable,
	1788	+ int max_coeff)
	1789	+{
1776	1790	/* read coded block flag */
1777	1791	if( get_cabac( &h->cabac, &h->cabac_state[get_cabac_cbf_ctx( h, cat, n, max_coeff, 1 ) ] ) == 0 ) {
1778	1792	h->non_zero_count_cache[scan8[n]] = 0;
@@ -1782,7 +1796,7 @@ static av_always_inline void decode_cabac_residual_dc( H264Context h, DCTELEM · 3 unmodified lines
1782	1796	}
1783	1797
1784	1798	static av_always_inline void
1785		−decode_cabac_residual_dc_422(H264Context h, DCTELEM block,
	1799	+decode_cabac_residual_dc_422(H264Context h, int16_t block,
1786	1800	int cat, int n, const uint8_t *scantable,
1787	1801	int max_coeff)
1788	1802	{
@@ -1794,7 +1808,13 @@ decode_cabac_residual_dc_422(H264Context h, DCTELEM block, · 5 unmodified lines
1794	1808	decode_cabac_residual_dc_internal_422(h, block, cat, n, scantable, max_coeff);
1795	1809	}
1796	1810
1797		−static av_always_inline void decode_cabac_residual_nondc( H264Context h, DCTELEM block, int cat, int n, const uint8_t scantable, const uint32_t qmul, int max_coeff ) {
	1811	+static av_always_inline void decode_cabac_residual_nondc(H264Context *h,
	1812	+ int16_t *block,
	1813	+ int cat, int n,
	1814	+ const uint8_t *scantable,
	1815	+ const uint32_t *qmul,
	1816	+ int max_coeff)
	1817	+{
1798	1818	/* read coded block flag */
1799	1819	if( (cat != 5 \|\| CHROMA444) && get_cabac( &h->cabac, &h->cabac_state[get_cabac_cbf_ctx( h, cat, n, max_coeff, 0 ) ] ) == 0 ) {
1800	1820	if( max_coeff == 64 ) {
@@ -2354,7 +2374,7 @@ decode_intra_mb: · 553 unmodified lines
2354	2374	if( cbp&0x20 ) {
2355	2375	int c, i, i8x8;
2356	2376	for( c = 0; c < 2; c++ ) {
2357		− DCTELEM mb = h->mb + (16(16 + 16*c) << pixel_shift);
	2377	+ int16_t mb = h->mb + (16(16 + 16*c) << pixel_shift);
2358	2378	qmul = h->dequant4_coeff[c+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[c]];
2359	2379	for (i8x8 = 0; i8x8 < 2; i8x8++) {
2360	2380	for (i = 0; i < 4; i++) {

libavcodec/h264_cavlc.c+3−3

@@ -442,7 +442,7 @@ static inline int get_level_prefix(GetBitContext *gb){ · 441 unmodified lines
442	442	* @param max_coeff number of coefficients in the block
443	443	* @return <0 if an error occurred
444	444	*/
445		−static int decode_residual(H264Context h, GetBitContext gb, DCTELEM block, int n, const uint8_t scantable, const uint32_t *qmul, int max_coeff){
	445	+static int decode_residual(H264Context h, GetBitContext gb, int16_t block, int n, const uint8_t scantable, const uint32_t *qmul, int max_coeff){
446	446	MpegEncContext * const s = &h->s;
447	447	static const int coeff_token_table_index[17]= {0, 0, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3};
448	448	int level[16];
@@ -662,7 +662,7 @@ static av_always_inline int decode_luma_residual(H264Context h, GetBitContext · 213 unmodified lines
662	662	for(i8x8=0; i8x8<4; i8x8++){
663	663	if(cbp & (1<<i8x8)){
664	664	if(IS_8x8DCT(mb_type)){
665		− DCTELEM buf = &h->mb[64i8x8+256*p << pixel_shift];
	665	+ int16_t buf = &h->mb[64i8x8+256*p << pixel_shift];
666	666	uint8_t *nnz;
667	667	for(i4x4=0; i4x4<4; i4x4++){
668	668	const int index= i4x4 + 4i8x8 + p16;
@@ -1134,7 +1134,7 @@ decode_intra_mb: · 465 unmodified lines
1134	1134	if(cbp&0x20){
1135	1135	for(chroma_idx=0; chroma_idx<2; chroma_idx++){
1136	1136	const uint32_t *qmul = h->dequant4_coeff[chroma_idx+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[chroma_idx]];
1137		− DCTELEM mb = h->mb + (16(16 + 16*chroma_idx) << pixel_shift);
	1137	+ int16_t mb = h->mb + (16(16 + 16*chroma_idx) << pixel_shift);
1138	1138	for (i8x8 = 0; i8x8 < 2; i8x8++) {
1139	1139	for (i4x4 = 0; i4x4 < 4; i4x4++) {
1140	1140	const int index = 16 + 16chroma_idx + 8i8x8 + i4x4;

libavcodec/h264_mb_template.c+1−1

@@ -52,7 +52,7 @@ static av_noinline void FUNC(hl_decode_mb)(H264Context *h) · 51 unmodified lines
52	52	const int transform_bypass = !SIMPLE && (s->qscale == 0 && h->sps.transform_bypass);
53	53	/* is_h264 should always be true if SVQ3 is disabled. */
54	54	const int is_h264 = !CONFIG_SVQ3_DECODER \|\| SIMPLE \|\| s->codec_id == AV_CODEC_ID_H264;
55		− void (idct_add)(uint8_t dst, DCTELEM *block, int stride);
	55	+ void (idct_add)(uint8_t dst, int16_t *block, int stride);
56	56	const int block_h = 16 >> s->chroma_y_shift;
57	57	const int chroma422 = CHROMA422;
58	58

libavcodec/h264dsp.h+11−13

@@ -29,8 +29,6 @@ · 28 unmodified lines
29	29
30	30	#include <stdint.h>
31	31
32		−#include "dsputil.h"
33		−
34	32	typedef void (h264_weight_func)(uint8_t block, int stride, int height,
35	33	int log2_denom, int weight, int offset);
36	34	typedef void (h264_biweight_func)(uint8_t dst, uint8_t *src,
@@ -80,29 +78,29 @@ typedef struct H264DSPContext { · 43 unmodified lines
80	78
81	79	/* IDCT */
82	80	void (h264_idct_add)(uint8_t dst /align 4/,
83		− DCTELEM block /align 16*/, int stride);
	81	+ int16_t block /align 16*/, int stride);
84	82	void (h264_idct8_add)(uint8_t dst /align 8/,
85		− DCTELEM block /align 16*/, int stride);
	83	+ int16_t block /align 16*/, int stride);
86	84	void (h264_idct_dc_add)(uint8_t dst /align 4/,
87		− DCTELEM block /align 16*/, int stride);
	85	+ int16_t block /align 16*/, int stride);
88	86	void (h264_idct8_dc_add)(uint8_t dst /align 8/,
89		− DCTELEM block /align 16*/, int stride);
	87	+ int16_t block /align 16*/, int stride);
90	88
91	89	void (h264_idct_add16)(uint8_t dst /align 16/, const int *blockoffset,
92		− DCTELEM block /align 16*/, int stride,
	90	+ int16_t block /align 16*/, int stride,
93	91	const uint8_t nnzc[15 * 8]);
94	92	void (h264_idct8_add4)(uint8_t dst /align 16/, const int *blockoffset,
95		− DCTELEM block /align 16*/, int stride,
	93	+ int16_t block /align 16*/, int stride,
96	94	const uint8_t nnzc[15 * 8]);
97	95	void (h264_idct_add8)(uint8_t dst /align 16/, const int blockoffset,
98		− DCTELEM block /align 16*/, int stride,
	96	+ int16_t block /align 16*/, int stride,
99	97	const uint8_t nnzc[15 * 8]);
100	98	void (h264_idct_add16intra)(uint8_t dst /align 16/, const int *blockoffset,
101		− DCTELEM block /align 16*/,
	99	+ int16_t block /align 16*/,
102	100	int stride, const uint8_t nnzc[15 * 8]);
103		− void (h264_luma_dc_dequant_idct)(DCTELEM output,
104		− DCTELEM input /align 16*/, int qmul);
105		− void (h264_chroma_dc_dequant_idct)(DCTELEM block, int qmul);
	101	+ void (h264_luma_dc_dequant_idct)(int16_t output,
	102	+ int16_t input /align 16*/, int qmul);
	103	+ void (h264_chroma_dc_dequant_idct)(int16_t block, int qmul);
106	104	} H264DSPContext;
107	105
108	106	void ff_h264dsp_init(H264DSPContext *c, const int bit_depth,

libavcodec/h264idct_template.c+12−12

@@ -47,7 +47,7 @@ static const uint8_t scan8[16*3]={ · 46 unmodified lines
47	47	};
48	48	#endif
49	49
50		−void FUNCC(ff_h264_idct_add)(uint8_t _dst, DCTELEM _block, int stride)
	50	+void FUNCC(ff_h264_idct_add)(uint8_t _dst, int16_t _block, int stride)
51	51	{
52	52	int i;
53	53	pixel dst = (pixel)_dst;
@@ -81,7 +81,7 @@ void FUNCC(ff_h264_idct_add)(uint8_t _dst, DCTELEM _block, int stride) · 27 unmodified lines
81	81	}
82	82	}
83	83
84		−void FUNCC(ff_h264_idct8_add)(uint8_t _dst, DCTELEM _block, int stride){
	84	+void FUNCC(ff_h264_idct8_add)(uint8_t _dst, int16_t _block, int stride){
85	85	int i;
86	86	pixel dst = (pixel)_dst;
87	87	dctcoef block = (dctcoef)_block;
@@ -154,7 +154,7 @@ void FUNCC(ff_h264_idct8_add)(uint8_t _dst, DCTELEM _block, int stride){ · 66 unmodified lines
154	154	}
155	155
156	156	// assumes all AC coefs are 0
157		−void FUNCC(ff_h264_idct_dc_add)(uint8_t _dst, DCTELEM block, int stride){
	157	+void FUNCC(ff_h264_idct_dc_add)(uint8_t _dst, int16_t block, int stride){
158	158	int i, j;
159	159	int dc = (((dctcoef*)block)[0] + 32) >> 6;
160	160	pixel dst = (pixel)_dst;
@@ -167,7 +167,7 @@ void FUNCC(ff_h264_idct_dc_add)(uint8_t _dst, DCTELEM block, int stride){ · 6 unmodified lines
167	167	}
168	168	}
169	169
170		−void FUNCC(ff_h264_idct8_dc_add)(uint8_t _dst, DCTELEM block, int stride){
	170	+void FUNCC(ff_h264_idct8_dc_add)(uint8_t _dst, int16_t block, int stride){
171	171	int i, j;
172	172	int dc = (((dctcoef*)block)[0] + 32) >> 6;
173	173	pixel dst = (pixel)_dst;
@@ -180,7 +180,7 @@ void FUNCC(ff_h264_idct8_dc_add)(uint8_t _dst, DCTELEM block, int stride){ · 6 unmodified lines
180	180	}
181	181	}
182	182
183		−void FUNCC(ff_h264_idct_add16)(uint8_t dst, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	183	+void FUNCC(ff_h264_idct_add16)(uint8_t dst, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
184	184	int i;
185	185	for(i=0; i<16; i++){
186	186	int nnz = nnzc[ scan8[i] ];
@@ -191,7 +191,7 @@ void FUNCC(ff_h264_idct_add16)(uint8_t dst, const int block_offset, DCTELEM *b · 4 unmodified lines
191	191	}
192	192	}
193	193
194		−void FUNCC(ff_h264_idct_add16intra)(uint8_t dst, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	194	+void FUNCC(ff_h264_idct_add16intra)(uint8_t dst, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
195	195	int i;
196	196	for(i=0; i<16; i++){
197	197	if(nnzc[ scan8[i] ]) FUNCC(ff_h264_idct_add )(dst + block_offset[i], block + i16sizeof(pixel), stride);
@@ -199,7 +199,7 @@ void FUNCC(ff_h264_idct_add16intra)(uint8_t dst, const int block_offset, DCTEL · 1 unmodified line
199	199	}
200	200	}
201	201
202		−void FUNCC(ff_h264_idct8_add4)(uint8_t dst, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	202	+void FUNCC(ff_h264_idct8_add4)(uint8_t dst, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
203	203	int i;
204	204	for(i=0; i<16; i+=4){
205	205	int nnz = nnzc[ scan8[i] ];
@@ -210,7 +210,7 @@ void FUNCC(ff_h264_idct8_add4)(uint8_t dst, const int block_offset, DCTELEM *b · 4 unmodified lines
210	210	}
211	211	}
212	212
213		−void FUNCC(ff_h264_idct_add8)(uint8_t *dest, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	213	+void FUNCC(ff_h264_idct_add8)(uint8_t *dest, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
214	214	int i, j;
215	215	for(j=1; j<3; j++){
216	216	for(i=j16; i<j16+4; i++){
@@ -222,7 +222,7 @@ void FUNCC(ff_h264_idct_add8)(uint8_t *dest, const int block_offset, DCTELEM * · 5 unmodified lines
222	222	}
223	223	}
224	224
225		−void FUNCC(ff_h264_idct_add8_422)(uint8_t *dest, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	225	+void FUNCC(ff_h264_idct_add8_422)(uint8_t *dest, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
226	226	int i, j;
227	227
228	228	for(j=1; j<3; j++){
@@ -248,7 +248,7 @@ void FUNCC(ff_h264_idct_add8_422)(uint8_t *dest, const int block_offset, DCTEL · 19 unmodified lines
248	248	* IDCT transforms the 16 dc values and dequantizes them.
249	249	* @param qmul quantization parameter
250	250	*/
251		−void FUNCC(ff_h264_luma_dc_dequant_idct)(DCTELEM _output, DCTELEM _input, int qmul){
	251	+void FUNCC(ff_h264_luma_dc_dequant_idct)(int16_t _output, int16_t _input, int qmul){
252	252	#define stride 16
253	253	int i;
254	254	int temp[16];
@@ -283,7 +283,7 @@ void FUNCC(ff_h264_luma_dc_dequant_idct)(DCTELEM _output, DCTELEM _input, int · 28 unmodified lines
283	283	#undef stride
284	284	}
285	285
286		−void FUNCC(ff_h264_chroma422_dc_dequant_idct)(DCTELEM *_block, int qmul){
	286	+void FUNCC(ff_h264_chroma422_dc_dequant_idct)(int16_t *_block, int qmul){
287	287	const int stride= 16*2;
288	288	const int xStride= 16;
289	289	int i;
@@ -310,7 +310,7 @@ void FUNCC(ff_h264_chroma422_dc_dequant_idct)(DCTELEM *_block, int qmul){ · 20 unmodified lines
310	310	}
311	311	}
312	312
313		−void FUNCC(ff_h264_chroma_dc_dequant_idct)(DCTELEM *_block, int qmul){
	313	+void FUNCC(ff_h264_chroma_dc_dequant_idct)(int16_t *_block, int qmul){
314	314	const int stride= 16*2;
315	315	const int xStride= 16;
316	316	int a,b,c,d,e;

libavcodec/h264pred.h+6−6

@@ -28,8 +28,8 @@ · 27 unmodified lines
28	28	#ifndef AVCODEC_H264PRED_H
29	29	#define AVCODEC_H264PRED_H
30	30
31		−#include "libavutil/common.h"
32		−#include "dsputil.h"
	31	+#include <stddef.h>
	32	+#include <stdint.h>
33	33
34	34	/**
35	35	* Prediction types
@@ -98,15 +98,15 @@ typedef struct H264PredContext { · 62 unmodified lines
98	98	void(pred16x16[4 + 3 + 2])(uint8_t src, ptrdiff_t stride);
99	99
100	100	void(pred4x4_add[2])(uint8_t pix /align 4/,
101		− const DCTELEM block /align 16*/, ptrdiff_t stride);
	101	+ const int16_t block /align 16*/, ptrdiff_t stride);
102	102	void(pred8x8l_add[2])(uint8_t pix /align 8/,
103		− const DCTELEM block /align 16*/, ptrdiff_t stride);
	103	+ const int16_t block /align 16*/, ptrdiff_t stride);
104	104	void(pred8x8_add[3])(uint8_t pix /align 8/,
105	105	const int *block_offset,
106		− const DCTELEM block /align 16*/, ptrdiff_t stride);
	106	+ const int16_t block /align 16*/, ptrdiff_t stride);
107	107	void(pred16x16_add[3])(uint8_t pix /align 16/,
108	108	const int *block_offset,
109		− const DCTELEM block /align 16*/, ptrdiff_t stride);
	109	+ const int16_t block /align 16*/, ptrdiff_t stride);
110	110	} H264PredContext;
111	111
112	112	void ff_h264_pred_init(H264PredContext *h, int codec_id,

libavcodec/h264pred_template.c+10−10

@@ -1131,7 +1131,7 @@ static void FUNCC(pred8x8l_horizontal_up)(uint8_t *_src, int has_topleft, · 1130 unmodified lines
1131	1131	#undef PL
1132	1132	#undef SRC
1133	1133
1134		−static void FUNCC(pred4x4_vertical_add)(uint8_t _pix, const DCTELEM _block,
	1134	+static void FUNCC(pred4x4_vertical_add)(uint8_t _pix, const int16_t _block,
1135	1135	ptrdiff_t stride)
1136	1136	{
1137	1137	int i;
@@ -1150,7 +1150,7 @@ static void FUNCC(pred4x4_vertical_add)(uint8_t _pix, const DCTELEM _block, · 12 unmodified lines
1150	1150	}
1151	1151	}
1152	1152
1153		−static void FUNCC(pred4x4_horizontal_add)(uint8_t _pix, const DCTELEM _block,
	1153	+static void FUNCC(pred4x4_horizontal_add)(uint8_t _pix, const int16_t _block,
1154	1154	ptrdiff_t stride)
1155	1155	{
1156	1156	int i;
@@ -1168,7 +1168,7 @@ static void FUNCC(pred4x4_horizontal_add)(uint8_t _pix, const DCTELEM _block, · 11 unmodified lines
1168	1168	}
1169	1169	}
1170	1170
1171		−static void FUNCC(pred8x8l_vertical_add)(uint8_t _pix, const DCTELEM _block,
	1171	+static void FUNCC(pred8x8l_vertical_add)(uint8_t _pix, const int16_t _block,
1172	1172	ptrdiff_t stride)
1173	1173	{
1174	1174	int i;
@@ -1191,7 +1191,7 @@ static void FUNCC(pred8x8l_vertical_add)(uint8_t _pix, const DCTELEM _block, · 16 unmodified lines
1191	1191	}
1192	1192	}
1193	1193
1194		−static void FUNCC(pred8x8l_horizontal_add)(uint8_t _pix, const DCTELEM _block,
	1194	+static void FUNCC(pred8x8l_horizontal_add)(uint8_t _pix, const int16_t _block,
1195	1195	ptrdiff_t stride)
1196	1196	{
1197	1197	int i;
@@ -1214,7 +1214,7 @@ static void FUNCC(pred8x8l_horizontal_add)(uint8_t _pix, const DCTELEM _block, · 16 unmodified lines
1214	1214	}
1215	1215
1216	1216	static void FUNCC(pred16x16_vertical_add)(uint8_t pix, const int block_offset,
1217		− const DCTELEM *block,
	1217	+ const int16_t *block,
1218	1218	ptrdiff_t stride)
1219	1219	{
1220	1220	int i;
@@ -1224,7 +1224,7 @@ static void FUNCC(pred16x16_vertical_add)(uint8_t pix, const int block_offset, · 3 unmodified lines
1224	1224
1225	1225	static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix,
1226	1226	const int *block_offset,
1227		− const DCTELEM *block,
	1227	+ const int16_t *block,
1228	1228	ptrdiff_t stride)
1229	1229	{
1230	1230	int i;
@@ -1233,7 +1233,7 @@ static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, · 2 unmodified lines
1233	1233	}
1234	1234
1235	1235	static void FUNCC(pred8x8_vertical_add)(uint8_t pix, const int block_offset,
1236		− const DCTELEM *block, ptrdiff_t stride)
	1236	+ const int16_t *block, ptrdiff_t stride)
1237	1237	{
1238	1238	int i;
1239	1239	for(i=0; i<4; i++)
@@ -1241,7 +1241,7 @@ static void FUNCC(pred8x8_vertical_add)(uint8_t pix, const int block_offset, · 1 unmodified line
1241	1241	}
1242	1242
1243	1243	static void FUNCC(pred8x16_vertical_add)(uint8_t pix, const int block_offset,
1244		− const DCTELEM *block, ptrdiff_t stride)
	1244	+ const int16_t *block, ptrdiff_t stride)
1245	1245	{
1246	1246	int i;
1247	1247	for(i=0; i<4; i++)
@@ -1251,7 +1251,7 @@ static void FUNCC(pred8x16_vertical_add)(uint8_t pix, const int block_offset, · 3 unmodified lines
1251	1251	}
1252	1252
1253	1253	static void FUNCC(pred8x8_horizontal_add)(uint8_t pix, const int block_offset,
1254		− const DCTELEM *block,
	1254	+ const int16_t *block,
1255	1255	ptrdiff_t stride)
1256	1256	{
1257	1257	int i;
@@ -1261,7 +1261,7 @@ static void FUNCC(pred8x8_horizontal_add)(uint8_t pix, const int block_offset, · 3 unmodified lines
1261	1261
1262	1262	static void FUNCC(pred8x16_horizontal_add)(uint8_t *pix,
1263	1263	const int *block_offset,
1264		− const DCTELEM *block, ptrdiff_t stride)
	1264	+ const int16_t *block, ptrdiff_t stride)
1265	1265	{
1266	1266	int i;
1267	1267	for(i=0; i<4; i++)

libavcodec/ituh263dec.c+3−3

@@ -438,7 +438,7 @@ static void h263_decode_dquant(MpegEncContext *s){ · 437 unmodified lines
438	438	ff_set_qscale(s, s->qscale);
439	439	}
440	440
441		−static int h263_decode_block(MpegEncContext * s, DCTELEM * block,
	441	+static int h263_decode_block(MpegEncContext * s, int16_t * block,
442	442	int n, int coded)
443	443	{
444	444	int code, level, i, j, last, run;
@@ -563,7 +563,7 @@ not_coded: · 118 unmodified lines
563	563
564	564	static int h263_skip_b_part(MpegEncContext *s, int cbp)
565	565	{
566		− LOCAL_ALIGNED_16(DCTELEM, dblock, [64]);
	566	+ LOCAL_ALIGNED_16(int16_t, dblock, [64]);
567	567	int i, mbi;
568	568
569	569	/* we have to set s->mb_intra to zero to decode B-part of PB-frame correctly
@@ -599,7 +599,7 @@ static int h263_get_modb(GetBitContext gb, int pb_frame, int cbpb) · 29 unmodified lines
599	599	}
600	600
601	601	int ff_h263_decode_mb(MpegEncContext *s,
602		− DCTELEM block[6][64])
	602	+ int16_t block[6][64])
603	603	{
604	604	int cbpc, cbpy, i, cbp, pred_x, pred_y, mx, my, dquant;
605	605	int16_t *mot_val;

libavcodec/ituh263enc.c+2−2

@@ -306,7 +306,7 @@ static const int dquant_code[5]= {1,0,9,2,3}; · 305 unmodified lines
306	306	* @param block the 8x8 block
307	307	* @param n block index (0-3 are luma, 4-5 are chroma)
308	308	*/
309		−static void h263_encode_block(MpegEncContext * s, DCTELEM * block, int n)
	309	+static void h263_encode_block(MpegEncContext * s, int16_t * block, int n)
310	310	{
311	311	int level, run, last, i, j, last_index, last_non_zero, sign, slevel, code;
312	312	RLTable *rl;
@@ -455,7 +455,7 @@ static void h263p_encode_umotion(MpegEncContext * s, int val) · 142 unmodified lines
455	455	}
456	456
457	457	void ff_h263_encode_mb(MpegEncContext * s,
458		− DCTELEM block[6][64],
	458	+ int16_t block[6][64],
459	459	int motion_x, int motion_y)
460	460	{
461	461	int cbpc, cbpy, i, cbp, pred_x, pred_y;

libavcodec/jfdctfst.c+9−9

@@ -136,17 +136,17 @@ · 135 unmodified lines
136	136	#endif
137	137
138	138
139		−/* Multiply a DCTELEM variable by an int32_t constant, and immediately
140		− * descale to yield a DCTELEM result.
	139	+/* Multiply a int16_t variable by an int32_t constant, and immediately
	140	+ * descale to yield a int16_t result.
141	141	*/
142	142
143		−#define MULTIPLY(var,const) ((DCTELEM) DESCALE((var) * (const), CONST_BITS))
	143	+#define MULTIPLY(var,const) ((int16_t) DESCALE((var) * (const), CONST_BITS))
144	144
145		−static av_always_inline void row_fdct(DCTELEM * data){
	145	+static av_always_inline void row_fdct(int16_t * data){
146	146	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
147	147	int tmp10, tmp11, tmp12, tmp13;
148	148	int z1, z2, z3, z4, z5, z11, z13;
149		− DCTELEM *dataptr;
	149	+ int16_t *dataptr;
150	150	int ctr;
151	151
152	152	/* Pass 1: process rows. */
@@ -205,12 +205,12 @@ static av_always_inline void row_fdct(DCTELEM * data){ · 52 unmodified lines
205	205	*/
206	206
207	207	GLOBAL(void)
208		−ff_fdct_ifast (DCTELEM * data)
	208	+ff_fdct_ifast (int16_t * data)
209	209	{
210	210	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
211	211	int tmp10, tmp11, tmp12, tmp13;
212	212	int z1, z2, z3, z4, z5, z11, z13;
213		− DCTELEM *dataptr;
	213	+ int16_t *dataptr;
214	214	int ctr;
215	215
216	216	row_fdct(data);
@@ -271,12 +271,12 @@ ff_fdct_ifast (DCTELEM * data) · 54 unmodified lines
271	271	*/
272	272
273	273	GLOBAL(void)
274		−ff_fdct_ifast248 (DCTELEM * data)
	274	+ff_fdct_ifast248 (int16_t * data)
275	275	{
276	276	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
277	277	int tmp10, tmp11, tmp12, tmp13;
278	278	int z1;
279		− DCTELEM *dataptr;
	279	+ int16_t *dataptr;
280	280	int ctr;
281	281
282	282	row_fdct(data);

libavcodec/jfdctint_template.c+14−14

@@ -184,12 +184,12 @@ · 183 unmodified lines
184	184	#endif
185	185
186	186
187		−static av_always_inline void FUNC(row_fdct)(DCTELEM *data)
	187	+static av_always_inline void FUNC(row_fdct)(int16_t *data)
188	188	{
189	189	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
190	190	int tmp10, tmp11, tmp12, tmp13;
191	191	int z1, z2, z3, z4, z5;
192		− DCTELEM *dataptr;
	192	+ int16_t *dataptr;
193	193	int ctr;
194	194
195	195	/* Pass 1: process rows. */
@@ -216,13 +216,13 @@ static av_always_inline void FUNC(row_fdct)(DCTELEM *data) · 20 unmodified lines
216	216	tmp11 = tmp1 + tmp2;
217	217	tmp12 = tmp1 - tmp2;
218	218
219		− dataptr[0] = (DCTELEM) ((tmp10 + tmp11) << PASS1_BITS);
220		− dataptr[4] = (DCTELEM) ((tmp10 - tmp11) << PASS1_BITS);
	219	+ dataptr[0] = (int16_t) ((tmp10 + tmp11) << PASS1_BITS);
	220	+ dataptr[4] = (int16_t) ((tmp10 - tmp11) << PASS1_BITS);
221	221
222	222	z1 = MULTIPLY(tmp12 + tmp13, FIX_0_541196100);
223		− dataptr[2] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
	223	+ dataptr[2] = (int16_t) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
224	224	CONST_BITS-PASS1_BITS);
225		− dataptr[6] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
	225	+ dataptr[6] = (int16_t) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
226	226	CONST_BITS-PASS1_BITS);
227	227
228	228	/* Odd part per figure 8 --- note paper omits factor of sqrt(2).
@@ -248,10 +248,10 @@ static av_always_inline void FUNC(row_fdct)(DCTELEM *data) · 19 unmodified lines
248	248	z3 += z5;
249	249	z4 += z5;
250	250
251		− dataptr[7] = (DCTELEM) DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS);
252		− dataptr[5] = (DCTELEM) DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS);
253		− dataptr[3] = (DCTELEM) DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS);
254		− dataptr[1] = (DCTELEM) DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS);
	251	+ dataptr[7] = (int16_t) DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS);
	252	+ dataptr[5] = (int16_t) DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS);
	253	+ dataptr[3] = (int16_t) DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS);
	254	+ dataptr[1] = (int16_t) DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS);
255	255
256	256	dataptr += DCTSIZE; /* advance pointer to next row */
257	257	}
@@ -262,12 +262,12 @@ static av_always_inline void FUNC(row_fdct)(DCTELEM *data) · 4 unmodified lines
262	262	*/
263	263
264	264	GLOBAL(void)
265		−FUNC(ff_jpeg_fdct_islow)(DCTELEM *data)
	265	+FUNC(ff_jpeg_fdct_islow)(int16_t *data)
266	266	{
267	267	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
268	268	int tmp10, tmp11, tmp12, tmp13;
269	269	int z1, z2, z3, z4, z5;
270		− DCTELEM *dataptr;
	270	+ int16_t *dataptr;
271	271	int ctr;
272	272
273	273	FUNC(row_fdct)(data);
@@ -344,12 +344,12 @@ FUNC(ff_jpeg_fdct_islow)(DCTELEM *data) · 70 unmodified lines
344	344	* you do even part two times.
345	345	*/
346	346	GLOBAL(void)
347		−FUNC(ff_fdct248_islow)(DCTELEM *data)
	347	+FUNC(ff_fdct248_islow)(int16_t *data)
348	348	{
349	349	int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
350	350	int tmp10, tmp11, tmp12, tmp13;
351	351	int z1;
352		− DCTELEM *dataptr;
	352	+ int16_t *dataptr;
353	353	int ctr;
354	354
355	355	FUNC(row_fdct)(data);

libavcodec/jrevdct.c+19−19

@@ -74,7 +74,7 @@ · 73 unmodified lines
74	74
75	75	#define RIGHT_SHIFT(x, n) ((x) >> (n))
76	76
77		−typedef DCTELEM DCTBLOCK[DCTSIZE2];
	77	+typedef int16_t DCTBLOCK[DCTSIZE2];
78	78
79	79	#define CONST_BITS 13
80	80
@@ -213,7 +213,7 @@ void ff_j_rev_dct(DCTBLOCK data) · 132 unmodified lines
213	213	int32_t tmp10, tmp11, tmp12, tmp13;
214	214	int32_t z1, z2, z3, z4, z5;
215	215	int32_t d0, d1, d2, d3, d4, d5, d6, d7;
216		− register DCTELEM *dataptr;
	216	+ register int16_t *dataptr;
217	217	int rowctr;
218	218
219	219	/* Pass 1: process rows. */
@@ -249,7 +249,7 @@ void ff_j_rev_dct(DCTBLOCK data) · 29 unmodified lines
249	249	/* AC terms all zero */
250	250	if (d0) {
251	251	/* Compute a 32 bit value to assign. */
252		− DCTELEM dcval = (DCTELEM) (d0 << PASS1_BITS);
	252	+ int16_t dcval = (int16_t) (d0 << PASS1_BITS);
253	253	register int v = (dcval & 0xffff) \| ((dcval << 16) & 0xffff0000);
254	254
255	255	idataptr[0] = v;
@@ -574,14 +574,14 @@ void ff_j_rev_dct(DCTBLOCK data) · 318 unmodified lines
574	574	}
575	575	/* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
576	576
577		− dataptr[0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
578		− dataptr[7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
579		− dataptr[1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
580		− dataptr[6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
581		− dataptr[2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
582		− dataptr[5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
583		− dataptr[3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
584		− dataptr[4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
	577	+ dataptr[0] = (int16_t) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
	578	+ dataptr[7] = (int16_t) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
	579	+ dataptr[1] = (int16_t) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
	580	+ dataptr[6] = (int16_t) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
	581	+ dataptr[2] = (int16_t) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
	582	+ dataptr[5] = (int16_t) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
	583	+ dataptr[3] = (int16_t) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
	584	+ dataptr[4] = (int16_t) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
585	585
586	586	dataptr += DCTSIZE; /* advance pointer to next row */
587	587	}
@@ -920,21 +920,21 @@ void ff_j_rev_dct(DCTBLOCK data) · 332 unmodified lines
920	920
921	921	/* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
922	922
923		− dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp3,
	923	+ dataptr[DCTSIZE*0] = (int16_t) DESCALE(tmp10 + tmp3,
924	924	CONST_BITS+PASS1_BITS+3);
925		− dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp10 - tmp3,
	925	+ dataptr[DCTSIZE*7] = (int16_t) DESCALE(tmp10 - tmp3,
926	926	CONST_BITS+PASS1_BITS+3);
927		− dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp11 + tmp2,
	927	+ dataptr[DCTSIZE*1] = (int16_t) DESCALE(tmp11 + tmp2,
928	928	CONST_BITS+PASS1_BITS+3);
929		− dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(tmp11 - tmp2,
	929	+ dataptr[DCTSIZE*6] = (int16_t) DESCALE(tmp11 - tmp2,
930	930	CONST_BITS+PASS1_BITS+3);
931		− dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(tmp12 + tmp1,
	931	+ dataptr[DCTSIZE*2] = (int16_t) DESCALE(tmp12 + tmp1,
932	932	CONST_BITS+PASS1_BITS+3);
933		− dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp12 - tmp1,
	933	+ dataptr[DCTSIZE*5] = (int16_t) DESCALE(tmp12 - tmp1,
934	934	CONST_BITS+PASS1_BITS+3);
935		− dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp13 + tmp0,
	935	+ dataptr[DCTSIZE*3] = (int16_t) DESCALE(tmp13 + tmp0,
936	936	CONST_BITS+PASS1_BITS+3);
937		− dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp13 - tmp0,
	937	+ dataptr[DCTSIZE*4] = (int16_t) DESCALE(tmp13 - tmp0,
938	938	CONST_BITS+PASS1_BITS+3);
939	939
940	940	dataptr++; /* advance pointer to next column */

libavcodec/mdec.c+4−4

@@ -45,14 +45,14 @@ typedef struct MDECContext{ · 44 unmodified lines
45	45	int mb_width;
46	46	int mb_height;
47	47	int mb_x, mb_y;
48		− DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
	48	+ DECLARE_ALIGNED(16, int16_t, block)[6][64];
49	49	uint8_t *bitstream_buffer;
50	50	unsigned int bitstream_buffer_size;
51	51	int block_last_index[6];
52	52	} MDECContext;
53	53
54	54	//very similar to MPEG-1
55		−static inline int mdec_decode_block_intra(MDECContext a, DCTELEM block, int n)
	55	+static inline int mdec_decode_block_intra(MDECContext a, int16_t block, int n)
56	56	{
57	57	int level, diff, i, j, run;
58	58	int component;
@@ -119,7 +119,7 @@ static inline int mdec_decode_block_intra(MDECContext a, DCTELEM block, int n) · 60 unmodified lines
119	119	return 0;
120	120	}
121	121
122		−static inline int decode_mb(MDECContext *a, DCTELEM block[6][64]){
	122	+static inline int decode_mb(MDECContext *a, int16_t block[6][64]){
123	123	int i;
124	124	const int block_index[6]= {5,4,0,1,2,3};
125	125
@@ -134,7 +134,7 @@ static inline int decode_mb(MDECContext *a, DCTELEM block[6][64]){ · 8 unmodified lines
134	134	}
135	135
136	136	static inline void idct_put(MDECContext *a, int mb_x, int mb_y){
137		− DCTELEM (*block)[64]= a->block;
	137	+ int16_t (*block)[64]= a->block;
138	138	int linesize= a->picture.linesize[0];
139	139
140	140	uint8_t dest_y = a->picture.data[0] + (mb_y 16* linesize ) + mb_x * 16;

libavcodec/mimic.c+2−2

@@ -47,7 +47,7 @@ typedef struct { · 46 unmodified lines
47	47	AVFrame buf_ptrs [16];
48	48	AVPicture flipped_ptrs[16];
49	49
50		− DECLARE_ALIGNED(16, DCTELEM, dct_block)[64];
	50	+ DECLARE_ALIGNED(16, int16_t, dct_block)[64];
51	51
52	52	GetBitContext gb;
53	53	ScanTable scantable;
@@ -183,7 +183,7 @@ static const int8_t vlcdec_lookup[9][64] = { · 129 unmodified lines
183	183
184	184	static int vlc_decode_block(MimicContext *ctx, int num_coeffs, int qscale)
185	185	{
186		− DCTELEM *block = ctx->dct_block;
	186	+ int16_t *block = ctx->dct_block;
187	187	unsigned int pos;
188	188
189	189	ctx->dsp.clear_block(block);

libavcodec/mjpegdec.c+6−6

@@ -408,7 +408,7 @@ static inline int mjpeg_decode_dc(MJpegDecodeContext *s, int dc_index) · 407 unmodified lines
408	408	}
409	409
410	410	/* decode block and dequantize */
411		−static int decode_block(MJpegDecodeContext s, DCTELEM block, int component,
	411	+static int decode_block(MJpegDecodeContext s, int16_t block, int component,
412	412	int dc_index, int ac_index, int16_t *quant_matrix)
413	413	{
414	414	int code, i, j, level, val;
@@ -456,7 +456,7 @@ static int decode_block(MJpegDecodeContext s, DCTELEM block, int component, · 41 unmodified lines
456	456	return 0;
457	457	}
458	458
459		−static int decode_dc_progressive(MJpegDecodeContext s, DCTELEM block,
	459	+static int decode_dc_progressive(MJpegDecodeContext s, int16_t block,
460	460	int component, int dc_index,
461	461	int16_t *quant_matrix, int Al)
462	462	{
@@ -474,7 +474,7 @@ static int decode_dc_progressive(MJpegDecodeContext s, DCTELEM block, · 11 unmodified lines
474	474	}
475	475
476	476	/* decode block and dequantize - progressive JPEG version */
477		−static int decode_block_progressive(MJpegDecodeContext s, DCTELEM block,
	477	+static int decode_block_progressive(MJpegDecodeContext s, int16_t block,
478	478	uint8_t *last_nnz, int ac_index,
479	479	int16_t *quant_matrix,
480	480	int ss, int se, int Al, int *EOBRUN)
@@ -572,7 +572,7 @@ for (; ; i++) { \ · 91 unmodified lines
572	572	}
573	573
574	574	/* decode block and dequantize - progressive JPEG refinement pass */
575		−static int decode_block_refinement(MJpegDecodeContext s, DCTELEM block,
	575	+static int decode_block_refinement(MJpegDecodeContext s, int16_t block,
576	576	uint8_t *last_nnz,
577	577	int ac_index, int16_t *quant_matrix,
578	578	int ss, int se, int Al, int *EOBRUN)
@@ -880,7 +880,7 @@ static int mjpeg_decode_scan(MJpegDecodeContext *s, int nb_components, int Ah, · 301 unmodified lines
880	880	} else {
881	881	int block_idx = s->block_stride[c] * (v * mb_y + y) +
882	882	(h * mb_x + x);
883		− DCTELEM *block = s->blocks[c][block_idx];
	883	+ int16_t *block = s->blocks[c][block_idx];
884	884	if (Ah)
885	885	block[0] += get_bits1(&s->gb) *
886	886	s->quant_matrixes[s->quant_index[c]][0] << Al;
@@ -957,7 +957,7 @@ static int mjpeg_decode_scan_progressive_ac(MJpegDecodeContext *s, int ss, · 70 unmodified lines
957	957	int block_offset = mb_y * linesize * 8;
958	958	uint8_t *ptr = data + block_offset;
959	959	int block_idx = mb_y * s->block_stride[c];
960		− DCTELEM (*block)[64] = &s->blocks[c][block_idx];
	960	+ int16_t (*block)[64] = &s->blocks[c][block_idx];
961	961	uint8_t *last_nnz = &s->last_nnz[c][block_idx];
962	962	for (mb_x = 0; mb_x < s->mb_width; mb_x++, block++, last_nnz++) {
963	963	const int copy_mb = mb_bitmask && !get_bits1(&mb_bitmask_gb);

libavcodec/mjpegdec.h+2−2

@@ -88,8 +88,8 @@ typedef struct MJpegDecodeContext { · 87 unmodified lines
88	88	int got_picture; ///< we found a SOF and picture is valid, too.
89	89	int linesize[MAX_COMPONENTS]; ///< linesize << interlaced
90	90	int8_t *qscale_table;
91		− DECLARE_ALIGNED(16, DCTELEM, block)[64];
92		− DCTELEM (*blocks[MAX_COMPONENTS])[64]; ///< intermediate sums (progressive mode)
	91	+ DECLARE_ALIGNED(16, int16_t, block)[64];
	92	+ int16_t (*blocks[MAX_COMPONENTS])[64]; ///< intermediate sums (progressive mode)
93	93	uint8_t *last_nnz[MAX_COMPONENTS];
94	94	uint64_t coefs_finished[MAX_COMPONENTS]; ///< bitmask of which coefs have been completely decoded (progressive mode)
95	95	ScanTable scantable;

libavcodec/mjpegenc.c+2−2

@@ -369,7 +369,7 @@ void ff_mjpeg_encode_dc(MpegEncContext *s, int val, · 368 unmodified lines
369	369	}
370	370	}
371	371
372		−static void encode_block(MpegEncContext s, DCTELEM block, int n)
	372	+static void encode_block(MpegEncContext s, int16_t block, int n)
373	373	{
374	374	int mant, nbits, code, i, j;
375	375	int component, dc, run, last_index, val;
@@ -427,7 +427,7 @@ static void encode_block(MpegEncContext s, DCTELEM block, int n) · 51 unmodified lines
427	427	put_bits(&s->pb, huff_size_ac[0], huff_code_ac[0]);
428	428	}
429	429
430		−void ff_mjpeg_encode_mb(MpegEncContext *s, DCTELEM block[6][64])
	430	+void ff_mjpeg_encode_mb(MpegEncContext *s, int16_t block[6][64])
431	431	{
432	432	int i;
433	433	for(i=0;i<5;i++) {

libavcodec/mjpegenc.h+3−2

@@ -33,7 +33,8 @@ · 32 unmodified lines
33	33	#ifndef AVCODEC_MJPEGENC_H
34	34	#define AVCODEC_MJPEGENC_H
35	35
36		−#include "dsputil.h"
	36	+#include <stdint.h>
	37	+
37	38	#include "mpegvideo.h"
38	39
39	40	typedef struct MJpegContext {
@@ -55,6 +56,6 @@ void ff_mjpeg_encode_picture_trailer(MpegEncContext *s); · 15 unmodified lines
55	56	void ff_mjpeg_encode_stuffing(PutBitContext *pbc);
56	57	void ff_mjpeg_encode_dc(MpegEncContext *s, int val,
57	58	uint8_t huff_size, uint16_t huff_code);
58		−void ff_mjpeg_encode_mb(MpegEncContext *s, DCTELEM block[6][64]);
	59	+void ff_mjpeg_encode_mb(MpegEncContext *s, int16_t block[6][64]);
59	60
60	61	#endif /* AVCODEC_MJPEGENC_H */

libavcodec/mpeg12.c+10−10

@@ -80,7 +80,7 @@ static int mpeg_decode_motion(MpegEncContext *s, int fcode, int pred) · 79 unmodified lines
80	80	return sign_extend(val, 5 + shift);
81	81	}
82	82
83		−static inline int mpeg1_decode_block_intra(MpegEncContext s, DCTELEM block, int n)
	83	+static inline int mpeg1_decode_block_intra(MpegEncContext s, int16_t block, int n)
84	84	{
85	85	int level, dc, diff, i, j, run;
86	86	int component;
@@ -151,12 +151,12 @@ static inline int mpeg1_decode_block_intra(MpegEncContext s, DCTELEM block, in · 64 unmodified lines
151	151	return 0;
152	152	}
153	153
154		−int ff_mpeg1_decode_block_intra(MpegEncContext s, DCTELEM block, int n)
	154	+int ff_mpeg1_decode_block_intra(MpegEncContext s, int16_t block, int n)
155	155	{
156	156	return mpeg1_decode_block_intra(s, block, n);
157	157	}
158	158
159		−static inline int mpeg1_decode_block_inter(MpegEncContext s, DCTELEM block, int n)
	159	+static inline int mpeg1_decode_block_inter(MpegEncContext s, int16_t block, int n)
160	160	{
161	161	int level, i, j, run;
162	162	RLTable *rl = &ff_rl_mpeg1;
@@ -231,7 +231,7 @@ end: · 68 unmodified lines
231	231	return 0;
232	232	}
233	233
234		−static inline int mpeg1_fast_decode_block_inter(MpegEncContext s, DCTELEM block, int n)
	234	+static inline int mpeg1_fast_decode_block_inter(MpegEncContext s, int16_t block, int n)
235	235	{
236	236	int level, i, j, run;
237	237	RLTable *rl = &ff_rl_mpeg1;
@@ -303,7 +303,7 @@ end: · 65 unmodified lines
303	303	}
304	304
305	305
306		−static inline int mpeg2_decode_block_non_intra(MpegEncContext s, DCTELEM block, int n)
	306	+static inline int mpeg2_decode_block_non_intra(MpegEncContext s, int16_t block, int n)
307	307	{
308	308	int level, i, j, run;
309	309	RLTable *rl = &ff_rl_mpeg1;
@@ -383,7 +383,7 @@ end: · 73 unmodified lines
383	383	}
384	384
385	385	static inline int mpeg2_fast_decode_block_non_intra(MpegEncContext *s,
386		− DCTELEM *block, int n)
	386	+ int16_t *block, int n)
387	387	{
388	388	int level, i, j, run;
389	389	RLTable *rl = &ff_rl_mpeg1;
@@ -444,7 +444,7 @@ end: · 54 unmodified lines
444	444	}
445	445
446	446
447		−static inline int mpeg2_decode_block_intra(MpegEncContext s, DCTELEM block, int n)
	447	+static inline int mpeg2_decode_block_intra(MpegEncContext s, int16_t block, int n)
448	448	{
449	449	int level, dc, diff, i, j, run;
450	450	int component;
@@ -522,7 +522,7 @@ static inline int mpeg2_decode_block_intra(MpegEncContext s, DCTELEM block, in · 71 unmodified lines
522	522	return 0;
523	523	}
524	524
525		−static inline int mpeg2_fast_decode_block_intra(MpegEncContext s, DCTELEM block, int n)
	525	+static inline int mpeg2_fast_decode_block_intra(MpegEncContext s, int16_t block, int n)
526	526	{
527	527	int level, dc, diff, j, run;
528	528	int component;
@@ -721,7 +721,7 @@ static inline int get_qscale(MpegEncContext *s) · 192 unmodified lines
721	721
722	722	static void exchange_uv(MpegEncContext *s)
723	723	{
724		− DCTELEM (*tmp)[64];
	724	+ int16_t (*tmp)[64];
725	725
726	726	tmp = s->pblocks[4];
727	727	s->pblocks[4] = s->pblocks[5];
@@ -734,7 +734,7 @@ static void exchange_uv(MpegEncContext *s) · 6 unmodified lines
734	734	#define MT_16X8 2
735	735	#define MT_DMV 3
736	736
737		−static int mpeg_decode_mb(MpegEncContext *s, DCTELEM block[12][64])
	737	+static int mpeg_decode_mb(MpegEncContext *s, int16_t block[12][64])
738	738	{
739	739	int i, j, k, cbp, val, mb_type, motion_type;
740	740	const int mb_block_count = 4 + (1 << s->chroma_format);

libavcodec/mpeg12.h+1−1

@@ -71,6 +71,6 @@ static inline int decode_dc(GetBitContext *gb, int component) · 70 unmodified lines
71	71	return diff;
72	72	}
73	73
74		−extern int ff_mpeg1_decode_block_intra(MpegEncContext s, DCTELEM block, int n);
	74	+extern int ff_mpeg1_decode_block_intra(MpegEncContext s, int16_t block, int n);
75	75
76	76	#endif /* AVCODEC_MPEG12_H */

libavcodec/mpeg12enc.c+4−4

@@ -51,7 +51,7 @@ static const uint8_t svcd_scan_offset_placeholder[14] = { · 50 unmodified lines
51	51	};
52	52
53	53	static void mpeg1_encode_block(MpegEncContext *s,
54		− DCTELEM *block,
	54	+ int16_t *block,
55	55	int component);
56	56	static void mpeg1_encode_motion(MpegEncContext *s, int val, int f_or_b_code); // RAL: f_code parameter added
57	57
@@ -432,7 +432,7 @@ static inline void put_mb_modes(MpegEncContext *s, int n, int bits, · 374 unmodified lines
432	432	}
433	433
434	434	static av_always_inline void mpeg1_encode_mb_internal(MpegEncContext *s,
435		− DCTELEM block[6][64],
	435	+ int16_t block[6][64],
436	436	int motion_x, int motion_y,
437	437	int mb_block_count)
438	438	{
@@ -656,7 +656,7 @@ static av_always_inline void mpeg1_encode_mb_internal(MpegEncContext *s, · 217 unmodified lines
656	656	}
657	657	}
658	658
659		−void ff_mpeg1_encode_mb(MpegEncContext *s, DCTELEM block[6][64], int motion_x, int motion_y)
	659	+void ff_mpeg1_encode_mb(MpegEncContext *s, int16_t block[6][64], int motion_x, int motion_y)
660	660	{
661	661	if (s->chroma_format == CHROMA_420) mpeg1_encode_mb_internal(s, block, motion_x, motion_y, 6);
662	662	else mpeg1_encode_mb_internal(s, block, motion_x, motion_y, 8);
@@ -839,7 +839,7 @@ static inline void encode_dc(MpegEncContext *s, int diff, int component) · 176 unmodified lines
839	839	}
840	840
841	841	static void mpeg1_encode_block(MpegEncContext *s,
842		− DCTELEM *block,
	842	+ int16_t *block,
843	843	int n)
844	844	{
845	845	int alevel, level, last_non_zero, dc, diff, i, j, run, last_index, sign;

libavcodec/mpeg4video.h+2−2

@@ -83,9 +83,9 @@ extern const uint16_t ff_mpeg4_resync_prefix[8]; · 82 unmodified lines
83	83	extern const uint8_t ff_mpeg4_dc_threshold[8];
84	84
85	85	void ff_mpeg4_encode_mb(MpegEncContext *s,
86		− DCTELEM block[6][64],
	86	+ int16_t block[6][64],
87	87	int motion_x, int motion_y);
88		−void ff_mpeg4_pred_ac(MpegEncContext * s, DCTELEM *block, int n,
	88	+void ff_mpeg4_pred_ac(MpegEncContext * s, int16_t *block, int n,
89	89	int dir);
90	90	void ff_set_mpeg4_time(MpegEncContext * s);
91	91	void ff_mpeg4_encode_picture_header(MpegEncContext *s, int picture_number);

libavcodec/mpeg4videodec.c+4−4

@@ -50,7 +50,7 @@ static const int mb_type_b_map[4]= { · 49 unmodified lines
50	50	* @param n block index (0-3 are luma, 4-5 are chroma)
51	51	* @param dir the ac prediction direction
52	52	*/
53		−void ff_mpeg4_pred_ac(MpegEncContext * s, DCTELEM *block, int n,
	53	+void ff_mpeg4_pred_ac(MpegEncContext * s, int16_t *block, int n,
54	54	int dir)
55	55	{
56	56	int i;
@@ -841,7 +841,7 @@ int ff_mpeg4_decode_partitions(MpegEncContext *s) · 784 unmodified lines
841	841	* Decode a block.
842	842	* @return <0 if an error occurred
843	843	*/
844		−static inline int mpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
	844	+static inline int mpeg4_decode_block(MpegEncContext * s, int16_t * block,
845	845	int n, int coded, int intra, int rvlc)
846	846	{
847	847	int level, i, last, run;
@@ -1065,7 +1065,7 @@ static inline int mpeg4_decode_block(MpegEncContext * s, DCTELEM * block, · 217 unmodified lines
1065	1065	* decode partition C of one MB.
1066	1066	* @return <0 if an error occurred
1067	1067	*/
1068		−static int mpeg4_decode_partitioned_mb(MpegEncContext *s, DCTELEM block[6][64])
	1068	+static int mpeg4_decode_partitioned_mb(MpegEncContext *s, int16_t block[6][64])
1069	1069	{
1070	1070	int cbp, mb_type;
1071	1071	const int xy= s->mb_x + s->mb_y*s->mb_stride;
@@ -1148,7 +1148,7 @@ static int mpeg4_decode_partitioned_mb(MpegEncContext *s, DCTELEM block[6][64]) · 76 unmodified lines
1148	1148	}
1149	1149
1150	1150	static int mpeg4_decode_mb(MpegEncContext *s,
1151		− DCTELEM block[6][64])
	1151	+ int16_t block[6][64])
1152	1152	{
1153	1153	int cbpc, cbpy, i, cbp, pred_x, pred_y, mx, my, dquant;
1154	1154	int16_t *mot_val;

libavcodec/mpeg4videoenc.c+8−8

@@ -59,7 +59,7 @@ max run: 29/41 · 58 unmodified lines
59	59	* Return the number of bits that encoding the 8x8 block in block would need.
60	60	* @param[in] block_last_index last index in scantable order that refers to a non zero element in block.
61	61	*/
62		−static inline int get_block_rate(MpegEncContext * s, DCTELEM block[64], int block_last_index, uint8_t scantable[64]){
	62	+static inline int get_block_rate(MpegEncContext * s, int16_t block[64], int block_last_index, uint8_t scantable[64]){
63	63	int last=0;
64	64	int j;
65	65	int rate=0;
@@ -91,7 +91,7 @@ static inline int get_block_rate(MpegEncContext * s, DCTELEM block[64], int bloc · 25 unmodified lines
91	91	* @param[out] st scantable for each 8x8 block
92	92	* @param[in] zigzag_last_index index referring to the last non zero coefficient in zigzag order
93	93	*/
94		−static inline void restore_ac_coeffs(MpegEncContext * s, DCTELEM block[6][64], const int dir[6], uint8_t *st[6], const int zigzag_last_index[6])
	94	+static inline void restore_ac_coeffs(MpegEncContext * s, int16_t block[6][64], const int dir[6], uint8_t *st[6], const int zigzag_last_index[6])
95	95	{
96	96	int i, n;
97	97	memcpy(s->block_last_index, zigzag_last_index, sizeof(int)*6);
@@ -122,7 +122,7 @@ static inline void restore_ac_coeffs(MpegEncContext * s, DCTELEM block[6][64], c · 24 unmodified lines
122	122	* @param[out] st scantable for each 8x8 block
123	123	* @param[out] zigzag_last_index index referring to the last non zero coefficient in zigzag order
124	124	*/
125		−static inline int decide_ac_pred(MpegEncContext * s, DCTELEM block[6][64], const int dir[6], uint8_t *st[6], int zigzag_last_index[6])
	125	+static inline int decide_ac_pred(MpegEncContext * s, int16_t block[6][64], const int dir[6], uint8_t *st[6], int zigzag_last_index[6])
126	126	{
127	127	int score= 0;
128	128	int i, n;
@@ -294,7 +294,7 @@ static inline int mpeg4_get_dc_length(int level, int n){ · 165 unmodified lines
294	294	* Encode an 8x8 block.
295	295	* @param n block index (0-3 are luma, 4-5 are chroma)
296	296	*/
297		−static inline void mpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n, int intra_dc,
	297	+static inline void mpeg4_encode_block(MpegEncContext * s, int16_t * block, int n, int intra_dc,
298	298	uint8_t scan_table, PutBitContext dc_pb, PutBitContext *ac_pb)
299	299	{
300	300	int i, last_non_zero;
@@ -345,7 +345,7 @@ static inline void mpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n · 44 unmodified lines
345	345	}
346	346	}
347	347
348		−static int mpeg4_get_block_length(MpegEncContext * s, DCTELEM * block, int n, int intra_dc,
	348	+static int mpeg4_get_block_length(MpegEncContext * s, int16_t * block, int n, int intra_dc,
349	349	uint8_t *scan_table)
350	350	{
351	351	int i, last_non_zero;
@@ -396,7 +396,7 @@ static int mpeg4_get_block_length(MpegEncContext * s, DCTELEM * block, int n, in · 44 unmodified lines
396	396	return len;
397	397	}
398	398
399		−static inline void mpeg4_encode_blocks(MpegEncContext * s, DCTELEM block[6][64], int intra_dc[6],
	399	+static inline void mpeg4_encode_blocks(MpegEncContext * s, int16_t block[6][64], int intra_dc[6],
400	400	uint8_t *scan_table, PutBitContext dc_pb, PutBitContext *ac_pb){
401	401	int i;
402	402
@@ -425,7 +425,7 @@ static inline void mpeg4_encode_blocks(MpegEncContext * s, DCTELEM block[6][64], · 22 unmodified lines
425	425	}
426	426	}
427	427
428		−static inline int get_b_cbp(MpegEncContext * s, DCTELEM block[6][64],
	428	+static inline int get_b_cbp(MpegEncContext * s, int16_t block[6][64],
429	429	int motion_x, int motion_y, int mb_type)
430	430	{
431	431	int cbp = 0, i;
@@ -469,7 +469,7 @@ static inline int get_b_cbp(MpegEncContext * s, DCTELEM block[6][64], · 37 unmodified lines
469	469	static const int dquant_code[5]= {1,0,9,2,3};
470	470
471	471	void ff_mpeg4_encode_mb(MpegEncContext * s,
472		− DCTELEM block[6][64],
	472	+ int16_t block[6][64],
473	473	int motion_x, int motion_y)
474	474	{
475	475	int cbpc, cbpy, pred_x, pred_y;

libavcodec/mpegvideo.c+22−22

@@ -43,19 +43,19 @@ · 42 unmodified lines
43	43	//#include <assert.h>
44	44
45	45	static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
46		− DCTELEM *block, int n, int qscale);
	46	+ int16_t *block, int n, int qscale);
47	47	static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
48		− DCTELEM *block, int n, int qscale);
	48	+ int16_t *block, int n, int qscale);
49	49	static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
50		− DCTELEM *block, int n, int qscale);
	50	+ int16_t *block, int n, int qscale);
51	51	static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
52		− DCTELEM *block, int n, int qscale);
	52	+ int16_t *block, int n, int qscale);
53	53	static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
54		− DCTELEM *block, int n, int qscale);
	54	+ int16_t *block, int n, int qscale);
55	55	static void dct_unquantize_h263_intra_c(MpegEncContext *s,
56		− DCTELEM *block, int n, int qscale);
	56	+ int16_t *block, int n, int qscale);
57	57	static void dct_unquantize_h263_inter_c(MpegEncContext *s,
58		− DCTELEM *block, int n, int qscale);
	58	+ int16_t *block, int n, int qscale);
59	59
60	60
61	61	/* enable all paranoid tests for rounding, overflows, etc... */
@@ -394,7 +394,7 @@ int ff_alloc_picture(MpegEncContext s, Picture pic, int shared) · 332 unmodified lines
394	394	}
395	395	if (s->avctx->debug&FF_DEBUG_DCT_COEFF) {
396	396	FF_ALLOCZ_OR_GOTO(s->avctx, pic->f.dct_coeff,
397		− 64 * mb_array_size * sizeof(DCTELEM) * 6, fail)
	397	+ 64 * mb_array_size * sizeof(int16_t) * 6, fail)
398	398	}
399	399	pic->f.qstride = s->mb_stride;
400	400	FF_ALLOCZ_OR_GOTO(s->avctx, pic->f.pan_scan,
@@ -471,7 +471,7 @@ static int init_duplicate_context(MpegEncContext s, MpegEncContext base) · 70 unmodified lines
471	471	2 * 64 * sizeof(int), fail)
472	472	}
473	473	}
474		− FF_ALLOCZ_OR_GOTO(s->avctx, s->blocks, 64 * 12 * 2 * sizeof(DCTELEM), fail)
	474	+ FF_ALLOCZ_OR_GOTO(s->avctx, s->blocks, 64 * 12 * 2 * sizeof(int16_t), fail)
475	475	s->block = s->blocks[0];
476	476
477	477	for (i = 0; i < 12; i++) {
@@ -2067,7 +2067,7 @@ unhandled: · 1589 unmodified lines
2067	2067
2068	2068	/* put block[] to dest[] */
2069	2069	static inline void put_dct(MpegEncContext *s,
2070		− DCTELEM block, int i, uint8_t dest, int line_size, int qscale)
	2070	+ int16_t block, int i, uint8_t dest, int line_size, int qscale)
2071	2071	{
2072	2072	s->dct_unquantize_intra(s, block, i, qscale);
2073	2073	s->dsp.idct_put (dest, line_size, block);
@@ -2075,7 +2075,7 @@ static inline void put_dct(MpegEncContext *s, · 1 unmodified line
2075	2075
2076	2076	/* add block[] to dest[] */
2077	2077	static inline void add_dct(MpegEncContext *s,
2078		− DCTELEM block, int i, uint8_t dest, int line_size)
	2078	+ int16_t block, int i, uint8_t dest, int line_size)
2079	2079	{
2080	2080	if (s->block_last_index[i] >= 0) {
2081	2081	s->dsp.idct_add (dest, line_size, block);
@@ -2083,7 +2083,7 @@ static inline void add_dct(MpegEncContext *s, · 1 unmodified line
2083	2083	}
2084	2084
2085	2085	static inline void add_dequant_dct(MpegEncContext *s,
2086		− DCTELEM block, int i, uint8_t dest, int line_size, int qscale)
	2086	+ int16_t block, int i, uint8_t dest, int line_size, int qscale)
2087	2087	{
2088	2088	if (s->block_last_index[i] >= 0) {
2089	2089	s->dct_unquantize_inter(s, block, i, qscale);
@@ -2136,7 +2136,7 @@ void ff_clean_intra_table_entries(MpegEncContext *s) · 46 unmodified lines
2136	2136	s->interlaced_dct : true if interlaced dct used (mpeg2)
2137	2137	*/
2138	2138	static av_always_inline
2139		−void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64],
	2139	+void MPV_decode_mb_internal(MpegEncContext *s, int16_t block[12][64],
2140	2140	int is_mpeg12)
2141	2141	{
2142	2142	const int mb_xy = s->mb_y * s->mb_stride + s->mb_x;
@@ -2148,7 +2148,7 @@ void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64], · 5 unmodified lines
2148	2148	if(s->avctx->debug&FF_DEBUG_DCT_COEFF) {
2149	2149	/* save DCT coefficients */
2150	2150	int i,j;
2151		− DCTELEM dct = &s->current_picture.f.dct_coeff[mb_xy 64 * 6];
	2151	+ int16_t dct = &s->current_picture.f.dct_coeff[mb_xy 64 * 6];
2152	2152	av_log(s->avctx, AV_LOG_DEBUG, "DCT coeffs of MB at %dx%d:\n", s->mb_x, s->mb_y);
2153	2153	for(i=0; i<6; i++){
2154	2154	for(j=0; j<64; j++){
@@ -2367,7 +2367,7 @@ skip_idct: · 212 unmodified lines
2367	2367	}
2368	2368	}
2369	2369
2370		−void ff_MPV_decode_mb(MpegEncContext *s, DCTELEM block[12][64]){
	2370	+void ff_MPV_decode_mb(MpegEncContext *s, int16_t block[12][64]){
2371	2371	#if !CONFIG_SMALL
2372	2372	if(s->out_format == FMT_MPEG1) {
2373	2373	MPV_decode_mb_internal(s, block, 1);
@@ -2506,7 +2506,7 @@ void ff_mpeg_flush(AVCodecContext *avctx){ · 132 unmodified lines
2506	2506	}
2507	2507
2508	2508	static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
2509		− DCTELEM *block, int n, int qscale)
	2509	+ int16_t *block, int n, int qscale)
2510	2510	{
2511	2511	int i, level, nCoeffs;
2512	2512	const uint16_t *quant_matrix;
@@ -2538,7 +2538,7 @@ static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s, · 25 unmodified lines
2538	2538	}
2539	2539
2540	2540	static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
2541		− DCTELEM *block, int n, int qscale)
	2541	+ int16_t *block, int n, int qscale)
2542	2542	{
2543	2543	int i, level, nCoeffs;
2544	2544	const uint16_t *quant_matrix;
@@ -2567,7 +2567,7 @@ static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s, · 22 unmodified lines
2567	2567	}
2568	2568
2569	2569	static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
2570		− DCTELEM *block, int n, int qscale)
	2570	+ int16_t *block, int n, int qscale)
2571	2571	{
2572	2572	int i, level, nCoeffs;
2573	2573	const uint16_t *quant_matrix;
@@ -2597,7 +2597,7 @@ static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s, · 23 unmodified lines
2597	2597	}
2598	2598
2599	2599	static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
2600		− DCTELEM *block, int n, int qscale)
	2600	+ int16_t *block, int n, int qscale)
2601	2601	{
2602	2602	int i, level, nCoeffs;
2603	2603	const uint16_t *quant_matrix;
@@ -2630,7 +2630,7 @@ static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s, · 26 unmodified lines
2630	2630	}
2631	2631
2632	2632	static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
2633		− DCTELEM *block, int n, int qscale)
	2633	+ int16_t *block, int n, int qscale)
2634	2634	{
2635	2635	int i, level, nCoeffs;
2636	2636	const uint16_t *quant_matrix;
@@ -2661,7 +2661,7 @@ static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s, · 24 unmodified lines
2661	2661	}
2662	2662
2663	2663	static void dct_unquantize_h263_intra_c(MpegEncContext *s,
2664		− DCTELEM *block, int n, int qscale)
	2664	+ int16_t *block, int n, int qscale)
2665	2665	{
2666	2666	int i, level, qmul, qadd;
2667	2667	int nCoeffs;
@@ -2698,7 +2698,7 @@ static void dct_unquantize_h263_intra_c(MpegEncContext *s, · 30 unmodified lines
2698	2698	}
2699	2699
2700	2700	static void dct_unquantize_h263_inter_c(MpegEncContext *s,
2701		− DCTELEM *block, int n, int qscale)
	2701	+ int16_t *block, int n, int qscale)
2702	2702	{
2703	2703	int i, level, qmul, qadd;
2704	2704	int nCoeffs;

libavcodec/mpegvideo.h+25−25

@@ -669,39 +669,39 @@ typedef struct MpegEncContext { · 668 unmodified lines
669	669
670	670	uint8_t *ptr_lastgob;
671	671	int swap_uv; //vcr2 codec is an MPEG-2 variant with U and V swapped
672		− DCTELEM (*pblocks[12])[64];
	672	+ int16_t (*pblocks[12])[64];
673	673
674		− DCTELEM (*block)[64]; ///< points to one of the following blocks
675		− DCTELEM (*blocks)[8][64]; // for HQ mode we need to keep the best block
676		− int (decode_mb)(struct MpegEncContext s, DCTELEM block[6][64]); // used by some codecs to avoid a switch()
	674	+ int16_t (*block)[64]; ///< points to one of the following blocks
	675	+ int16_t (*blocks)[8][64]; // for HQ mode we need to keep the best block
	676	+ int (decode_mb)(struct MpegEncContext s, int16_t block[6][64]); // used by some codecs to avoid a switch()
677	677	#define SLICE_OK 0
678	678	#define SLICE_ERROR -1
679	679	#define SLICE_END -2 ///<end marker found
680	680	#define SLICE_NOEND -3 ///<no end marker or error found but mb count exceeded
681	681
682	682	void (dct_unquantize_mpeg1_intra)(struct MpegEncContext s,
683		− DCTELEM block/align 16*/, int n, int qscale);
	683	+ int16_t block/align 16*/, int n, int qscale);
684	684	void (dct_unquantize_mpeg1_inter)(struct MpegEncContext s,
685		− DCTELEM block/align 16*/, int n, int qscale);
	685	+ int16_t block/align 16*/, int n, int qscale);
686	686	void (dct_unquantize_mpeg2_intra)(struct MpegEncContext s,
687		− DCTELEM block/align 16*/, int n, int qscale);
	687	+ int16_t block/align 16*/, int n, int qscale);
688	688	void (dct_unquantize_mpeg2_inter)(struct MpegEncContext s,
689		− DCTELEM block/align 16*/, int n, int qscale);
	689	+ int16_t block/align 16*/, int n, int qscale);
690	690	void (dct_unquantize_h263_intra)(struct MpegEncContext s,
691		− DCTELEM block/align 16*/, int n, int qscale);
	691	+ int16_t block/align 16*/, int n, int qscale);
692	692	void (dct_unquantize_h263_inter)(struct MpegEncContext s,
693		− DCTELEM block/align 16*/, int n, int qscale);
	693	+ int16_t block/align 16*/, int n, int qscale);
694	694	void (dct_unquantize_h261_intra)(struct MpegEncContext s,
695		− DCTELEM block/align 16*/, int n, int qscale);
	695	+ int16_t block/align 16*/, int n, int qscale);
696	696	void (dct_unquantize_h261_inter)(struct MpegEncContext s,
697		− DCTELEM block/align 16*/, int n, int qscale);
	697	+ int16_t block/align 16*/, int n, int qscale);
698	698	void (dct_unquantize_intra)(struct MpegEncContext s, // unquantizer to use (mpeg4 can use both)
699		− DCTELEM block/align 16*/, int n, int qscale);
	699	+ int16_t block/align 16*/, int n, int qscale);
700	700	void (dct_unquantize_inter)(struct MpegEncContext s, // unquantizer to use (mpeg4 can use both)
701		− DCTELEM block/align 16*/, int n, int qscale);
702		− int (dct_quantize)(struct MpegEncContext s, DCTELEM block/align 16/, int n, int qscale, int overflow);
703		− int (fast_dct_quantize)(struct MpegEncContext s, DCTELEM block/align 16/, int n, int qscale, int overflow);
704		− void (denoise_dct)(struct MpegEncContext s, DCTELEM *block);
	701	+ int16_t block/align 16*/, int n, int qscale);
	702	+ int (dct_quantize)(struct MpegEncContext s, int16_t block/align 16/, int n, int qscale, int overflow);
	703	+ int (fast_dct_quantize)(struct MpegEncContext s, int16_t block/align 16/, int n, int qscale, int overflow);
	704	+ void (denoise_dct)(struct MpegEncContext s, int16_t *block);
705	705
706	706	int mpv_flags; ///< flags set by private options
707	707	int quantizer_noise_shaping;
@@ -764,7 +764,7 @@ int ff_MPV_common_init(MpegEncContext *s); · 56 unmodified lines
764	764	int ff_mpv_frame_size_alloc(MpegEncContext *s, int linesize);
765	765	int ff_MPV_common_frame_size_change(MpegEncContext *s);
766	766	void ff_MPV_common_end(MpegEncContext *s);
767		−void ff_MPV_decode_mb(MpegEncContext *s, DCTELEM block[12][64]);
	767	+void ff_MPV_decode_mb(MpegEncContext *s, int16_t block[12][64]);
768	768	int ff_MPV_frame_start(MpegEncContext s, AVCodecContext avctx);
769	769	void ff_MPV_frame_end(MpegEncContext *s);
770	770	int ff_MPV_encode_init(AVCodecContext *avctx);
@@ -784,7 +784,7 @@ void ff_print_debug_info(MpegEncContext s, AVFrame pict); · 13 unmodified lines
784	784	void ff_write_quant_matrix(PutBitContext pb, uint16_t matrix);
785	785	void ff_release_unused_pictures(MpegEncContext *s, int remove_current);
786	786	int ff_find_unused_picture(MpegEncContext *s, int shared);
787		−void ff_denoise_dct(MpegEncContext s, DCTELEM block);
	787	+void ff_denoise_dct(MpegEncContext s, int16_t block);
788	788	int ff_update_duplicate_context(MpegEncContext dst, MpegEncContext src);
789	789	int ff_MPV_lowest_referenced_row(MpegEncContext *s, int dir);
790	790	void ff_MPV_report_decode_progress(MpegEncContext *s);
@@ -799,7 +799,7 @@ void ff_er_add_slice(MpegEncContext *s, int startx, int starty, int endx, int en · 8 unmodified lines
799	799	int ff_dct_common_init(MpegEncContext *s);
800	800	void ff_convert_matrix(DSPContext dsp, int (qmat)[64], uint16_t (*qmat16)[2][64],
801	801	const uint16_t *quant_matrix, int bias, int qmin, int qmax, int intra);
802		−int ff_dct_quantize_c(MpegEncContext s, DCTELEM block, int n, int qscale, int *overflow);
	802	+int ff_dct_quantize_c(MpegEncContext s, int16_t block, int n, int qscale, int *overflow);
803	803
804	804	void ff_init_block_index(MpegEncContext *s);
805	805	void ff_copy_picture(Picture dst, Picture src);
@@ -874,7 +874,7 @@ extern const uint8_t * const ff_mpeg2_dc_scale_table[4]; · 68 unmodified lines
874	874
875	875	void ff_mpeg1_encode_picture_header(MpegEncContext *s, int picture_number);
876	876	void ff_mpeg1_encode_mb(MpegEncContext *s,
877		− DCTELEM block[6][64],
	877	+ int16_t block[6][64],
878	878	int motion_x, int motion_y);
879	879	void ff_mpeg1_encode_init(MpegEncContext *s);
880	880	void ff_mpeg1_encode_slice_header(MpegEncContext *s);
@@ -889,7 +889,7 @@ extern const uint8_t ff_h263_loop_filter_strength[32]; · 8 unmodified lines
889	889	void ff_h261_loop_filter(MpegEncContext *s);
890	890	void ff_h261_reorder_mb_index(MpegEncContext* s);
891	891	void ff_h261_encode_mb(MpegEncContext *s,
892		− DCTELEM block[6][64],
	892	+ int16_t block[6][64],
893	893	int motion_x, int motion_y);
894	894	void ff_h261_encode_picture_header(MpegEncContext * s, int picture_number);
895	895	void ff_h261_encode_init(MpegEncContext *s);
@@ -906,7 +906,7 @@ void ff_rv20_encode_picture_header(MpegEncContext *s, int picture_number); · 10 unmodified lines
906	906	void ff_msmpeg4_encode_picture_header(MpegEncContext * s, int picture_number);
907	907	void ff_msmpeg4_encode_ext_header(MpegEncContext * s);
908	908	void ff_msmpeg4_encode_mb(MpegEncContext * s,
909		− DCTELEM block[6][64],
	909	+ int16_t block[6][64],
910	910	int motion_x, int motion_y);
911	911	int ff_msmpeg4_decode_picture_header(MpegEncContext * s);
912	912	int ff_msmpeg4_decode_ext_header(MpegEncContext * s, int buf_size);
@@ -914,14 +914,14 @@ int ff_msmpeg4_decode_init(AVCodecContext *avctx); · 1 unmodified line
914	914	void ff_msmpeg4_encode_init(MpegEncContext *s);
915	915	int ff_wmv2_decode_picture_header(MpegEncContext * s);
916	916	int ff_wmv2_decode_secondary_picture_header(MpegEncContext * s);
917		−void ff_wmv2_add_mb(MpegEncContext s, DCTELEM block[6][64], uint8_t dest_y, uint8_t dest_cb, uint8_t dest_cr);
	917	+void ff_wmv2_add_mb(MpegEncContext s, int16_t block[6][64], uint8_t dest_y, uint8_t dest_cb, uint8_t dest_cr);
918	918	void ff_mspel_motion(MpegEncContext *s,
919	919	uint8_t dest_y, uint8_t dest_cb, uint8_t *dest_cr,
920	920	uint8_t *ref_picture, op_pixels_func (pix_op)[4],
921	921	int motion_x, int motion_y, int h);
922	922	int ff_wmv2_encode_picture_header(MpegEncContext * s, int picture_number);
923	923	void ff_wmv2_encode_mb(MpegEncContext * s,
924		− DCTELEM block[6][64],
	924	+ int16_t block[6][64],
925	925	int motion_x, int motion_y);
926	926
927	927	#endif /* AVCODEC_MPEGVIDEO_H */

libavcodec/mpegvideo_enc.c+17−17

@@ -51,10 +51,10 @@ · 50 unmodified lines
51	51	//#include <assert.h>
52	52
53	53	static int encode_picture(MpegEncContext *s, int picture_number);
54		−static int dct_quantize_refine(MpegEncContext s, DCTELEM block, int16_t weight, DCTELEM orig, int n, int qscale);
	54	+static int dct_quantize_refine(MpegEncContext s, int16_t block, int16_t weight, int16_t orig, int n, int qscale);
55	55	static int sse_mb(MpegEncContext *s);
56		−static void denoise_dct_c(MpegEncContext s, DCTELEM block);
57		−static int dct_quantize_trellis_c(MpegEncContext s, DCTELEM block, int n, int qscale, int *overflow);
	56	+static void denoise_dct_c(MpegEncContext s, int16_t block);
	57	+static int dct_quantize_trellis_c(MpegEncContext s, int16_t block, int n, int qscale, int *overflow);
58	58
59	59	/* enable all paranoid tests for rounding, overflows, etc... */
60	60	//#define PARANOID
@@ -1625,7 +1625,7 @@ static inline void dct_single_coeff_elimination(MpegEncContext *s, · 1564 unmodified lines
1625	1625	int score = 0;
1626	1626	int run = 0;
1627	1627	int i;
1628		− DCTELEM *block = s->block[n];
	1628	+ int16_t *block = s->block[n];
1629	1629	const int last_index = s->block_last_index[n];
1630	1630	int skip_dc;
1631	1631
@@ -1665,7 +1665,7 @@ static inline void dct_single_coeff_elimination(MpegEncContext *s, · 33 unmodified lines
1665	1665	s->block_last_index[n] = -1;
1666	1666	}
1667	1667
1668		−static inline void clip_coeffs(MpegEncContext s, DCTELEM block,
	1668	+static inline void clip_coeffs(MpegEncContext s, int16_t block,
1669	1669	int last_index)
1670	1670	{
1671	1671	int i;
@@ -1729,7 +1729,7 @@ static av_always_inline void encode_mb_internal(MpegEncContext *s, · 57 unmodified lines
1729	1729	int mb_block_count)
1730	1730	{
1731	1731	int16_t weight[8][64];
1732		− DCTELEM orig[8][64];
	1732	+ int16_t orig[8][64];
1733	1733	const int mb_x = s->mb_x;
1734	1734	const int mb_y = s->mb_y;
1735	1735	int i;
@@ -1979,7 +1979,7 @@ static av_always_inline void encode_mb_internal(MpegEncContext *s, · 243 unmodified lines
1979	1979	get_visual_weight(weight[7], ptr_cr + (dct_offset >> 1),
1980	1980	wrap_c);
1981	1981	}
1982		− memcpy(orig[0], s->block[0], sizeof(DCTELEM) * 64 * mb_block_count);
	1982	+ memcpy(orig[0], s->block[0], sizeof(int16_t) * 64 * mb_block_count);
1983	1983	}
1984	1984
1985	1985	/* DCT & quantize */
@@ -2716,7 +2716,7 @@ static int encode_thread(AVCodecContext c, void arg){ · 730 unmodified lines
2716	2716	if(best_s.mv_type==MV_TYPE_16X16){ //FIXME move 4mv after QPRD
2717	2717	const int last_qp= backup_s.qscale;
2718	2718	int qpi, qp, dc[6];
2719		− DCTELEM ac[6][16];
	2719	+ int16_t ac[6][16];
2720	2720	const int mvdir= (best_s.mv_dir&MV_DIR_BACKWARD) ? 1 : 0;
2721	2721	static const int dquant_tab[4]={-1,1,-2,2};
2722	2722
@@ -2741,7 +2741,7 @@ static int encode_thread(AVCodecContext c, void arg){ · 18 unmodified lines
2741	2741	if(s->mb_intra && s->dc_val[0]){
2742	2742	for(i=0; i<6; i++){
2743	2743	dc[i]= s->dc_val[0][ s->block_index[i] ];
2744		− memcpy(ac[i], s->ac_val[0][s->block_index[i]], sizeof(DCTELEM)*16);
	2744	+ memcpy(ac[i], s->ac_val[0][s->block_index[i]], sizeof(int16_t)*16);
2745	2745	}
2746	2746	}
2747	2747
@@ -2751,7 +2751,7 @@ static int encode_thread(AVCodecContext c, void arg){ · 3 unmodified lines
2751	2751	if(s->mb_intra && s->dc_val[0]){
2752	2752	for(i=0; i<6; i++){
2753	2753	s->dc_val[0][ s->block_index[i] ]= dc[i];
2754		− memcpy(s->ac_val[0][s->block_index[i]], ac[i], sizeof(DCTELEM)*16);
	2754	+ memcpy(s->ac_val[0][s->block_index[i]], ac[i], sizeof(int16_t)*16);
2755	2755	}
2756	2756	}
2757	2757	}
@@ -3327,7 +3327,7 @@ static int encode_picture(MpegEncContext *s, int picture_number) · 569 unmodified lines
3327	3327	return 0;
3328	3328	}
3329	3329
3330		−static void denoise_dct_c(MpegEncContext s, DCTELEM block){
	3330	+static void denoise_dct_c(MpegEncContext s, int16_t block){
3331	3331	const int intra= s->mb_intra;
3332	3332	int i;
3333	3333
@@ -3352,7 +3352,7 @@ static void denoise_dct_c(MpegEncContext s, DCTELEM block){ · 18 unmodified lines
3352	3352	}
3353	3353
3354	3354	static int dct_quantize_trellis_c(MpegEncContext *s,
3355		− DCTELEM *block, int n,
	3355	+ int16_t *block, int n,
3356	3356	int qscale, int *overflow){
3357	3357	const int *qmat;
3358	3358	const uint8_t *scantable= s->intra_scantable.scantable;
@@ -3459,7 +3459,7 @@ static int dct_quantize_trellis_c(MpegEncContext *s, · 100 unmodified lines
3459	3459	*overflow= s->max_qcoeff < max; //overflow might have happened
3460	3460
3461	3461	if(last_non_zero < start_i){
3462		− memset(block + start_i, 0, (64-start_i)*sizeof(DCTELEM));
	3462	+ memset(block + start_i, 0, (64-start_i)*sizeof(int16_t));
3463	3463	return last_non_zero;
3464	3464	}
3465	3465
@@ -3591,7 +3591,7 @@ static int dct_quantize_trellis_c(MpegEncContext *s, · 125 unmodified lines
3591	3591
3592	3592	dc= FFABS(block[0]);
3593	3593	last_non_zero= last_i - 1;
3594		− memset(block + start_i, 0, (64-start_i)*sizeof(DCTELEM));
	3594	+ memset(block + start_i, 0, (64-start_i)*sizeof(int16_t));
3595	3595
3596	3596	if(last_non_zero < start_i)
3597	3597	return last_non_zero;
@@ -3666,10 +3666,10 @@ static void build_basis(uint8_t *perm){ · 68 unmodified lines
3666	3666	}
3667	3667
3668	3668	static int dct_quantize_refine(MpegEncContext *s, //FIXME breaks denoise?
3669		− DCTELEM block, int16_t weight, DCTELEM *orig,
	3669	+ int16_t block, int16_t weight, int16_t *orig,
3670	3670	int n, int qscale){
3671	3671	int16_t rem[64];
3672		− LOCAL_ALIGNED_16(DCTELEM, d1, [64]);
	3672	+ LOCAL_ALIGNED_16(int16_t, d1, [64]);
3673	3673	const uint8_t *scantable= s->intra_scantable.scantable;
3674	3674	const uint8_t *perm_scantable= s->intra_scantable.permutated;
3675	3675	// unsigned int threshold1, threshold2;
@@ -4039,7 +4039,7 @@ STOP_TIMER("iterative search") · 363 unmodified lines
4039	4039	}
4040	4040
4041	4041	int ff_dct_quantize_c(MpegEncContext *s,
4042		− DCTELEM *block, int n,
	4042	+ int16_t *block, int n,
4043	4043	int qscale, int *overflow)
4044	4044	{
4045	4045	int i, j, level, last_non_zero, q, start_i;

libavcodec/mpegvideo_xvmc.c+1−1

@@ -44,7 +44,7 @@ void ff_xvmc_init_block(MpegEncContext *s) · 43 unmodified lines
44	44	struct xvmc_pix_fmt render = (struct xvmc_pix_fmt)s->current_picture.f.data[2];
45	45	assert(render && render->xvmc_id == AV_XVMC_ID);
46	46
47		− s->block = (DCTELEM ()[64])(render->data_blocks + render->next_free_data_block_num 64);
	47	+ s->block = (int16_t ()[64])(render->data_blocks + render->next_free_data_block_num 64);
48	48	}
49	49
50	50	/**

libavcodec/msmpeg4.c+3−3

@@ -406,7 +406,7 @@ static int msmpeg4v2_decode_motion(MpegEncContext * s, int pred, int f_code) · 405 unmodified lines
406	406	return val;
407	407	}
408	408
409		−static int msmpeg4v12_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
	409	+static int msmpeg4v12_decode_mb(MpegEncContext *s, int16_t block[6][64])
410	410	{
411	411	int cbp, code, i;
412	412
@@ -492,7 +492,7 @@ static int msmpeg4v12_decode_mb(MpegEncContext *s, DCTELEM block[6][64]) · 79 unmodified lines
492	492	return 0;
493	493	}
494	494
495		−static int msmpeg4v34_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
	495	+static int msmpeg4v34_decode_mb(MpegEncContext *s, int16_t block[6][64])
496	496	{
497	497	int cbp, code, i;
498	498	uint8_t *coded_val;
@@ -934,7 +934,7 @@ static int msmpeg4_decode_dc(MpegEncContext * s, int n, int *dir_ptr) · 435 unmodified lines
934	934	}
935	935
936	936	//#define ERROR_DETAILS
937		−int ff_msmpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
	937	+int ff_msmpeg4_decode_block(MpegEncContext * s, int16_t * block,
938	938	int n, int coded, const uint8_t *scan_table)
939	939	{
940	940	int level, i, last, run, run_diff;

libavcodec/msmpeg4.h+3−4

@@ -26,7 +26,6 @@ · 25 unmodified lines
26	26
27	27	#include "config.h"
28	28	#include "avcodec.h"
29		−#include "dsputil.h"
30	29	#include "mpegvideo.h"
31	30	#include "msmpeg4data.h"
32	31	#include "put_bits.h"
@@ -45,17 +44,17 @@ extern VLC ff_inter_intra_vlc; · 12 unmodified lines
45	44
46	45	void ff_msmpeg4_code012(PutBitContext *pb, int n);
47	46	void ff_msmpeg4_common_init(MpegEncContext *s);
48		−void ff_msmpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n);
	47	+void ff_msmpeg4_encode_block(MpegEncContext * s, int16_t * block, int n);
49	48	void ff_msmpeg4_handle_slices(MpegEncContext *s);
50	49	void ff_msmpeg4_encode_motion(MpegEncContext * s, int mx, int my);
51	50	int ff_msmpeg4_coded_block_pred(MpegEncContext * s, int n,
52	51	uint8_t **coded_block_ptr);
53	52	int ff_msmpeg4_decode_motion(MpegEncContext * s, int mx_ptr, int my_ptr);
54		−int ff_msmpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
	53	+int ff_msmpeg4_decode_block(MpegEncContext * s, int16_t * block,
55	54	int n, int coded, const uint8_t *scan_table);
56	55	int ff_msmpeg4_pred_dc(MpegEncContext *s, int n,
57	56	int16_t *dc_val_ptr, int dir_ptr);
58		−int ff_wmv2_decode_mb(MpegEncContext *s, DCTELEM block[6][64]);
	57	+int ff_wmv2_decode_mb(MpegEncContext *s, int16_t block[6][64]);
59	58
60	59	#define CONFIG_MSMPEG4_DECODER (CONFIG_MSMPEG4V1_DECODER \|\| \
61	60	CONFIG_MSMPEG4V2_DECODER \|\| \

libavcodec/msmpeg4enc.c+2−2

@@ -368,7 +368,7 @@ static void msmpeg4v2_encode_motion(MpegEncContext * s, int val) · 367 unmodified lines
368	368	}
369	369
370	370	void ff_msmpeg4_encode_mb(MpegEncContext * s,
371		− DCTELEM block[6][64],
	371	+ int16_t block[6][64],
372	372	int motion_x, int motion_y)
373	373	{
374	374	int cbp, coded_cbp, i;
@@ -569,7 +569,7 @@ static void msmpeg4_encode_dc(MpegEncContext * s, int level, int n, int *dir_ptr · 194 unmodified lines
569	569	/* Encoding of a block. Very similar to MPEG4 except for a different
570	570	escape coding (same as H263) and more vlc tables.
571	571	*/
572		−void ff_msmpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n)
	572	+void ff_msmpeg4_encode_block(MpegEncContext * s, int16_t * block, int n)
573	573	{
574	574	int level, run, last, i, j, last_index;
575	575	int last_non_zero, sign, slevel;

libavcodec/ppc/dsputil_altivec.c+3−3

@@ -476,7 +476,7 @@ static int pix_sum_altivec(uint8_t * pix, int line_size) · 475 unmodified lines
476	476	return s;
477	477	}
478	478
479		−static void get_pixels_altivec(DCTELEM restrict block, const uint8_t pixels, int line_size)
	479	+static void get_pixels_altivec(int16_t restrict block, const uint8_t pixels, int line_size)
480	480	{
481	481	int i;
482	482	vector unsigned char perm = vec_lvsl(0, pixels);
@@ -502,7 +502,7 @@ static void get_pixels_altivec(DCTELEM restrict block, const uint8_t pixels, i · 19 unmodified lines
502	502	}
503	503	}
504	504
505		−static void diff_pixels_altivec(DCTELEM restrict block, const uint8_t s1,
	505	+static void diff_pixels_altivec(int16_t restrict block, const uint8_t s1,
506	506	const uint8_t *s2, int stride)
507	507	{
508	508	int i;
@@ -576,7 +576,7 @@ static void diff_pixels_altivec(DCTELEM restrict block, const uint8_t s1, · 67 unmodified lines
576	576	}
577	577
578	578
579		−static void clear_block_altivec(DCTELEM *block) {
	579	+static void clear_block_altivec(int16_t *block) {
580	580	LOAD_ZERO;
581	581	vec_st(zero_s16v, 0, block);
582	582	vec_st(zero_s16v, 16, block);

libavcodec/ppc/dsputil_ppc.c+7−7

@@ -47,7 +47,7 @@ distinguish, and use dcbz (32 bytes) or dcbzl (one cache line) as required. · 46 unmodified lines
47	47	see <http://developer.apple.com/technotes/tn/tn2087.html>
48	48	and <http://developer.apple.com/technotes/tn/tn2086.html>
49	49	*/
50		−static void clear_blocks_dcbz32_ppc(DCTELEM *blocks)
	50	+static void clear_blocks_dcbz32_ppc(int16_t *blocks)
51	51	{
52	52	register int misal = ((unsigned long)blocks & 0x00000010);
53	53	register int i = 0;
@@ -58,7 +58,7 @@ static void clear_blocks_dcbz32_ppc(DCTELEM *blocks) · 4 unmodified lines
58	58	((unsigned long*)blocks)[3] = 0L;
59	59	i += 16;
60	60	}
61		− for ( ; i < sizeof(DCTELEM)664-31 ; i += 32) {
	61	+ for ( ; i < sizeof(int16_t)664-31 ; i += 32) {
62	62	__asm__ volatile("dcbz %0,%1" : : "b" (blocks), "r" (i) : "memory");
63	63	}
64	64	if (misal) {
@@ -73,7 +73,7 @@ static void clear_blocks_dcbz32_ppc(DCTELEM *blocks) · 8 unmodified lines
73	73	/* same as above, when dcbzl clear a whole 128B cache line
74	74	i.e. the PPC970 aka G5 */
75	75	#if HAVE_DCBZL
76		−static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
	76	+static void clear_blocks_dcbz128_ppc(int16_t *blocks)
77	77	{
78	78	register int misal = ((unsigned long)blocks & 0x0000007f);
79	79	register int i = 0;
@@ -81,17 +81,17 @@ static void clear_blocks_dcbz128_ppc(DCTELEM *blocks) · 1 unmodified line
81	81	// we could probably also optimize this case,
82	82	// but there's not much point as the machines
83	83	// aren't available yet (2003-06-26)
84		− memset(blocks, 0, sizeof(DCTELEM)664);
	84	+ memset(blocks, 0, sizeof(int16_t)664);
85	85	}
86	86	else
87		− for ( ; i < sizeof(DCTELEM)664 ; i += 128) {
	87	+ for ( ; i < sizeof(int16_t)664 ; i += 128) {
88	88	__asm__ volatile("dcbzl %0,%1" : : "b" (blocks), "r" (i) : "memory");
89	89	}
90	90	}
91	91	#else
92		−static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
	92	+static void clear_blocks_dcbz128_ppc(int16_t *blocks)
93	93	{
94		− memset(blocks, 0, sizeof(DCTELEM)664);
	94	+ memset(blocks, 0, sizeof(int16_t)664);
95	95	}
96	96	#endif
97	97

libavcodec/ppc/h264_altivec.c+9−9

@@ -315,7 +315,7 @@ H264_MC(avg_, 16, altivec) · 314 unmodified lines
315	315	va_u32 = vec_splat((vec_u32)va_u8, 0); \
316	316	vec_ste(va_u32, element, (uint32_t*)dst);
317	317
318		−static void ff_h264_idct_add_altivec(uint8_t dst, DCTELEM block, int stride)
	318	+static void ff_h264_idct_add_altivec(uint8_t dst, int16_t block, int stride)
319	319	{
320	320	vec_s16 va0, va1, va2, va3;
321	321	vec_s16 vz0, vz1, vz2, vz3;
@@ -429,7 +429,7 @@ static void ff_h264_idct_add_altivec(uint8_t dst, DCTELEM block, int stride) · 107 unmodified lines
429	429	vec_st( hv, 0, dest ); \
430	430	}
431	431
432		−static void ff_h264_idct8_add_altivec( uint8_t dst, DCTELEM dct, int stride ) {
	432	+static void ff_h264_idct8_add_altivec( uint8_t dst, int16_t dct, int stride ) {
433	433	vec_s16 s0, s1, s2, s3, s4, s5, s6, s7;
434	434	vec_s16 d0, d1, d2, d3, d4, d5, d6, d7;
435	435	vec_s16 idct0, idct1, idct2, idct3, idct4, idct5, idct6, idct7;
@@ -473,7 +473,7 @@ static void ff_h264_idct8_add_altivec( uint8_t dst, DCTELEM dct, int stride ) · 37 unmodified lines
473	473	ALTIVEC_STORE_SUM_CLIP(&dst[7*stride], idct7, perm_ldv, perm_stv, sel);
474	474	}
475	475
476		−static av_always_inline void h264_idct_dc_add_internal(uint8_t dst, DCTELEM block, int stride, int size)
	476	+static av_always_inline void h264_idct_dc_add_internal(uint8_t dst, int16_t block, int stride, int size)
477	477	{
478	478	vec_s16 dc16;
479	479	vec_u8 dcplus, dcminus, v0, v1, v2, v3, aligner;
@@ -518,17 +518,17 @@ static av_always_inline void h264_idct_dc_add_internal(uint8_t dst, DCTELEM bl · 38 unmodified lines
518	518	}
519	519	}
520	520
521		−static void h264_idct_dc_add_altivec(uint8_t dst, DCTELEM block, int stride)
	521	+static void h264_idct_dc_add_altivec(uint8_t dst, int16_t block, int stride)
522	522	{
523	523	h264_idct_dc_add_internal(dst, block, stride, 4);
524	524	}
525	525
526		−static void ff_h264_idct8_dc_add_altivec(uint8_t dst, DCTELEM block, int stride)
	526	+static void ff_h264_idct8_dc_add_altivec(uint8_t dst, int16_t block, int stride)
527	527	{
528	528	h264_idct_dc_add_internal(dst, block, stride, 8);
529	529	}
530	530
531		−static void ff_h264_idct_add16_altivec(uint8_t dst, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	531	+static void ff_h264_idct_add16_altivec(uint8_t dst, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
532	532	int i;
533	533	for(i=0; i<16; i++){
534	534	int nnz = nnzc[ scan8[i] ];
@@ -539,7 +539,7 @@ static void ff_h264_idct_add16_altivec(uint8_t dst, const int block_offset, DC · 4 unmodified lines
539	539	}
540	540	}
541	541
542		−static void ff_h264_idct_add16intra_altivec(uint8_t dst, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	542	+static void ff_h264_idct_add16intra_altivec(uint8_t dst, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
543	543	int i;
544	544	for(i=0; i<16; i++){
545	545	if(nnzc[ scan8[i] ]) ff_h264_idct_add_altivec(dst + block_offset[i], block + i*16, stride);
@@ -547,7 +547,7 @@ static void ff_h264_idct_add16intra_altivec(uint8_t dst, const int block_offse · 1 unmodified line
547	547	}
548	548	}
549	549
550		−static void ff_h264_idct8_add4_altivec(uint8_t dst, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	550	+static void ff_h264_idct8_add4_altivec(uint8_t dst, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
551	551	int i;
552	552	for(i=0; i<16; i+=4){
553	553	int nnz = nnzc[ scan8[i] ];
@@ -558,7 +558,7 @@ static void ff_h264_idct8_add4_altivec(uint8_t dst, const int block_offset, DC · 4 unmodified lines
558	558	}
559	559	}
560	560
561		−static void ff_h264_idct_add8_altivec(uint8_t *dest, const int block_offset, DCTELEM block, int stride, const uint8_t nnzc[158]){
	561	+static void ff_h264_idct_add8_altivec(uint8_t *dest, const int block_offset, int16_t block, int stride, const uint8_t nnzc[158]){
562	562	int i, j;
563	563	for (j = 1; j < 3; j++) {
564	564	for(i = j * 16; i < j * 16 + 4; i++){

libavcodec/ppc/mpegvideo_altivec.c+1−1

@@ -35,7 +35,7 @@ · 34 unmodified lines
35	35	/* AltiVec version of dct_unquantize_h263
36	36	this code assumes `block' is 16 bytes-aligned */
37	37	static void dct_unquantize_h263_altivec(MpegEncContext *s,
38		− DCTELEM *block, int n, int qscale)
	38	+ int16_t *block, int n, int qscale)
39	39	{
40	40	int i, level, qmul, qadd;
41	41	int nCoeffs;

libavcodec/ppc/vc1dsp_altivec.c+2−2

@@ -129,7 +129,7 @@ do { \ · 128 unmodified lines
129	129
130	130	/** Do inverse transform on 8x8 block
131	131	*/
132		−static void vc1_inv_trans_8x8_altivec(DCTELEM block[64])
	132	+static void vc1_inv_trans_8x8_altivec(int16_t block[64])
133	133	{
134	134	vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
135	135	vector signed int s0, s1, s2, s3, s4, s5, s6, s7;
@@ -224,7 +224,7 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64]) · 88 unmodified lines
224	224
225	225	/** Do inverse transform on 8x4 part of block
226	226	*/
227		−static void vc1_inv_trans_8x4_altivec(uint8_t dest, int stride, DCTELEM block)
	227	+static void vc1_inv_trans_8x4_altivec(uint8_t dest, int stride, int16_t block)
228	228	{
229	229	vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
230	230	vector signed int s0, s1, s2, s3, s4, s5, s6, s7;

libavcodec/ppc/vp3dsp_altivec.c+2−2

@@ -116,7 +116,7 @@ static inline vec_s16 M16(vec_s16 a, vec_s16 C) · 115 unmodified lines
116	116	#define ADD8(a) vec_add(a, eight)
117	117	#define SHIFT4(a) vec_sra(a, four)
118	118
119		−static void vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
	119	+static void vp3_idct_put_altivec(uint8_t *dst, int stride, int16_t block[64])
120	120	{
121	121	vec_u8 t;
122	122	IDCT_START
@@ -145,7 +145,7 @@ static void vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64]) · 22 unmodified lines
145	145	memset(block, 0, sizeof(block) 64);
146	146	}
147	147
148		−static void vp3_idct_add_altivec(uint8_t *dst, int stride, DCTELEM block[64])
	148	+static void vp3_idct_add_altivec(uint8_t *dst, int stride, int16_t block[64])
149	149	{
150	150	LOAD_ZERO;
151	151	vec_u8 t, vdst;

libavcodec/proresdec.c+6−5

@@ -34,6 +34,7 @@ · 33 unmodified lines
34	34
35	35	#include "libavutil/intmath.h"
36	36	#include "avcodec.h"
	37	+#include "dsputil.h"
37	38	#include "internal.h"
38	39	#include "proresdata.h"
39	40	#include "proresdsp.h"
@@ -45,7 +46,7 @@ typedef struct { · 5 unmodified lines
45	46	int x_pos, y_pos;
46	47	int slice_width;
47	48	int prev_slice_sf; ///< scalefactor of the previous decoded slice
48		− DECLARE_ALIGNED(16, DCTELEM, blocks)[8 * 4 * 64];
	49	+ DECLARE_ALIGNED(16, int16_t, blocks)[8 * 4 * 64];
49	50	DECLARE_ALIGNED(16, int16_t, qmat_luma_scaled)[64];
50	51	DECLARE_ALIGNED(16, int16_t, qmat_chroma_scaled)[64];
51	52	} ProresThreadData;
@@ -340,10 +341,10 @@ static inline int decode_vlc_codeword(GetBitContext *gb, unsigned codebook) · 288 unmodified lines
340	341	/**
341	342	* Decode DC coefficients for all blocks in a slice.
342	343	*/
343		−static inline void decode_dc_coeffs(GetBitContext gb, DCTELEM out,
	344	+static inline void decode_dc_coeffs(GetBitContext gb, int16_t out,
344	345	int nblocks)
345	346	{
346		− DCTELEM prev_dc;
	347	+ int16_t prev_dc;
347	348	int i, sign;
348	349	int16_t delta;
349	350	unsigned int code;
@@ -368,7 +369,7 @@ static inline void decode_dc_coeffs(GetBitContext gb, DCTELEM out, · 18 unmodified lines
368	369	/**
369	370	* Decode AC coefficients for all blocks in a slice.
370	371	*/
371		−static inline void decode_ac_coeffs(GetBitContext gb, DCTELEM out,
	372	+static inline void decode_ac_coeffs(GetBitContext gb, int16_t out,
372	373	int blocks_per_slice,
373	374	int plane_size_factor,
374	375	const uint8_t *scan)
@@ -421,7 +422,7 @@ static void decode_slice_plane(ProresContext ctx, ProresThreadData td, · 46 unmodified lines
421	422	const int16_t *qmat, int is_chroma)
422	423	{
423	424	GetBitContext gb;
424		− DCTELEM *block_ptr;
	425	+ int16_t *block_ptr;
425	426	int mb_num, blocks_per_slice;
426	427
427	428	blocks_per_slice = mbs_per_slice * blocks_per_mb;

libavcodec/proresdsp.c+4−3

@@ -20,6 +20,7 @@ · 19 unmodified lines
20	20	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21	21	*/
22	22
	23	+#include "dsputil.h"
23	24	#include "proresdsp.h"
24	25	#include "simple_idct.h"
25	26	#include "libavutil/common.h"
@@ -34,7 +35,7 @@ · 8 unmodified lines
34	35	/**
35	36	* Add bias value, clamp and output pixels of a slice
36	37	*/
37		−static void put_pixels(uint16_t dst, int stride, const DCTELEM in)
	38	+static void put_pixels(uint16_t dst, int stride, const int16_t in)
38	39	{
39	40	int x, y, src_offset, dst_offset;
40	41
@@ -47,7 +48,7 @@ static void put_pixels(uint16_t dst, int stride, const DCTELEM in) · 6 unmodified lines
47	48	}
48	49	}
49	50
50		−static void prores_idct_put_c(uint16_t out, int linesize, DCTELEM block, const int16_t *qmat)
	51	+static void prores_idct_put_c(uint16_t out, int linesize, int16_t block, const int16_t *qmat)
51	52	{
52	53	ff_prores_idct(block, qmat);
53	54	put_pixels(out, linesize >> 1, block);
@@ -55,7 +56,7 @@ static void prores_idct_put_c(uint16_t out, int linesize, DCTELEM block, const · 1 unmodified line
55	56	#endif
56	57
57	58	#if CONFIG_PRORES_ENCODER
58		−static void prores_fdct_c(const uint16_t src, int linesize, DCTELEM block)
	59	+static void prores_fdct_c(const uint16_t src, int linesize, int16_t block)
59	60	{
60	61	int x, y;
61	62	const uint16_t *tsrc = src;

libavcodec/proresdsp.h+3−3

@@ -23,7 +23,7 @@ · 22 unmodified lines
23	23	#ifndef AVCODEC_PRORESDSP_H
24	24	#define AVCODEC_PRORESDSP_H
25	25
26		−#include "dsputil.h"
	26	+#include <stdint.h>
27	27
28	28	#define PRORES_BITS_PER_SAMPLE 10 ///< output precision of prores decoder
29	29
@@ -32,8 +32,8 @@ typedef struct ProresDSPContext { · 2 unmodified lines
32	32	uint8_t idct_permutation[64];
33	33	int dct_permutation_type;
34	34	uint8_t dct_permutation[64];
35		− void (* idct_put) (uint16_t out, int linesize, DCTELEM block, const int16_t *qmat);
36		− void (* fdct) (const uint16_t src, int linesize, DCTELEM block);
	35	+ void (* idct_put) (uint16_t out, int linesize, int16_t block, const int16_t *qmat);
	36	+ void (* fdct) (const uint16_t src, int linesize, int16_t block);
37	37	} ProresDSPContext;
38	38
39	39	void ff_proresdsp_init(ProresDSPContext *dsp);

libavcodec/proresenc.c+9−8

@@ -22,6 +22,7 @@ · 21 unmodified lines
22	22
23	23	#include "libavutil/opt.h"
24	24	#include "avcodec.h"
	25	+#include "dsputil.h"
25	26	#include "put_bits.h"
26	27	#include "bytestream.h"
27	28	#include "internal.h"
@@ -167,7 +168,7 @@ struct TrellisNode { · 139 unmodified lines
167	168	#define MAX_STORED_Q 16
168	169
169	170	typedef struct ProresThreadData {
170		− DECLARE_ALIGNED(16, DCTELEM, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
	171	+ DECLARE_ALIGNED(16, int16_t, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
171	172	DECLARE_ALIGNED(16, uint16_t, emu_buf)[16 * 16];
172	173	int16_t custom_q[64];
173	174	struct TrellisNode *nodes;
@@ -175,7 +176,7 @@ typedef struct ProresThreadData { · 1 unmodified line
175	176
176	177	typedef struct ProresContext {
177	178	AVClass *class;
178		− DECLARE_ALIGNED(16, DCTELEM, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
	179	+ DECLARE_ALIGNED(16, int16_t, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
179	180	DECLARE_ALIGNED(16, uint16_t, emu_buf)[16*16];
180	181	int16_t quants[MAX_STORED_Q][64];
181	182	int16_t custom_q[64];
@@ -210,7 +211,7 @@ typedef struct ProresContext { · 28 unmodified lines
210	211
211	212	static void get_slice_data(ProresContext ctx, const uint16_t src,
212	213	int linesize, int x, int y, int w, int h,
213		− DCTELEM blocks, uint16_t emu_buf,
	214	+ int16_t blocks, uint16_t emu_buf,
214	215	int mbs_per_slice, int blocks_per_mb, int is_chroma)
215	216	{
216	217	const uint16_t *esrc;
@@ -314,7 +315,7 @@ static inline void encode_vlc_codeword(PutBitContext *pb, unsigned codebook, int · 97 unmodified lines
314	315	#define GET_SIGN(x) ((x) >> 31)
315	316	#define MAKE_CODE(x) (((x) << 1) ^ GET_SIGN(x))
316	317
317		−static void encode_dcs(PutBitContext pb, DCTELEM blocks,
	318	+static void encode_dcs(PutBitContext pb, int16_t blocks,
318	319	int blocks_per_slice, int scale)
319	320	{
320	321	int i;
@@ -340,7 +341,7 @@ static void encode_dcs(PutBitContext pb, DCTELEM blocks, · 19 unmodified lines
340	341	}
341	342	}
342	343
343		−static void encode_acs(PutBitContext pb, DCTELEM blocks,
	344	+static void encode_acs(PutBitContext pb, int16_t blocks,
344	345	int blocks_per_slice,
345	346	int plane_size_factor,
346	347	const uint8_t scan, const int16_t qmat)
@@ -376,7 +377,7 @@ static void encode_acs(PutBitContext pb, DCTELEM blocks, · 29 unmodified lines
376	377
377	378	static int encode_slice_plane(ProresContext ctx, PutBitContext pb,
378	379	const uint16_t *src, int linesize,
379		− int mbs_per_slice, DCTELEM *blocks,
	380	+ int mbs_per_slice, int16_t *blocks,
380	381	int blocks_per_mb, int plane_size_factor,
381	382	const int16_t *qmat)
382	383	{
@@ -478,7 +479,7 @@ static inline int estimate_vlc(unsigned codebook, int val) · 95 unmodified lines
478	479	}
479	480	}
480	481
481		−static int estimate_dcs(int error, DCTELEM blocks, int blocks_per_slice,
	482	+static int estimate_dcs(int error, int16_t blocks, int blocks_per_slice,
482	483	int scale)
483	484	{
484	485	int i;
@@ -509,7 +510,7 @@ static int estimate_dcs(int error, DCTELEM blocks, int blocks_per_slice, · 24 unmodified lines
509	510	return bits;
510	511	}
511	512
512		−static int estimate_acs(int error, DCTELEM blocks, int blocks_per_slice,
	513	+static int estimate_acs(int error, int16_t blocks, int blocks_per_slice,
513	514	int plane_size_factor,
514	515	const uint8_t scan, const int16_t qmat)
515	516	{

libavcodec/rtjpeg.c+3−3

@@ -44,7 +44,7 @@ · 43 unmodified lines
44	44	* aligned this could be done faster in a different way, e.g. as it is done
45	45	* in MPlayer libmpcodecs/native/rtjpegn.c.
46	46	*/
47		−static inline int get_block(GetBitContext gb, DCTELEM block, const uint8_t *scan,
	47	+static inline int get_block(GetBitContext gb, int16_t block, const uint8_t *scan,
48	48	const uint32_t *quant) {
49	49	int coeff, i, n;
50	50	int8_t ac;
@@ -61,7 +61,7 @@ static inline int get_block(GetBitContext gb, DCTELEM block, const uint8_t *sc · 10 unmodified lines
61	61
62	62	// normally we would only need to clear the (63 - coeff) last values,
63	63	// but since we do not know where they are we just clear the whole block
64		− memset(block, 0, 64 * sizeof(DCTELEM));
	64	+ memset(block, 0, 64 * sizeof(int16_t));
65	65
66	66	// 2 bits per coefficient
67	67	while (coeff) {
@@ -121,7 +121,7 @@ int ff_rtjpeg_decode_frame_yuv420(RTJpegContext c, AVFrame f, · 53 unmodified lines
121	121	if (res > 0) \
122	122	c->dsp->idct_put(dst, stride, block); \
123	123	} while (0)
124		− DCTELEM *block = c->block;
	124	+ int16_t *block = c->block;
125	125	BLOCK(c->lquant, y1, f->linesize[0]);
126	126	y1 += 8;
127	127	BLOCK(c->lquant, y1, f->linesize[0]);

libavcodec/rtjpeg.h+1−1

@@ -35,7 +35,7 @@ typedef struct RTJpegContext { · 34 unmodified lines
35	35	uint8_t scan[64];
36	36	uint32_t lquant[64];
37	37	uint32_t cquant[64];
38		− DECLARE_ALIGNED(16, DCTELEM, block)[64];
	38	+ DECLARE_ALIGNED(16, int16_t, block)[64];
39	39	} RTJpegContext;
40	40
41	41	void ff_rtjpeg_decode_init(RTJpegContext c, DSPContext dsp,

libavcodec/rv34.c+10−10

@@ -216,7 +216,7 @@ static int rv34_decode_cbp(GetBitContext gb, RV34VLC vlc, int table) · 215 unmodified lines
216	216	/**
217	217	* Get one coefficient value from the bistream and store it.
218	218	*/
219		−static inline void decode_coeff(DCTELEM dst, int coef, int esc, GetBitContext gb, VLC* vlc, int q)
	219	+static inline void decode_coeff(int16_t dst, int coef, int esc, GetBitContext gb, VLC* vlc, int q)
220	220	{
221	221	if(coef){
222	222	if(coef == esc){
@@ -236,7 +236,7 @@ static inline void decode_coeff(DCTELEM dst, int coef, int esc, GetBitContext · 13 unmodified lines
236	236	/**
237	237	* Decode 2x2 subblock of coefficients.
238	238	*/
239		−static inline void decode_subblock(DCTELEM dst, int code, const int is_block2, GetBitContext gb, VLC *vlc, int q)
	239	+static inline void decode_subblock(int16_t dst, int code, const int is_block2, GetBitContext gb, VLC *vlc, int q)
240	240	{
241	241	int flags = modulo_three_table[code];
242	242
@@ -254,13 +254,13 @@ static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2, · 11 unmodified lines
254	254	/**
255	255	* Decode a single coefficient.
256	256	*/
257		−static inline void decode_subblock1(DCTELEM dst, int code, GetBitContext gb, VLC *vlc, int q)
	257	+static inline void decode_subblock1(int16_t dst, int code, GetBitContext gb, VLC *vlc, int q)
258	258	{
259	259	int coeff = modulo_three_table[code] >> 6;
260	260	decode_coeff(dst, coeff, 3, gb, vlc, q);
261	261	}
262	262
263		−static inline void decode_subblock3(DCTELEM dst, int code, GetBitContext gb, VLC *vlc,
	263	+static inline void decode_subblock3(int16_t dst, int code, GetBitContext gb, VLC *vlc,
264	264	int q_dc, int q_ac1, int q_ac2)
265	265	{
266	266	int flags = modulo_three_table[code];
@@ -282,7 +282,7 @@ static inline void decode_subblock3(DCTELEM dst, int code, GetBitContext gb, V · 15 unmodified lines
282	282	* o--o
283	283	*/
284	284
285		−static int rv34_decode_block(DCTELEM dst, GetBitContext gb, RV34VLC *rvlc, int fc, int sc, int q_dc, int q_ac1, int q_ac2)
	285	+static int rv34_decode_block(int16_t dst, GetBitContext gb, RV34VLC *rvlc, int fc, int sc, int q_dc, int q_ac1, int q_ac2)
286	286	{
287	287	int code, pattern, has_ac = 1;
288	288
@@ -994,7 +994,7 @@ static inline void rv34_process_block(RV34DecContext *r, · 705 unmodified lines
994	994	int fc, int sc, int q_dc, int q_ac)
995	995	{
996	996	MpegEncContext *s = &r->s;
997		− DCTELEM *ptr = s->block[0];
	997	+ int16_t *ptr = s->block[0];
998	998	int has_ac = rv34_decode_block(ptr, &s->gb, r->cur_vlcs,
999	999	fc, sc, q_dc, q_ac, q_ac);
1000	1000	if(has_ac){
@@ -1007,13 +1007,13 @@ static inline void rv34_process_block(RV34DecContext *r, · 6 unmodified lines
1007	1007
1008	1008	static void rv34_output_i16x16(RV34DecContext r, int8_t intra_types, int cbp)
1009	1009	{
1010		− LOCAL_ALIGNED_16(DCTELEM, block16, [16]);
	1010	+ LOCAL_ALIGNED_16(int16_t, block16, [16]);
1011	1011	MpegEncContext *s = &r->s;
1012	1012	GetBitContext *gb = &s->gb;
1013	1013	int q_dc = rv34_qscale_tab[ r->luma_dc_quant_i[s->qscale] ],
1014	1014	q_ac = rv34_qscale_tab[s->qscale];
1015	1015	uint8_t *dst = s->dest[0];
1016		− DCTELEM *ptr = s->block[0];
	1016	+ int16_t *ptr = s->block[0];
1017	1017	int i, j, itype, has_ac;
1018	1018
1019	1019	memset(block16, 0, 16 * sizeof(*block16));
@@ -1179,7 +1179,7 @@ static int rv34_decode_inter_macroblock(RV34DecContext r, int8_t intra_types) · 159 unmodified lines
1179	1179	MpegEncContext *s = &r->s;
1180	1180	GetBitContext *gb = &s->gb;
1181	1181	uint8_t *dst = s->dest[0];
1182		− DCTELEM *ptr = s->block[0];
	1182	+ int16_t *ptr = s->block[0];
1183	1183	int mb_pos = s->mb_x + s->mb_y * s->mb_stride;
1184	1184	int cbp, cbp2;
1185	1185	int q_dc, q_ac, has_ac;
@@ -1219,7 +1219,7 @@ static int rv34_decode_inter_macroblock(RV34DecContext r, int8_t intra_types) · 33 unmodified lines
1219	1219
1220	1220	if(r->is16){
1221	1221	// Only for RV34_MB_P_MIX16x16
1222		− LOCAL_ALIGNED_16(DCTELEM, block16, [16]);
	1222	+ LOCAL_ALIGNED_16(int16_t, block16, [16]);
1223	1223	memset(block16, 0, 16 * sizeof(*block16));
1224	1224	q_dc = rv34_qscale_tab[ r->luma_dc_quant_p[s->qscale] ];
1225	1225	q_ac = rv34_qscale_tab[s->qscale];

libavcodec/rv34dsp.c+6−6

@@ -33,7 +33,7 @@ · 32 unmodified lines
33	33	* @{
34	34	*/
35	35
36		−static av_always_inline void rv34_row_transform(int temp[16], DCTELEM *block)
	36	+static av_always_inline void rv34_row_transform(int temp[16], int16_t *block)
37	37	{
38	38	int i;
39	39
@@ -54,12 +54,12 @@ static av_always_inline void rv34_row_transform(int temp[16], DCTELEM *block) · 14 unmodified lines
54	54	* Real Video 3.0/4.0 inverse transform + sample reconstruction
55	55	* Code is almost the same as in SVQ3, only scaling is different.
56	56	*/
57		−static void rv34_idct_add_c(uint8_t dst, ptrdiff_t stride, DCTELEM block){
	57	+static void rv34_idct_add_c(uint8_t dst, ptrdiff_t stride, int16_t block){
58	58	int temp[16];
59	59	int i;
60	60
61	61	rv34_row_transform(temp, block);
62		− memset(block, 0, 16*sizeof(DCTELEM));
	62	+ memset(block, 0, 16*sizeof(int16_t));
63	63
64	64	for(i = 0; i < 4; i++){
65	65	const int z0 = 13(temp[40+i] + temp[4*2+i]) + 0x200;
@@ -82,7 +82,7 @@ static void rv34_idct_add_c(uint8_t dst, ptrdiff_t stride, DCTELEM block){ · 16 unmodified lines
82	82	* Code is almost the same as rv34_inv_transform()
83	83	* but final coefficients are multiplied by 1.5 and have no rounding.
84	84	*/
85		−static void rv34_inv_transform_noround_c(DCTELEM *block){
	85	+static void rv34_inv_transform_noround_c(int16_t *block){
86	86	int temp[16];
87	87	int i;
88	88
@@ -115,9 +115,9 @@ static void rv34_idct_dc_add_c(uint8_t *dst, ptrdiff_t stride, int dc) · 26 unmodified lines
115	115	}
116	116	}
117	117
118		−static void rv34_inv_transform_dc_noround_c(DCTELEM *block)
	118	+static void rv34_inv_transform_dc_noround_c(int16_t *block)
119	119	{
120		− DCTELEM dc = (13 * 13 * 3 * block[0]) >> 11;
	120	+ int16_t dc = (13 * 13 * 3 * block[0]) >> 11;
121	121	int i, j;
122	122
123	123	for (i = 0; i < 4; i++, block += 4)

libavcodec/rv34dsp.h+2−2

@@ -34,9 +34,9 @@ typedef void (rv40_weight_func)(uint8_t dst/align width (8 or 16)/, · 33 unmodified lines
34	34	uint8_t src2/align width (8 or 16)*/,
35	35	int w1, int w2, ptrdiff_t stride);
36	36
37		−typedef void (rv34_inv_transform_func)(DCTELEM block);
	37	+typedef void (rv34_inv_transform_func)(int16_t block);
38	38
39		−typedef void (rv34_idct_add_func)(uint8_t dst, ptrdiff_t stride, DCTELEM *block);
	39	+typedef void (rv34_idct_add_func)(uint8_t dst, ptrdiff_t stride, int16_t *block);
40	40	typedef void (rv34_idct_dc_add_func)(uint8_t dst, ptrdiff_t stride,
41	41	int dc);
42	42

libavcodec/sh4/dsputil_sh4.c+4−4

@@ -47,12 +47,12 @@ static void memzero_align8(void *dst,size_t size) · 46 unmodified lines
47	47	fp_single_leave(fpscr);
48	48	}
49	49
50		−static void clear_blocks_sh4(DCTELEM *blocks)
	50	+static void clear_blocks_sh4(int16_t *blocks)
51	51	{
52		− memzero_align8(blocks,sizeof(DCTELEM)664);
	52	+ memzero_align8(blocks,sizeof(int16_t)664);
53	53	}
54	54
55		−static void idct_put(uint8_t dest, int line_size, DCTELEM block)
	55	+static void idct_put(uint8_t dest, int line_size, int16_t block)
56	56	{
57	57	int i;
58	58	uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
@@ -70,7 +70,7 @@ static void idct_put(uint8_t dest, int line_size, DCTELEM block) · 11 unmodified lines
70	70	block+=8;
71	71	}
72	72	}
73		−static void idct_add(uint8_t dest, int line_size, DCTELEM block)
	73	+static void idct_add(uint8_t dest, int line_size, int16_t block)
74	74	{
75	75	int i;
76	76	uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;

libavcodec/sh4/dsputil_sh4.h+1−1

@@ -22,7 +22,7 @@ · 21 unmodified lines
22	22	#include "libavcodec/avcodec.h"
23	23	#include "libavcodec/dsputil.h"
24	24
25		−void ff_idct_sh4(DCTELEM *block);
	25	+void ff_idct_sh4(int16_t *block);
26	26	void ff_dsputil_init_align(DSPContext* c, AVCodecContext *avctx);
27	27
28	28	#endif /* AVCODEC_SH4_DSPUTIL_SH4_H */

libavcodec/sh4/idct_sh4.c+1−1

@@ -89,7 +89,7 @@ static const float odd_table[] __attribute__ ((aligned(8))) = { · 88 unmodified lines
89	89
90	90	//optimized
91	91
92		−void ff_idct_sh4(DCTELEM *block)
	92	+void ff_idct_sh4(int16_t *block)
93	93	{
94	94	DEFREG;
95	95

libavcodec/simple_idct.c+9−9

@@ -50,7 +50,7 @@ · 49 unmodified lines
50	50	and the butterfly must be multiplied by 0.5 * sqrt(2.0) */
51	51	#define C_SHIFT (4+1+12)
52	52
53		−static inline void idct4col_put(uint8_t dest, int line_size, const DCTELEM col)
	53	+static inline void idct4col_put(uint8_t dest, int line_size, const int16_t col)
54	54	{
55	55	int c0, c1, c2, c3, a0, a1, a2, a3;
56	56
@@ -86,10 +86,10 @@ static inline void idct4col_put(uint8_t dest, int line_size, const DCTELEM col · 29 unmodified lines
86	86	/* XXX: I think a 1.0/sqrt(2) normalization should be needed to
87	87	compensate the extra butterfly stage - I don't have the full DV
88	88	specification */
89		−void ff_simple_idct248_put(uint8_t dest, int line_size, DCTELEM block)
	89	+void ff_simple_idct248_put(uint8_t dest, int line_size, int16_t block)
90	90	{
91	91	int i;
92		− DCTELEM *ptr;
	92	+ int16_t *ptr;
93	93
94	94	/* butterfly */
95	95	ptr = block;
@@ -129,7 +129,7 @@ void ff_simple_idct248_put(uint8_t dest, int line_size, DCTELEM block) · 33 unmodified lines
129	129	#define C2 C_FIX(0.2705980501)
130	130	#define C3 C_FIX(0.5)
131	131	#define C_SHIFT (4+1+12)
132		−static inline void idct4col_add(uint8_t dest, int line_size, const DCTELEM col)
	132	+static inline void idct4col_add(uint8_t dest, int line_size, const int16_t col)
133	133	{
134	134	int c0, c1, c2, c3, a0, a1, a2, a3;
135	135
@@ -156,7 +156,7 @@ static inline void idct4col_add(uint8_t dest, int line_size, const DCTELEM col · 20 unmodified lines
156	156	#define R2 R_FIX(0.2705980501)
157	157	#define R3 R_FIX(0.5)
158	158	#define R_SHIFT 11
159		−static inline void idct4row(DCTELEM *row)
	159	+static inline void idct4row(int16_t *row)
160	160	{
161	161	int c0, c1, c2, c3, a0, a1, a2, a3;
162	162
@@ -174,7 +174,7 @@ static inline void idct4row(DCTELEM *row) · 11 unmodified lines
174	174	row[3]= (c0 - c1) >> R_SHIFT;
175	175	}
176	176
177		−void ff_simple_idct84_add(uint8_t dest, int line_size, DCTELEM block)
	177	+void ff_simple_idct84_add(uint8_t dest, int line_size, int16_t block)
178	178	{
179	179	int i;
180	180
@@ -189,7 +189,7 @@ void ff_simple_idct84_add(uint8_t dest, int line_size, DCTELEM block) · 8 unmodified lines
189	189	}
190	190	}
191	191
192		−void ff_simple_idct48_add(uint8_t dest, int line_size, DCTELEM block)
	192	+void ff_simple_idct48_add(uint8_t dest, int line_size, int16_t block)
193	193	{
194	194	int i;
195	195
@@ -204,7 +204,7 @@ void ff_simple_idct48_add(uint8_t dest, int line_size, DCTELEM block) · 8 unmodified lines
204	204	}
205	205	}
206	206
207		−void ff_simple_idct44_add(uint8_t dest, int line_size, DCTELEM block)
	207	+void ff_simple_idct44_add(uint8_t dest, int line_size, int16_t block)
208	208	{
209	209	int i;
210	210
@@ -219,7 +219,7 @@ void ff_simple_idct44_add(uint8_t dest, int line_size, DCTELEM block) · 8 unmodified lines
219	219	}
220	220	}
221	221
222		−void ff_prores_idct(DCTELEM block, const int16_t qmat)
	222	+void ff_prores_idct(int16_t block, const int16_t qmat)
223	223	{
224	224	int i;
225	225

libavcodec/simple_idct.h+11−12

@@ -29,30 +29,29 @@ · 28 unmodified lines
29	29	#define AVCODEC_SIMPLE_IDCT_H
30	30
31	31	#include <stdint.h>
32		−#include "dsputil.h"
33	32
34		−void ff_simple_idct_put_8(uint8_t dest, int line_size, DCTELEM block);
35		−void ff_simple_idct_add_8(uint8_t dest, int line_size, DCTELEM block);
36		−void ff_simple_idct_8(DCTELEM *block);
	33	+void ff_simple_idct_put_8(uint8_t dest, int line_size, int16_t block);
	34	+void ff_simple_idct_add_8(uint8_t dest, int line_size, int16_t block);
	35	+void ff_simple_idct_8(int16_t *block);
37	36
38		−void ff_simple_idct_put_10(uint8_t dest, int line_size, DCTELEM block);
39		−void ff_simple_idct_add_10(uint8_t dest, int line_size, DCTELEM block);
40		−void ff_simple_idct_10(DCTELEM *block);
	37	+void ff_simple_idct_put_10(uint8_t dest, int line_size, int16_t block);
	38	+void ff_simple_idct_add_10(uint8_t dest, int line_size, int16_t block);
	39	+void ff_simple_idct_10(int16_t *block);
41	40	/**
42	41	* Special version of ff_simple_idct_10() which does dequantization
43	42	* and scales by a factor of 2 more between the two IDCTs to account
44	43	* for larger scale of input coefficients.
45	44	*/
46		−void ff_prores_idct(DCTELEM block, const int16_t qmat);
	45	+void ff_prores_idct(int16_t block, const int16_t qmat);
47	46
48	47	void ff_simple_idct_mmx(int16_t *block);
49	48	void ff_simple_idct_add_mmx(uint8_t dest, int line_size, int16_t block);
50	49	void ff_simple_idct_put_mmx(uint8_t dest, int line_size, int16_t block);
51	50
52		−void ff_simple_idct248_put(uint8_t dest, int line_size, DCTELEM block);
	51	+void ff_simple_idct248_put(uint8_t dest, int line_size, int16_t block);
53	52
54		−void ff_simple_idct84_add(uint8_t dest, int line_size, DCTELEM block);
55		−void ff_simple_idct48_add(uint8_t dest, int line_size, DCTELEM block);
56		−void ff_simple_idct44_add(uint8_t dest, int line_size, DCTELEM block);
	53	+void ff_simple_idct84_add(uint8_t dest, int line_size, int16_t block);
	54	+void ff_simple_idct48_add(uint8_t dest, int line_size, int16_t block);
	55	+void ff_simple_idct44_add(uint8_t dest, int line_size, int16_t block);
57	56
58	57	#endif /* AVCODEC_SIMPLE_IDCT_H */

libavcodec/simple_idct_template.c+7−7

@@ -85,7 +85,7 @@ · 84 unmodified lines
85	85
86	86	#endif
87	87
88		−static inline void FUNC(idctRowCondDC)(DCTELEM *row, int extra_shift)
	88	+static inline void FUNC(idctRowCondDC)(int16_t *row, int extra_shift)
89	89	{
90	90	int a0, a1, a2, a3, b0, b1, b2, b3;
91	91
@@ -221,7 +221,7 @@ static inline void FUNC(idctRowCondDC)(DCTELEM *row, int extra_shift) · 129 unmodified lines
221	221	} while (0)
222	222
223	223	static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size,
224		− DCTELEM *col)
	224	+ int16_t *col)
225	225	{
226	226	int a0, a1, a2, a3, b0, b1, b2, b3;
227	227
@@ -245,7 +245,7 @@ static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size, · 17 unmodified lines
245	245	}
246	246
247	247	static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size,
248		− DCTELEM *col)
	248	+ int16_t *col)
249	249	{
250	250	int a0, a1, a2, a3, b0, b1, b2, b3;
251	251
@@ -268,7 +268,7 @@ static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size, · 16 unmodified lines
268	268	dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT));
269	269	}
270	270
271		−static inline void FUNC(idctSparseCol)(DCTELEM *col)
	271	+static inline void FUNC(idctSparseCol)(int16_t *col)
272	272	{
273	273	int a0, a1, a2, a3, b0, b1, b2, b3;
274	274
@@ -284,7 +284,7 @@ static inline void FUNC(idctSparseCol)(DCTELEM *col) · 9 unmodified lines
284	284	col[56] = ((a0 - b0) >> COL_SHIFT);
285	285	}
286	286
287		−void FUNC(ff_simple_idct_put)(uint8_t dest_, int line_size, DCTELEM block)
	287	+void FUNC(ff_simple_idct_put)(uint8_t dest_, int line_size, int16_t block)
288	288	{
289	289	pixel dest = (pixel )dest_;
290	290	int i;
@@ -298,7 +298,7 @@ void FUNC(ff_simple_idct_put)(uint8_t dest_, int line_size, DCTELEM block) · 7 unmodified lines
298	298	FUNC(idctSparseColPut)(dest + i, line_size, block + i);
299	299	}
300	300
301		−void FUNC(ff_simple_idct_add)(uint8_t dest_, int line_size, DCTELEM block)
	301	+void FUNC(ff_simple_idct_add)(uint8_t dest_, int line_size, int16_t block)
302	302	{
303	303	pixel dest = (pixel )dest_;
304	304	int i;
@@ -312,7 +312,7 @@ void FUNC(ff_simple_idct_add)(uint8_t dest_, int line_size, DCTELEM block) · 7 unmodified lines
312	312	FUNC(idctSparseColAdd)(dest + i, line_size, block + i);
313	313	}
314	314
315		−void FUNC(ff_simple_idct)(DCTELEM *block)
	315	+void FUNC(ff_simple_idct)(int16_t *block)
316	316	{
317	317	int i;
318	318

libavcodec/sparc/dsputil_vis.h+3−4

@@ -20,10 +20,9 @@ · 19 unmodified lines
20	20	#define AVCODEC_SPARC_DSPUTIL_VIS_H
21	21
22	22	#include <stdint.h>
23		−#include "libavcodec/dsputil.h"
24	23
25		−void ff_simple_idct_put_vis(uint8_t dest, int line_size, DCTELEM data);
26		−void ff_simple_idct_add_vis(uint8_t dest, int line_size, DCTELEM data);
27		−void ff_simple_idct_vis(DCTELEM *data);
	24	+void ff_simple_idct_put_vis(uint8_t dest, int line_size, int16_t data);
	25	+void ff_simple_idct_add_vis(uint8_t dest, int line_size, int16_t data);
	26	+void ff_simple_idct_vis(int16_t *data);
28	27
29	28	#endif /* AVCODEC_SPARC_DSPUTIL_VIS_H */

libavcodec/sparc/simple_idct_vis.c+3−3

@@ -388,7 +388,7 @@ static const DECLARE_ALIGNED(8, uint16_t, expand)[4] = { · 387 unmodified lines
388	388	"st %%f14, [%12+" dest "] \n\t"\
389	389
390	390
391		−void ff_simple_idct_vis(DCTELEM *data) {
	391	+void ff_simple_idct_vis(int16_t *data) {
392	392	int out1, out2, out3, out4;
393	393	DECLARE_ALIGNED(8, int16_t, temp)[8*8];
394	394
@@ -428,7 +428,7 @@ void ff_simple_idct_vis(DCTELEM *data) { · 33 unmodified lines
428	428	);
429	429	}
430	430
431		−void ff_simple_idct_put_vis(uint8_t dest, int line_size, DCTELEM data) {
	431	+void ff_simple_idct_put_vis(uint8_t dest, int line_size, int16_t data) {
432	432	int out1, out2, out3, out4, out5;
433	433	int r1, r2, r3, r4, r5, r6, r7;
434	434
@@ -478,7 +478,7 @@ void ff_simple_idct_put_vis(uint8_t dest, int line_size, DCTELEM data) { · 43 unmodified lines
478	478	);
479	479	}
480	480
481		−void ff_simple_idct_add_vis(uint8_t dest, int line_size, DCTELEM data) {
	481	+void ff_simple_idct_add_vis(uint8_t dest, int line_size, int16_t data) {
482	482	int out1, out2, out3, out4, out5, out6;
483	483	int r1, r2, r3, r4, r5, r6, r7;
484	484

libavcodec/svq3.c+3−3

@@ -137,7 +137,7 @@ static const uint32_t svq3_dequant_coeff[32] = { · 136 unmodified lines
137	137	61694, 68745, 77615, 89113, 100253, 109366, 126635, 141533
138	138	};
139	139
140		−void ff_svq3_luma_dc_dequant_idct_c(DCTELEM output, DCTELEM input, int qp)
	140	+void ff_svq3_luma_dc_dequant_idct_c(int16_t output, int16_t input, int qp)
141	141	{
142	142	const int qmul = svq3_dequant_coeff[qp];
143	143	#define stride 16
@@ -172,7 +172,7 @@ void ff_svq3_luma_dc_dequant_idct_c(DCTELEM output, DCTELEM input, int qp) · 28 unmodified lines
172	172	}
173	173	#undef stride
174	174
175		−void ff_svq3_add_idct_c(uint8_t dst, DCTELEM block,
	175	+void ff_svq3_add_idct_c(uint8_t dst, int16_t block,
176	176	int stride, int qp, int dc)
177	177	{
178	178	const int qmul = svq3_dequant_coeff[qp];
@@ -210,7 +210,7 @@ void ff_svq3_add_idct_c(uint8_t dst, DCTELEM block, · 31 unmodified lines
210	210	}
211	211	}
212	212
213		−static inline int svq3_decode_block(GetBitContext gb, DCTELEM block,
	213	+static inline int svq3_decode_block(GetBitContext gb, int16_t block,
214	214	int index, const int type)
215	215	{
216	216	static const uint8_t *const scan_patterns[4] =

libavcodec/vc1.h+1−1

@@ -385,7 +385,7 @@ typedef struct VC1Context{ · 384 unmodified lines
385	385	int bi_type;
386	386	int x8_type;
387	387
388		− DCTELEM (*block)[6][64];
	388	+ int16_t (*block)[6][64];
389	389	int n_allocated_blks, cur_blk_idx, left_blk_idx, topleft_blk_idx, top_blk_idx;
390	390	uint32_t cbp_base, cbp;
391	391	uint8_t is_intra_base, is_intra;

libavcodec/vc1dec.c+5−5

@@ -2538,7 +2538,7 @@ static void vc1_decode_ac_coeff(VC1Context v, int last, int *skip, · 2537 unmodified lines
2538	2538	* @param coded are AC coeffs present or not
2539	2539	* @param codingset set of VLC to decode data
2540	2540	*/
2541		−static int vc1_decode_i_block(VC1Context *v, DCTELEM block[64], int n,
	2541	+static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n,
2542	2542	int coded, int codingset)
2543	2543	{
2544	2544	GetBitContext *gb = &v->s.gb;
@@ -2701,7 +2701,7 @@ not_coded: · 156 unmodified lines
2701	2701	* @param codingset set of VLC to decode data
2702	2702	* @param mquant quantizer value for this macroblock
2703	2703	*/
2704		−static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n,
	2704	+static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n,
2705	2705	int coded, int codingset, int mquant)
2706	2706	{
2707	2707	GetBitContext *gb = &v->s.gb;
@@ -2913,7 +2913,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n, · 205 unmodified lines
2913	2913	* @param mquant block quantizer
2914	2914	* @param codingset set of VLC to decode data
2915	2915	*/
2916		−static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n,
	2916	+static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n,
2917	2917	int coded, int mquant, int codingset)
2918	2918	{
2919	2919	GetBitContext *gb = &v->s.gb;
@@ -3123,7 +3123,7 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n, · 203 unmodified lines
3123	3123
3124	3124	/** Decode P block
3125	3125	*/
3126		−static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n,
	3126	+static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n,
3127	3127	int mquant, int ttmb, int first_block,
3128	3128	uint8_t *dst, int linesize, int skip_block,
3129	3129	int *ttmb_out)
@@ -4505,7 +4505,7 @@ static void vc1_decode_i_blocks_adv(VC1Context *v) · 1375 unmodified lines
4505	4505	s->mb_x = 0;
4506	4506	ff_init_block_index(s);
4507	4507	for (;s->mb_x < s->mb_width; s->mb_x++) {
4508		− DCTELEM (*block)[64] = v->block[v->cur_blk_idx];
	4508	+ int16_t (*block)[64] = v->block[v->cur_blk_idx];
4509	4509	ff_update_block_index(s);
4510	4510	s->dsp.clear_blocks(block[0]);
4511	4511	mb_pos = s->mb_x + s->mb_y * s->mb_stride;

libavcodec/vc1dsp.c+14−14

@@ -79,7 +79,7 @@ static void vc1_h_overlap_c(uint8_t* src, int stride) · 78 unmodified lines
79	79	}
80	80	}
81	81
82		−static void vc1_v_s_overlap_c(DCTELEM top, DCTELEM bottom)
	82	+static void vc1_v_s_overlap_c(int16_t top, int16_t bottom)
83	83	{
84	84	int i;
85	85	int a, b, c, d;
@@ -105,7 +105,7 @@ static void vc1_v_s_overlap_c(DCTELEM top, DCTELEM bottom) · 19 unmodified lines
105	105	}
106	106	}
107	107
108		−static void vc1_h_s_overlap_c(DCTELEM left, DCTELEM right)
	108	+static void vc1_h_s_overlap_c(int16_t left, int16_t right)
109	109	{
110	110	int i;
111	111	int a, b, c, d;
@@ -229,7 +229,7 @@ static void vc1_h_loop_filter16_c(uint8_t *src, int stride, int pq) · 117 unmodified lines
229	229
230	230	/** Do inverse transform on 8x8 block
231	231	*/
232		−static void vc1_inv_trans_8x8_dc_c(uint8_t dest, int linesize, DCTELEM block)
	232	+static void vc1_inv_trans_8x8_dc_c(uint8_t dest, int linesize, int16_t block)
233	233	{
234	234	int i;
235	235	int dc = block[0];
@@ -248,11 +248,11 @@ static void vc1_inv_trans_8x8_dc_c(uint8_t dest, int linesize, DCTELEM block) · 12 unmodified lines
248	248	}
249	249	}
250	250
251		−static void vc1_inv_trans_8x8_c(DCTELEM block[64])
	251	+static void vc1_inv_trans_8x8_c(int16_t block[64])
252	252	{
253	253	int i;
254	254	register int t1,t2,t3,t4,t5,t6,t7,t8;
255		− DCTELEM src, dst, temp[64];
	255	+ int16_t src, dst, temp[64];
256	256
257	257	src = block;
258	258	dst = temp;
@@ -319,7 +319,7 @@ static void vc1_inv_trans_8x8_c(DCTELEM block[64]) · 60 unmodified lines
319	319
320	320	/** Do inverse transform on 8x4 part of block
321	321	*/
322		−static void vc1_inv_trans_8x4_dc_c(uint8_t dest, int linesize, DCTELEM block)
	322	+static void vc1_inv_trans_8x4_dc_c(uint8_t dest, int linesize, int16_t block)
323	323	{
324	324	int i;
325	325	int dc = block[0];
@@ -338,11 +338,11 @@ static void vc1_inv_trans_8x4_dc_c(uint8_t dest, int linesize, DCTELEM block) · 12 unmodified lines
338	338	}
339	339	}
340	340
341		−static void vc1_inv_trans_8x4_c(uint8_t dest, int linesize, DCTELEM block)
	341	+static void vc1_inv_trans_8x4_c(uint8_t dest, int linesize, int16_t block)
342	342	{
343	343	int i;
344	344	register int t1,t2,t3,t4,t5,t6,t7,t8;
345		− DCTELEM src, dst;
	345	+ int16_t src, dst;
346	346
347	347	src = block;
348	348	dst = block;
@@ -394,7 +394,7 @@ static void vc1_inv_trans_8x4_c(uint8_t dest, int linesize, DCTELEM block) · 45 unmodified lines
394	394
395	395	/** Do inverse transform on 4x8 parts of block
396	396	*/
397		−static void vc1_inv_trans_4x8_dc_c(uint8_t dest, int linesize, DCTELEM block)
	397	+static void vc1_inv_trans_4x8_dc_c(uint8_t dest, int linesize, int16_t block)
398	398	{
399	399	int i;
400	400	int dc = block[0];
@@ -409,11 +409,11 @@ static void vc1_inv_trans_4x8_dc_c(uint8_t dest, int linesize, DCTELEM block) · 8 unmodified lines
409	409	}
410	410	}
411	411
412		−static void vc1_inv_trans_4x8_c(uint8_t dest, int linesize, DCTELEM block)
	412	+static void vc1_inv_trans_4x8_c(uint8_t dest, int linesize, int16_t block)
413	413	{
414	414	int i;
415	415	register int t1,t2,t3,t4,t5,t6,t7,t8;
416		− DCTELEM src, dst;
	416	+ int16_t src, dst;
417	417
418	418	src = block;
419	419	dst = block;
@@ -465,7 +465,7 @@ static void vc1_inv_trans_4x8_c(uint8_t dest, int linesize, DCTELEM block) · 45 unmodified lines
465	465
466	466	/** Do inverse transform on 4x4 part of block
467	467	*/
468		−static void vc1_inv_trans_4x4_dc_c(uint8_t dest, int linesize, DCTELEM block)
	468	+static void vc1_inv_trans_4x4_dc_c(uint8_t dest, int linesize, int16_t block)
469	469	{
470	470	int i;
471	471	int dc = block[0];
@@ -480,11 +480,11 @@ static void vc1_inv_trans_4x4_dc_c(uint8_t dest, int linesize, DCTELEM block) · 8 unmodified lines
480	480	}
481	481	}
482	482
483		−static void vc1_inv_trans_4x4_c(uint8_t dest, int linesize, DCTELEM block)
	483	+static void vc1_inv_trans_4x4_c(uint8_t dest, int linesize, int16_t block)
484	484	{
485	485	int i;
486	486	register int t1,t2,t3,t4;
487		− DCTELEM src, dst;
	487	+ int16_t src, dst;
488	488
489	489	src = block;
490	490	dst = block;

libavcodec/vc1dsp.h+10−10

@@ -32,18 +32,18 @@ · 31 unmodified lines
32	32
33	33	typedef struct VC1DSPContext {
34	34	/* vc1 functions */
35		− void (vc1_inv_trans_8x8)(DCTELEM b);
36		− void (vc1_inv_trans_8x4)(uint8_t dest, int line_size, DCTELEM *block);
37		− void (vc1_inv_trans_4x8)(uint8_t dest, int line_size, DCTELEM *block);
38		− void (vc1_inv_trans_4x4)(uint8_t dest, int line_size, DCTELEM *block);
39		− void (vc1_inv_trans_8x8_dc)(uint8_t dest, int line_size, DCTELEM *block);
40		− void (vc1_inv_trans_8x4_dc)(uint8_t dest, int line_size, DCTELEM *block);
41		− void (vc1_inv_trans_4x8_dc)(uint8_t dest, int line_size, DCTELEM *block);
42		− void (vc1_inv_trans_4x4_dc)(uint8_t dest, int line_size, DCTELEM *block);
	35	+ void (vc1_inv_trans_8x8)(int16_t b);
	36	+ void (vc1_inv_trans_8x4)(uint8_t dest, int line_size, int16_t *block);
	37	+ void (vc1_inv_trans_4x8)(uint8_t dest, int line_size, int16_t *block);
	38	+ void (vc1_inv_trans_4x4)(uint8_t dest, int line_size, int16_t *block);
	39	+ void (vc1_inv_trans_8x8_dc)(uint8_t dest, int line_size, int16_t *block);
	40	+ void (vc1_inv_trans_8x4_dc)(uint8_t dest, int line_size, int16_t *block);
	41	+ void (vc1_inv_trans_4x8_dc)(uint8_t dest, int line_size, int16_t *block);
	42	+ void (vc1_inv_trans_4x4_dc)(uint8_t dest, int line_size, int16_t *block);
43	43	void (vc1_v_overlap)(uint8_t src, int stride);
44	44	void (vc1_h_overlap)(uint8_t src, int stride);
45		− void (vc1_v_s_overlap)(DCTELEM top, DCTELEM *bottom);
46		− void (vc1_h_s_overlap)(DCTELEM left, DCTELEM *right);
	45	+ void (vc1_v_s_overlap)(int16_t top, int16_t *bottom);
	46	+ void (vc1_h_s_overlap)(int16_t left, int16_t *right);
47	47	void (vc1_v_loop_filter4)(uint8_t src, int stride, int pq);
48	48	void (vc1_h_loop_filter4)(uint8_t src, int stride, int pq);
49	49	void (vc1_v_loop_filter8)(uint8_t src, int stride, int pq);

libavcodec/vp3.c+4−4

@@ -138,7 +138,7 @@ typedef struct Vp3DecodeContext { · 137 unmodified lines
138	138	DSPContext dsp;
139	139	VideoDSPContext vdsp;
140	140	VP3DSPContext vp3dsp;
141		− DECLARE_ALIGNED(16, DCTELEM, block)[64];
	141	+ DECLARE_ALIGNED(16, int16_t, block)[64];
142	142	int flipped_image;
143	143	int last_slice_end;
144	144	int skip_loop_filter;
@@ -920,7 +920,7 @@ static int unpack_vlcs(Vp3DecodeContext s, GetBitContext gb, · 775 unmodified lines
920	920	int i, j = 0;
921	921	int token;
922	922	int zero_run = 0;
923		− DCTELEM coeff = 0;
	923	+ int16_t coeff = 0;
924	924	int bits_to_get;
925	925	int blocks_ended;
926	926	int coeff_i = 0;
@@ -1350,7 +1350,7 @@ static void apply_loop_filter(Vp3DecodeContext *s, int plane, int ystart, int ye · 423 unmodified lines
1350	1350	* for the next block in coding order
1351	1351	*/
1352	1352	static inline int vp3_dequant(Vp3DecodeContext s, Vp3Fragment frag,
1353		− int plane, int inter, DCTELEM block[64])
	1353	+ int plane, int inter, int16_t block[64])
1354	1354	{
1355	1355	int16_t *dequantizer = s->qmat[frag->qpi][inter][plane];
1356	1356	uint8_t *perm = s->scantable.permutated;
@@ -1459,7 +1459,7 @@ static void await_reference_row(Vp3DecodeContext s, Vp3Fragment fragment, int · 102 unmodified lines
1459	1459	static void render_slice(Vp3DecodeContext *s, int slice)
1460	1460	{
1461	1461	int x, y, i, j, fragment;
1462		− DCTELEM *block = s->block;
	1462	+ int16_t *block = s->block;
1463	1463	int motion_x = 0xdeadbeef, motion_y = 0xdeadbeef;
1464	1464	int motion_halfpel_index;
1465	1465	uint8_t *motion_source;

libavcodec/vp3dsp.c+8−3

@@ -213,18 +213,23 @@ static av_always_inline void idct(uint8_t dst, int stride, int16_t input, int · 212 unmodified lines
213	213	}
214	214	}
215	215
216		−static void vp3_idct_put_c(uint8_t dest/align 8/, int line_size, DCTELEM block/align 16/){
	216	+static void vp3_idct_put_c(uint8_t dest/align 8*/, int line_size,
	217	+ int16_t block/align 16*/)
	218	+{
217	219	idct(dest, line_size, block, 1);
218	220	memset(block, 0, sizeof(block) 64);
219	221	}
220	222
221		−static void vp3_idct_add_c(uint8_t dest/align 8/, int line_size, DCTELEM block/align 16/){
	223	+static void vp3_idct_add_c(uint8_t dest/align 8*/, int line_size,
	224	+ int16_t block/align 16*/)
	225	+{
222	226	idct(dest, line_size, block, 2);
223	227	memset(block, 0, sizeof(block) 64);
224	228	}
225	229
226	230	static void vp3_idct_dc_add_c(uint8_t dest/align 8*/, int line_size,
227		− DCTELEM block/align 16*/){
	231	+ int16_t block/align 16*/)
	232	+{
228	233	int i, dc = (block[0] + 15) >> 5;
229	234
230	235	for(i = 0; i < 8; i++){

libavcodec/vp3dsp.h+3−4

@@ -21,7 +21,6 @@ · 20 unmodified lines
21	21
22	22	#include <stddef.h>
23	23	#include <stdint.h>
24		−#include "dsputil.h"
25	24
26	25	typedef struct VP3DSPContext {
27	26	/**
@@ -39,9 +38,9 @@ typedef struct VP3DSPContext { · 11 unmodified lines
39	38	const uint8_t *b,
40	39	ptrdiff_t stride, int h);
41	40
42		− void (idct_put)(uint8_t dest, int line_size, DCTELEM *block);
43		− void (idct_add)(uint8_t dest, int line_size, DCTELEM *block);
44		− void (idct_dc_add)(uint8_t dest, int line_size, DCTELEM *block);
	41	+ void (idct_put)(uint8_t dest, int line_size, int16_t *block);
	42	+ void (idct_add)(uint8_t dest, int line_size, int16_t *block);
	43	+ void (idct_dc_add)(uint8_t dest, int line_size, int16_t *block);
45	44	void (v_loop_filter)(uint8_t src, int stride, int *bounding_values);
46	45	void (h_loop_filter)(uint8_t src, int stride, int *bounding_values);
47	46

libavcodec/vp56.h+3−3

@@ -67,7 +67,7 @@ typedef struct VP56RangeCoder { · 66 unmodified lines
67	67	typedef struct VP56RefDc {
68	68	uint8_t not_null_dc;
69	69	VP56Frame ref_frame;
70		− DCTELEM dc_coeff;
	70	+ int16_t dc_coeff;
71	71	} VP56RefDc;
72	72
73	73	typedef struct VP56Macroblock {
@@ -124,12 +124,12 @@ struct vp56_context { · 50 unmodified lines
124	124	VP56RefDc *above_blocks;
125	125	VP56RefDc left_block[4];
126	126	int above_block_idx[6];
127		− DCTELEM prev_dc[3][3]; /* [plan][ref_frame] */
	127	+ int16_t prev_dc[3][3]; /* [plan][ref_frame] */
128	128
129	129	/* blocks / macroblock */
130	130	VP56mb mb_type;
131	131	VP56Macroblock *macroblocks;
132		− DECLARE_ALIGNED(16, DCTELEM, block_coeff)[6][64];
	132	+ DECLARE_ALIGNED(16, int16_t, block_coeff)[6][64];
133	133
134	134	/* motion vectors */
135	135	VP56mv mv[6]; /* vectors for each block in MB */

libavcodec/vp8.c+2−2

@@ -760,7 +760,7 @@ void decode_mb_mode(VP8Context s, VP8Macroblock mb, int mb_x, int mb_y, · 759 unmodified lines
760	760	* @return 0 if no coeffs were decoded
761	761	* otherwise, the index of the last coeff decoded plus one
762	762	*/
763		−static int decode_block_coeffs_internal(VP56RangeCoder *r, DCTELEM block[16],
	763	+static int decode_block_coeffs_internal(VP56RangeCoder *r, int16_t block[16],
764	764	uint8_t probs[16][3][NUM_DCT_TOKENS-1],
765	765	int i, uint8_t *token_prob, int16_t qmul[2])
766	766	{
@@ -828,7 +828,7 @@ skip_eob: · 61 unmodified lines
828	828	* otherwise, the index of the last coeff decoded plus one
829	829	*/
830	830	static av_always_inline
831		−int decode_block_coeffs(VP56RangeCoder *c, DCTELEM block[16],
	831	+int decode_block_coeffs(VP56RangeCoder *c, int16_t block[16],
832	832	uint8_t probs[16][3][NUM_DCT_TOKENS-1],
833	833	int i, int zero_nhood, int16_t qmul[2])
834	834	{

libavcodec/vp8.h+2−2

@@ -94,8 +94,8 @@ typedef struct VP8Macroblock { · 93 unmodified lines
94	94	} VP8Macroblock;
95	95
96	96	typedef struct VP8ThreadData {
97		− DECLARE_ALIGNED(16, DCTELEM, block)[6][4][16];
98		− DECLARE_ALIGNED(16, DCTELEM, block_dc)[16];
	97	+ DECLARE_ALIGNED(16, int16_t, block)[6][4][16];
	98	+ DECLARE_ALIGNED(16, int16_t, block_dc)[16];
99	99	/**
100	100	* This is the index plus one of the last non-zero coeff
101	101	* for each of the blocks in the current macroblock.

libavcodec/vp8dsp.c+7−7

@@ -29,7 +29,7 @@ · 28 unmodified lines
29	29	#include "libavutil/common.h"
30	30
31	31	// TODO: Maybe add dequant
32		−static void vp8_luma_dc_wht_c(DCTELEM block[4][4][16], DCTELEM dc[16])
	32	+static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
33	33	{
34	34	int i, t0, t1, t2, t3;
35	35
@@ -62,7 +62,7 @@ static void vp8_luma_dc_wht_c(DCTELEM block[4][4][16], DCTELEM dc[16]) · 26 unmodified lines
62	62	}
63	63	}
64	64
65		−static void vp8_luma_dc_wht_dc_c(DCTELEM block[4][4][16], DCTELEM dc[16])
	65	+static void vp8_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
66	66	{
67	67	int i, val = (dc[0] + 3) >> 3;
68	68	dc[0] = 0;
@@ -78,10 +78,10 @@ static void vp8_luma_dc_wht_dc_c(DCTELEM block[4][4][16], DCTELEM dc[16]) · 9 unmodified lines
78	78	#define MUL_20091(a) ((((a)*20091) >> 16) + (a))
79	79	#define MUL_35468(a) (((a)*35468) >> 16)
80	80
81		−static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
	81	+static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
82	82	{
83	83	int i, t0, t1, t2, t3;
84		− DCTELEM tmp[16];
	84	+ int16_t tmp[16];
85	85
86	86	for (i = 0; i < 4; i++) {
87	87	t0 = block[04+i] + block[24+i];
@@ -113,7 +113,7 @@ static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride) · 25 unmodified lines
113	113	}
114	114	}
115	115
116		−static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
	116	+static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
117	117	{
118	118	int i, dc = (block[0] + 4) >> 3;
119	119	block[0] = 0;
@@ -127,7 +127,7 @@ static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride) · 7 unmodified lines
127	127	}
128	128	}
129	129
130		−static void vp8_idct_dc_add4uv_c(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride)
	130	+static void vp8_idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)
131	131	{
132	132	vp8_idct_dc_add_c(dst+stride*0+0, block[0], stride);
133	133	vp8_idct_dc_add_c(dst+stride*0+4, block[1], stride);
@@ -135,7 +135,7 @@ static void vp8_idct_dc_add4uv_c(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t s · 1 unmodified line
135	135	vp8_idct_dc_add_c(dst+stride*4+4, block[3], stride);
136	136	}
137	137
138		−static void vp8_idct_dc_add4y_c(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride)
	138	+static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)
139	139	{
140	140	vp8_idct_dc_add_c(dst+ 0, block[0], stride);
141	141	vp8_idct_dc_add_c(dst+ 4, block[1], stride);

libavcodec/vp8dsp.h+8−7

@@ -27,20 +27,21 @@ · 26 unmodified lines
27	27	#ifndef AVCODEC_VP8DSP_H
28	28	#define AVCODEC_VP8DSP_H
29	29
30		−#include "dsputil.h"
	30	+#include <stddef.h>
	31	+#include <stdint.h>
31	32
32	33	typedef void (vp8_mc_func)(uint8_t dst/align 8/, ptrdiff_t dstStride,
33	34	uint8_t src/align 1*/, ptrdiff_t srcStride,
34	35	int h, int x, int y);
35	36
36	37	typedef struct VP8DSPContext {
37		− void (*vp8_luma_dc_wht)(DCTELEM block[4][4][16], DCTELEM dc[16]);
38		− void (*vp8_luma_dc_wht_dc)(DCTELEM block[4][4][16], DCTELEM dc[16]);
39		− void (vp8_idct_add)(uint8_t dst, DCTELEM block[16], ptrdiff_t stride);
40		− void (vp8_idct_dc_add)(uint8_t dst, DCTELEM block[16], ptrdiff_t stride);
41		− void (vp8_idct_dc_add4y)(uint8_t dst, DCTELEM block[4][16],
	38	+ void (*vp8_luma_dc_wht)(int16_t block[4][4][16], int16_t dc[16]);
	39	+ void (*vp8_luma_dc_wht_dc)(int16_t block[4][4][16], int16_t dc[16]);
	40	+ void (vp8_idct_add)(uint8_t dst, int16_t block[16], ptrdiff_t stride);
	41	+ void (vp8_idct_dc_add)(uint8_t dst, int16_t block[16], ptrdiff_t stride);
	42	+ void (vp8_idct_dc_add4y)(uint8_t dst, int16_t block[4][16],
42	43	ptrdiff_t stride);
43		− void (vp8_idct_dc_add4uv)(uint8_t dst, DCTELEM block[4][16],
	44	+ void (vp8_idct_dc_add4uv)(uint8_t dst, int16_t block[4][16],
44	45	ptrdiff_t stride);
45	46
46	47	// loop filter applied to edges between macroblocks

libavcodec/wmv2.c+2−2

@@ -48,7 +48,7 @@ av_cold void ff_wmv2_common_init(Wmv2Context * w){ · 47 unmodified lines
48	48	s->dsp.idct = NULL;
49	49	}
50	50
51		−static void wmv2_add_block(Wmv2Context w, DCTELEM block1, uint8_t *dst, int stride, int n){
	51	+static void wmv2_add_block(Wmv2Context w, int16_t block1, uint8_t *dst, int stride, int n){
52	52	MpegEncContext * const s= &w->s;
53	53
54	54	if (s->block_last_index[n] >= 0) {
@@ -72,7 +72,7 @@ static void wmv2_add_block(Wmv2Context w, DCTELEM block1, uint8_t *dst, int st · 17 unmodified lines
72	72	}
73	73	}
74	74
75		−void ff_wmv2_add_mb(MpegEncContext s, DCTELEM block1[6][64], uint8_t dest_y, uint8_t dest_cb, uint8_t dest_cr){
	75	+void ff_wmv2_add_mb(MpegEncContext s, int16_t block1[6][64], uint8_t dest_y, uint8_t dest_cb, uint8_t dest_cr){
76	76	Wmv2Context * const w= (Wmv2Context*)s;
77	77
78	78	wmv2_add_block(w, block1[0], dest_y , s->linesize, 0);

libavcodec/wmv2.h+1−1

@@ -52,7 +52,7 @@ typedef struct Wmv2Context{ · 51 unmodified lines
52	52	int hshift;
53	53
54	54	ScanTable abt_scantable[2];
55		− DECLARE_ALIGNED(16, DCTELEM, abt_block2)[6][64];
	55	+ DECLARE_ALIGNED(16, int16_t, abt_block2)[6][64];
56	56	}Wmv2Context;
57	57
58	58	void ff_wmv2_common_init(Wmv2Context * w);

libavcodec/wmv2dec.c+2−2

@@ -291,7 +291,7 @@ static int16_t wmv2_pred_motion(Wmv2Context w, int px, int py){ · 290 unmodified lines
291	291	return mot_val;
292	292	}
293	293
294		−static inline int wmv2_decode_inter_block(Wmv2Context w, DCTELEM block, int n, int cbp){
	294	+static inline int wmv2_decode_inter_block(Wmv2Context w, int16_t block, int n, int cbp){
295	295	MpegEncContext * const s= &w->s;
296	296	static const int sub_cbp_table[3]= {2,3,1};
297	297	int sub_cbp;
@@ -331,7 +331,7 @@ static inline int wmv2_decode_inter_block(Wmv2Context w, DCTELEM block, int n, · 33 unmodified lines
331	331	}
332	332
333	333
334		−int ff_wmv2_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
	334	+int ff_wmv2_decode_mb(MpegEncContext *s, int16_t block[6][64])
335	335	{
336	336	Wmv2Context * const w= (Wmv2Context*)s;
337	337	int cbp, code, i;

libavcodec/wmv2dsp.c+3−2

@@ -19,6 +19,7 @@ · 18 unmodified lines
19	19	#include "libavutil/attributes.h"
20	20	#include "libavutil/common.h"
21	21	#include "avcodec.h"
	22	+#include "dsputil.h"
22	23	#include "wmv2dsp.h"
23	24
24	25	#define W0 2048
@@ -91,7 +92,7 @@ static void wmv2_idct_col(short * b) · 66 unmodified lines
91	92	b[8 * 7] = (a0 + a2 - a1 - a5 + (1 << 13)) >> 14;
92	93	}
93	94
94		−static void wmv2_idct_add_c(uint8_t dest, int line_size, DCTELEM block)
	95	+static void wmv2_idct_add_c(uint8_t dest, int line_size, int16_t block)
95	96	{
96	97	int i;
97	98
@@ -114,7 +115,7 @@ static void wmv2_idct_add_c(uint8_t dest, int line_size, DCTELEM block) · 16 unmodified lines
114	115	}
115	116	}
116	117
117		−static void wmv2_idct_put_c(uint8_t dest, int line_size, DCTELEM block)
	118	+static void wmv2_idct_put_c(uint8_t dest, int line_size, int16_t block)
118	119	{
119	120	int i;
120	121

libavcodec/wmv2dsp.h+2−3

@@ -20,11 +20,10 @@ · 19 unmodified lines
20	20	#define AVCODEC_WMV2DSP_H
21	21
22	22	#include <stdint.h>
23		−#include "dsputil.h"
24	23
25	24	typedef struct WMV2DSPContext {
26		− void (idct_add)(uint8_t dest, int line_size, DCTELEM *block);
27		− void (idct_put)(uint8_t dest, int line_size, DCTELEM *block);
	25	+ void (idct_add)(uint8_t dest, int line_size, int16_t *block);
	26	+ void (idct_put)(uint8_t dest, int line_size, int16_t *block);
28	27
29	28	int idct_perm;
30	29	} WMV2DSPContext;

libavcodec/wmv2enc.c+1−1

@@ -148,7 +148,7 @@ int ff_wmv2_encode_picture_header(MpegEncContext * s, int picture_number) · 147 unmodified lines
148	148	* useless M$ crap features. It is duplicated here in case someone wants
149	149	* to add support for these crap features. */
150	150	void ff_wmv2_encode_mb(MpegEncContext * s,
151		− DCTELEM block[6][64],
	151	+ int16_t block[6][64],
152	152	int motion_x, int motion_y)
153	153	{
154	154	Wmv2Context * const w= (Wmv2Context*)s;

libavcodec/x86/dnxhdenc.c+1−1

@@ -26,7 +26,7 @@ · 25 unmodified lines
26	26
27	27	#if HAVE_SSE2_INLINE
28	28
29		−static void get_pixels_8x4_sym_sse2(DCTELEM block, const uint8_t pixels, int line_size)
	29	+static void get_pixels_8x4_sym_sse2(int16_t block, const uint8_t pixels, int line_size)
30	30	{
31	31	__asm__ volatile(
32	32	"pxor %%xmm5, %%xmm5 \n\t"

libavcodec/x86/dsputil_mmx.c+8−8

@@ -225,10 +225,10 @@ DECLARE_ALIGNED(16, const double, ff_pd_2)[2] = { 2.0, 2.0 }; · 224 unmodified lines
225	225	/***********************************/
226	226	/* standard MMX */
227	227
228		−void ff_put_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels,
	228	+void ff_put_pixels_clamped_mmx(const int16_t block, uint8_t pixels,
229	229	int line_size)
230	230	{
231		− const DCTELEM *p;
	231	+ const int16_t *p;
232	232	uint8_t *pix;
233	233
234	234	/* read the pixels */
@@ -300,7 +300,7 @@ void ff_put_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels, · 65 unmodified lines
300	300	"movq %%mm3, (%0, %3, 2) \n\t" \
301	301	"movq %%mm4, (%0, %1) \n\t"
302	302
303		−void ff_put_signed_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels,
	303	+void ff_put_signed_pixels_clamped_mmx(const int16_t block, uint8_t pixels,
304	304	int line_size)
305	305	{
306	306	x86_reg line_skip = line_size;
@@ -317,10 +317,10 @@ void ff_put_signed_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels, · 10 unmodified lines
317	317	: "memory");
318	318	}
319	319
320		−void ff_add_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels,
	320	+void ff_add_pixels_clamped_mmx(const int16_t block, uint8_t pixels,
321	321	int line_size)
322	322	{
323		− const DCTELEM *p;
	323	+ const int16_t *p;
324	324	uint8_t *pix;
325	325	int i;
326	326
@@ -422,7 +422,7 @@ static void put_pixels16_mmx(uint8_t block, const uint8_t pixels, · 95 unmodified lines
422	422	}
423	423
424	424	#define CLEAR_BLOCKS(name, n) \
425		−static void name(DCTELEM *blocks) \
	425	+static void name(int16_t *blocks) \
426	426	{ \
427	427	__asm__ volatile ( \
428	428	"pxor %%mm7, %%mm7 \n\t" \
@@ -442,7 +442,7 @@ static void name(DCTELEM *blocks) \ · 13 unmodified lines
442	442	CLEAR_BLOCKS(clear_blocks_mmx, 6)
443	443	CLEAR_BLOCKS(clear_block_mmx, 1)
444	444
445		−static void clear_block_sse(DCTELEM *block)
	445	+static void clear_block_sse(int16_t *block)
446	446	{
447	447	__asm__ volatile (
448	448	"xorps %%xmm0, %%xmm0 \n"
@@ -459,7 +459,7 @@ static void clear_block_sse(DCTELEM *block) · 10 unmodified lines
459	459	);
460	460	}
461	461
462		−static void clear_blocks_sse(DCTELEM *blocks)
	462	+static void clear_blocks_sse(int16_t *blocks)
463	463	{
464	464	__asm__ volatile (
465	465	"xorps %%xmm0, %%xmm0 \n"

libavcodec/x86/dsputil_mmx.h+3−3

@@ -83,9 +83,9 @@ extern const double ff_pd_2[2]; · 82 unmodified lines
83	83	void ff_dsputilenc_init_mmx(DSPContext* c, AVCodecContext *avctx);
84	84	void ff_dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx);
85	85
86		−void ff_add_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels, int line_size);
87		−void ff_put_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels, int line_size);
88		−void ff_put_signed_pixels_clamped_mmx(const DCTELEM block, uint8_t pixels, int line_size);
	86	+void ff_add_pixels_clamped_mmx(const int16_t block, uint8_t pixels, int line_size);
	87	+void ff_put_pixels_clamped_mmx(const int16_t block, uint8_t pixels, int line_size);
	88	+void ff_put_signed_pixels_clamped_mmx(const int16_t block, uint8_t pixels, int line_size);
89	89
90	90	void ff_put_cavs_qpel8_mc00_mmxext(uint8_t dst, uint8_t src, int stride);
91	91	void ff_avg_cavs_qpel8_mc00_mmxext(uint8_t dst, uint8_t src, int stride);

libavcodec/x86/dsputilenc.asm+2−2

@@ -335,7 +335,7 @@ cglobal sse16, 5, 5, 8 · 334 unmodified lines
335	335	RET
336	336
337	337	INIT_MMX mmx
338		−; get_pixels_mmx(DCTELEM block, const uint8_t pixels, int line_size)
	338	+; get_pixels_mmx(int16_t block, const uint8_t pixels, int line_size)
339	339	cglobal get_pixels, 3,4
340	340	movsxdifnidn r2, r2d
341	341	add r0, 128
@@ -392,7 +392,7 @@ cglobal get_pixels, 3, 4 · 50 unmodified lines
392	392	RET
393	393
394	394	INIT_MMX mmx
395		−; diff_pixels_mmx(DCTELEM block, const uint8_t s1, const unint8_t *s2, stride)
	395	+; diff_pixels_mmx(int16_t block, const uint8_t s1, const unint8_t *s2, stride)
396	396	cglobal diff_pixels, 4,5
397	397	movsxdifnidn r3, r3d
398	398	pxor m7, m7

libavcodec/x86/dsputilenc_mmx.c+4−4

@@ -30,9 +30,9 @@ · 29 unmodified lines
30	30	#include "libavcodec/mathops.h"
31	31	#include "dsputil_mmx.h"
32	32
33		−void ff_get_pixels_mmx(DCTELEM block, const uint8_t pixels, int line_size);
34		−void ff_get_pixels_sse2(DCTELEM block, const uint8_t pixels, int line_size);
35		−void ff_diff_pixels_mmx(DCTELEM block, const uint8_t s1, const uint8_t *s2, int stride);
	33	+void ff_get_pixels_mmx(int16_t block, const uint8_t pixels, int line_size);
	34	+void ff_get_pixels_sse2(int16_t block, const uint8_t pixels, int line_size);
	35	+void ff_diff_pixels_mmx(int16_t block, const uint8_t s1, const uint8_t *s2, int stride);
36	36	int ff_pix_sum16_mmx(uint8_t * pix, int line_size);
37	37	int ff_pix_norm1_mmx(uint8_t *pix, int line_size);
38	38
@@ -797,7 +797,7 @@ static void sub_hfyu_median_prediction_mmxext(uint8_t dst, const uint8_t src1, · 758 unmodified lines
797	797	HSUM(%%xmm0, %%xmm1, %0)
798	798
799	799	#define DCT_SAD_FUNC(cpu) \
800		−static int sum_abs_dctelem_##cpu(DCTELEM *block){\
	800	+static int sum_abs_dctelem_##cpu(int16_t *block){\
801	801	int sum;\
802	802	__asm__ volatile(\
803	803	DCT_SAD\

libavcodec/x86/h264_idct.asm+13−13

@@ -302,7 +302,7 @@ cglobal h264_idct8_dc_add_8, 3, 3, 0 · 301 unmodified lines
302	302
303	303	INIT_MMX mmx
304	304	; ff_h264_idct_add16_mmx(uint8_t dst, const int block_offset,
305		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	305	+; int16_t block, int stride, const uint8_t nnzc[68])
306	306	cglobal h264_idct_add16_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, nnzc, cntr, coeff, picreg
307	307	xor r5, r5
308	308	%ifdef PIC
@@ -324,7 +324,7 @@ cglobal h264_idct_add16_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, · 15 unmodified lines
324	324	REP_RET
325	325
326	326	; ff_h264_idct8_add4_mmx(uint8_t dst, const int block_offset,
327		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	327	+; int16_t block, int stride, const uint8_t nnzc[68])
328	328	cglobal h264_idct8_add4_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, nnzc, cntr, coeff, picreg
329	329	%assign pad 128+4-(stack_offset&7)
330	330	SUB rsp, pad
@@ -357,7 +357,7 @@ cglobal h264_idct8_add4_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, · 26 unmodified lines
357	357
358	358	INIT_MMX mmxext
359	359	; ff_h264_idct_add16_mmxext(uint8_t dst, const int block_offset,
360		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	360	+; int16_t block, int stride, const uint8_t nnzc[68])
361	361	cglobal h264_idct_add16_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
362	362	xor r5, r5
363	363	%ifdef PIC
@@ -402,7 +402,7 @@ cglobal h264_idct_add16_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride · 38 unmodified lines
402	402
403	403	INIT_MMX mmx
404	404	; ff_h264_idct_add16intra_mmx(uint8_t dst, const int block_offset,
405		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	405	+; int16_t block, int stride, const uint8_t nnzc[68])
406	406	cglobal h264_idct_add16intra_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, nnzc, cntr, coeff, picreg
407	407	xor r5, r5
408	408	%ifdef PIC
@@ -426,7 +426,7 @@ cglobal h264_idct_add16intra_8, 5, 7 + npicregs, 0, dst, block_offset, block, st · 17 unmodified lines
426	426
427	427	INIT_MMX mmxext
428	428	; ff_h264_idct_add16intra_mmxext(uint8_t dst, const int block_offset,
429		−; DCTELEM *block, int stride,
	429	+; int16_t *block, int stride,
430	430	; const uint8_t nnzc[6*8])
431	431	cglobal h264_idct_add16intra_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
432	432	xor r5, r5
@@ -469,7 +469,7 @@ cglobal h264_idct_add16intra_8, 5, 8 + npicregs, 0, dst1, block_offset, block, s · 36 unmodified lines
469	469	REP_RET
470	470
471	471	; ff_h264_idct8_add4_mmxext(uint8_t dst, const int block_offset,
472		−; DCTELEM *block, int stride,
	472	+; int16_t *block, int stride,
473	473	; const uint8_t nnzc[6*8])
474	474	cglobal h264_idct8_add4_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
475	475	%assign pad 128+4-(stack_offset&7)
@@ -530,7 +530,7 @@ cglobal h264_idct8_add4_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride · 54 unmodified lines
530	530
531	531	INIT_XMM sse2
532	532	; ff_h264_idct8_add4_sse2(uint8_t dst, const int block_offset,
533		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	533	+; int16_t block, int stride, const uint8_t nnzc[68])
534	534	cglobal h264_idct8_add4_8, 5, 8 + npicregs, 10, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
535	535	xor r5, r5
536	536	%ifdef PIC
@@ -605,7 +605,7 @@ h264_idct_add8_mmx_plane: · 68 unmodified lines
605	605	rep ret
606	606
607	607	; ff_h264_idct_add8_mmx(uint8_t *dest, const int block_offset,
608		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	608	+; int16_t block, int stride, const uint8_t nnzc[68])
609	609	cglobal h264_idct_add8_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
610	610	mov r5, 16
611	611	add r2, 512
@@ -669,7 +669,7 @@ h264_idct_add8_mmxext_plane: · 57 unmodified lines
669	669
670	670	INIT_MMX mmxext
671	671	; ff_h264_idct_add8_mmxext(uint8_t *dest, const int block_offset,
672		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	672	+; int16_t block, int stride, const uint8_t nnzc[68])
673	673	cglobal h264_idct_add8_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
674	674	mov r5, 16
675	675	add r2, 512
@@ -746,7 +746,7 @@ h264_add8x4_idct_sse2: · 70 unmodified lines
746	746	%endmacro
747	747
748	748	; ff_h264_idct_add16_sse2(uint8_t dst, const int block_offset,
749		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	749	+; int16_t block, int stride, const uint8_t nnzc[68])
750	750	cglobal h264_idct_add16_8, 5, 5 + ARCH_X86_64, 8
751	751	%if ARCH_X86_64
752	752	mov r5, r0
@@ -793,7 +793,7 @@ cglobal h264_idct_add16_8, 5, 5 + ARCH_X86_64, 8 · 40 unmodified lines
793	793	%endmacro
794	794
795	795	; ff_h264_idct_add16intra_sse2(uint8_t dst, const int block_offset,
796		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	796	+; int16_t block, int stride, const uint8_t nnzc[68])
797	797	cglobal h264_idct_add16intra_8, 5, 7 + ARCH_X86_64, 8
798	798	%if ARCH_X86_64
799	799	mov r7, r0
@@ -844,7 +844,7 @@ cglobal h264_idct_add16intra_8, 5, 7 + ARCH_X86_64, 8 · 44 unmodified lines
844	844	%endmacro
845	845
846	846	; ff_h264_idct_add8_sse2(uint8_t *dest, const int block_offset,
847		−; DCTELEM block, int stride, const uint8_t nnzc[68])
	847	+; int16_t block, int stride, const uint8_t nnzc[68])
848	848	cglobal h264_idct_add8_8, 5, 7 + ARCH_X86_64, 8
849	849	add r2, 512
850	850	%if ARCH_X86_64
@@ -861,7 +861,7 @@ cglobal h264_idct_add8_8, 5, 7 + ARCH_X86_64, 8 · 10 unmodified lines
861	861	add8_sse2_cycle 3, 0x64
862	862	RET
863	863
864		−;void ff_h264_luma_dc_dequant_idct_mmx(DCTELEM output, DCTELEM input, int qmul)
	864	+;void ff_h264_luma_dc_dequant_idct_mmx(int16_t output, int16_t input, int qmul)
865	865
866	866	%macro WALSH4_1D 5
867	867	SUMSUB_BADC w, %4, %3, %2, %1, %5

libavcodec/x86/h264_intrapred_init.c+1−0

@@ -20,6 +20,7 @@ · 19 unmodified lines
20	20
21	21	#include "libavutil/cpu.h"
22	22	#include "libavutil/x86/cpu.h"
	23	+#include "libavcodec/avcodec.h"
23	24	#include "libavcodec/h264pred.h"
24	25
25	26	#define PRED4x4(TYPE, DEPTH, OPT) \

libavcodec/x86/h264dsp_init.c+4−4

@@ -48,7 +48,7 @@ IDCT_ADD_FUNC(8, 10, avx) · 47 unmodified lines
48	48	#define IDCT_ADD_REP_FUNC(NUM, REP, DEPTH, OPT) \
49	49	void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \
50	50	(uint8_t dst, const int block_offset, \
51		− DCTELEM block, int stride, const uint8_t nnzc[6 8]);
	51	+ int16_t block, int stride, const uint8_t nnzc[6 8]);
52	52
53	53	IDCT_ADD_REP_FUNC(8, 4, 8, mmx)
54	54	IDCT_ADD_REP_FUNC(8, 4, 8, mmxext)
@@ -70,7 +70,7 @@ IDCT_ADD_REP_FUNC(, 16intra, 10, avx) · 15 unmodified lines
70	70	#define IDCT_ADD_REP_FUNC2(NUM, REP, DEPTH, OPT) \
71	71	void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \
72	72	(uint8_t *dst, const int block_offset, \
73		− DCTELEM block, int stride, const uint8_t nnzc[6 8]);
	73	+ int16_t block, int stride, const uint8_t nnzc[6 8]);
74	74
75	75	IDCT_ADD_REP_FUNC2(, 8, 8, mmx)
76	76	IDCT_ADD_REP_FUNC2(, 8, 8, mmxext)
@@ -78,8 +78,8 @@ IDCT_ADD_REP_FUNC2(, 8, 8, sse2) · 1 unmodified line
78	78	IDCT_ADD_REP_FUNC2(, 8, 10, sse2)
79	79	IDCT_ADD_REP_FUNC2(, 8, 10, avx)
80	80
81		−void ff_h264_luma_dc_dequant_idct_mmx(DCTELEM output, DCTELEM input, int qmul);
82		−void ff_h264_luma_dc_dequant_idct_sse2(DCTELEM output, DCTELEM input, int qmul);
	81	+void ff_h264_luma_dc_dequant_idct_mmx(int16_t output, int16_t input, int qmul);
	82	+void ff_h264_luma_dc_dequant_idct_sse2(int16_t output, int16_t input, int qmul);
83	83
84	84	/***********************************/
85	85	/* deblocking */

libavcodec/x86/idct_mmx_xvid.c+4−4

@@ -531,25 +531,25 @@ __asm__ volatile( · 530 unmodified lines
531	531	:: "r"(block), "r"(rounder_0), "r"(tab_i_04_xmm), "r"(tg_1_16));
532	532	}
533	533
534		−void ff_idct_xvid_mmx_put(uint8_t dest, int line_size, DCTELEM block)
	534	+void ff_idct_xvid_mmx_put(uint8_t dest, int line_size, int16_t block)
535	535	{
536	536	ff_idct_xvid_mmx(block);
537	537	ff_put_pixels_clamped_mmx(block, dest, line_size);
538	538	}
539	539
540		−void ff_idct_xvid_mmx_add(uint8_t dest, int line_size, DCTELEM block)
	540	+void ff_idct_xvid_mmx_add(uint8_t dest, int line_size, int16_t block)
541	541	{
542	542	ff_idct_xvid_mmx(block);
543	543	ff_add_pixels_clamped_mmx(block, dest, line_size);
544	544	}
545	545
546		−void ff_idct_xvid_mmxext_put(uint8_t dest, int line_size, DCTELEM block)
	546	+void ff_idct_xvid_mmxext_put(uint8_t dest, int line_size, int16_t block)
547	547	{
548	548	ff_idct_xvid_mmxext(block);
549	549	ff_put_pixels_clamped_mmx(block, dest, line_size);
550	550	}
551	551
552		−void ff_idct_xvid_mmxext_add(uint8_t dest, int line_size, DCTELEM block)
	552	+void ff_idct_xvid_mmxext_add(uint8_t dest, int line_size, int16_t block)
553	553	{
554	554	ff_idct_xvid_mmxext(block);
555	555	ff_add_pixels_clamped_mmx(block, dest, line_size);

libavcodec/x86/idct_xvid.h+4−6

@@ -28,15 +28,13 @@ · 27 unmodified lines
28	28
29	29	#include <stdint.h>
30	30
31		−#include "libavcodec/dsputil.h"
32		−
33	31	void ff_idct_xvid_mmx(short *block);
34		−void ff_idct_xvid_mmx_put(uint8_t dest, int line_size, DCTELEM block);
35		−void ff_idct_xvid_mmx_add(uint8_t dest, int line_size, DCTELEM block);
	32	+void ff_idct_xvid_mmx_put(uint8_t dest, int line_size, int16_t block);
	33	+void ff_idct_xvid_mmx_add(uint8_t dest, int line_size, int16_t block);
36	34
37	35	void ff_idct_xvid_mmxext(short *block);
38		−void ff_idct_xvid_mmxext_put(uint8_t dest, int line_size, DCTELEM block);
39		−void ff_idct_xvid_mmxext_add(uint8_t dest, int line_size, DCTELEM block);
	36	+void ff_idct_xvid_mmxext_put(uint8_t dest, int line_size, int16_t block);
	37	+void ff_idct_xvid_mmxext_add(uint8_t dest, int line_size, int16_t block);
40	38
41	39	void ff_idct_xvid_sse2(short *block);
42	40	void ff_idct_xvid_sse2_put(uint8_t dest, int line_size, short block);

libavcodec/x86/mpegvideo.c+8−8

@@ -29,7 +29,7 @@ · 28 unmodified lines
29	29	#if HAVE_INLINE_ASM
30	30
31	31	static void dct_unquantize_h263_intra_mmx(MpegEncContext *s,
32		− DCTELEM *block, int n, int qscale)
	32	+ int16_t *block, int n, int qscale)
33	33	{
34	34	x86_reg level, qmul, qadd, nCoeffs;
35	35
@@ -104,7 +104,7 @@ __asm__ volatile( · 68 unmodified lines
104	104
105	105
106	106	static void dct_unquantize_h263_inter_mmx(MpegEncContext *s,
107		− DCTELEM *block, int n, int qscale)
	107	+ int16_t *block, int n, int qscale)
108	108	{
109	109	x86_reg qmul, qadd, nCoeffs;
110	110
@@ -195,7 +195,7 @@ __asm__ volatile( · 84 unmodified lines
195	195	high3 += tlow1
196	196	*/
197	197	static void dct_unquantize_mpeg1_intra_mmx(MpegEncContext *s,
198		− DCTELEM *block, int n, int qscale)
	198	+ int16_t *block, int n, int qscale)
199	199	{
200	200	x86_reg nCoeffs;
201	201	const uint16_t *quant_matrix;
@@ -264,7 +264,7 @@ __asm__ volatile( · 62 unmodified lines
264	264	}
265	265
266	266	static void dct_unquantize_mpeg1_inter_mmx(MpegEncContext *s,
267		− DCTELEM *block, int n, int qscale)
	267	+ int16_t *block, int n, int qscale)
268	268	{
269	269	x86_reg nCoeffs;
270	270	const uint16_t *quant_matrix;
@@ -330,7 +330,7 @@ __asm__ volatile( · 59 unmodified lines
330	330	}
331	331
332	332	static void dct_unquantize_mpeg2_intra_mmx(MpegEncContext *s,
333		− DCTELEM *block, int n, int qscale)
	333	+ int16_t *block, int n, int qscale)
334	334	{
335	335	x86_reg nCoeffs;
336	336	const uint16_t *quant_matrix;
@@ -396,7 +396,7 @@ __asm__ volatile( · 59 unmodified lines
396	396	}
397	397
398	398	static void dct_unquantize_mpeg2_inter_mmx(MpegEncContext *s,
399		− DCTELEM *block, int n, int qscale)
	399	+ int16_t *block, int n, int qscale)
400	400	{
401	401	x86_reg nCoeffs;
402	402	const uint16_t *quant_matrix;
@@ -472,7 +472,7 @@ __asm__ volatile( · 69 unmodified lines
472	472	);
473	473	}
474	474
475		−static void denoise_dct_mmx(MpegEncContext s, DCTELEM block){
	475	+static void denoise_dct_mmx(MpegEncContext s, int16_t block){
476	476	const int intra= s->mb_intra;
477	477	int *sum= s->dct_error_sum[intra];
478	478	uint16_t *offset= s->dct_offset[intra];
@@ -526,7 +526,7 @@ static void denoise_dct_mmx(MpegEncContext s, DCTELEM block){ · 47 unmodified lines
526	526	);
527	527	}
528	528
529		−static void denoise_dct_sse2(MpegEncContext s, DCTELEM block){
	529	+static void denoise_dct_sse2(MpegEncContext s, int16_t block){
530	530	const int intra= s->mb_intra;
531	531	int *sum= s->dct_error_sum[intra];
532	532	uint16_t *offset= s->dct_offset[intra];

libavcodec/x86/mpegvideoenc_template.c+1−1

@@ -92,7 +92,7 @@ · 91 unmodified lines
92	92	#endif
93	93
94	94	static int RENAME(dct_quantize)(MpegEncContext *s,
95		− DCTELEM *block, int n,
	95	+ int16_t *block, int n,
96	96	int qscale, int *overflow)
97	97	{
98	98	x86_reg last_non_zero_p1;

libavcodec/x86/proresdsp.asm+1−1

@@ -327,7 +327,7 @@ section .text align=16 · 326 unmodified lines
327	327	%endmacro
328	328
329	329	; void prores_idct_put_10_<opt>(uint8_t *pixels, int stride,
330		−; DCTELEM block, const int16_t qmat);
	330	+; int16_t block, const int16_t qmat);
331	331	%macro idct_put_fn 1
332	332	cglobal prores_idct_put_10, 4, 4, %1
333	333	movsxd r1, r1d

libavcodec/x86/proresdsp_init.c+4−3

@@ -21,14 +21,15 @@ · 20 unmodified lines
21	21	*/
22	22
23	23	#include "libavutil/x86/cpu.h"
	24	+#include "libavcodec/dsputil.h"
24	25	#include "libavcodec/proresdsp.h"
25	26
26	27	void ff_prores_idct_put_10_sse2(uint16_t *dst, int linesize,
27		− DCTELEM block, const int16_t qmat);
	28	+ int16_t block, const int16_t qmat);
28	29	void ff_prores_idct_put_10_sse4(uint16_t *dst, int linesize,
29		− DCTELEM block, const int16_t qmat);
	30	+ int16_t block, const int16_t qmat);
30	31	void ff_prores_idct_put_10_avx (uint16_t *dst, int linesize,
31		− DCTELEM block, const int16_t qmat);
	32	+ int16_t block, const int16_t qmat);
32	33
33	34	void ff_proresdsp_x86_init(ProresDSPContext *dsp)
34	35	{

libavcodec/x86/rv34dsp.asm+1−1

@@ -133,7 +133,7 @@ cglobal rv34_idct_dc_add, 3, 3 · 132 unmodified lines
133	133	mova mm5, [pd_512] ; 0x200
134	134	%endmacro
135	135
136		−; ff_rv34_idct_add_mmxext(uint8_t dst, ptrdiff_t stride, DCTELEM block);
	136	+; ff_rv34_idct_add_mmxext(uint8_t dst, ptrdiff_t stride, int16_t block);
137	137	%macro COL_TRANSFORM 4
138	138	pshufw mm3, %2, 0xDD ; col. 1,3,1,3
139	139	pshufw %2, %2, 0x88 ; col. 0,2,0,2

libavcodec/x86/rv34dsp_init.c+3−3

@@ -25,11 +25,11 @@ · 24 unmodified lines
25	25	#include "libavcodec/dsputil.h"
26	26	#include "libavcodec/rv34dsp.h"
27	27
28		−void ff_rv34_idct_dc_mmxext(DCTELEM *block);
29		−void ff_rv34_idct_dc_noround_mmxext(DCTELEM *block);
	28	+void ff_rv34_idct_dc_mmxext(int16_t *block);
	29	+void ff_rv34_idct_dc_noround_mmxext(int16_t *block);
30	30	void ff_rv34_idct_dc_add_mmx(uint8_t *dst, ptrdiff_t stride, int dc);
31	31	void ff_rv34_idct_dc_add_sse4(uint8_t *dst, ptrdiff_t stride, int dc);
32		−void ff_rv34_idct_add_mmxext(uint8_t dst, ptrdiff_t stride, DCTELEM block);
	32	+void ff_rv34_idct_add_mmxext(uint8_t dst, ptrdiff_t stride, int16_t block);
33	33
34	34	av_cold void ff_rv34dsp_init_x86(RV34DSPContext* c, DSPContext *dsp)
35	35	{

libavcodec/x86/simple_idct.c+2−2

@@ -1155,12 +1155,12 @@ void ff_simple_idct_mmx(int16_t *block) · 1154 unmodified lines
1155	1155
1156	1156	//FIXME merge add/put into the idct
1157	1157
1158		−void ff_simple_idct_put_mmx(uint8_t dest, int line_size, DCTELEM block)
	1158	+void ff_simple_idct_put_mmx(uint8_t dest, int line_size, int16_t block)
1159	1159	{
1160	1160	idct(block);
1161	1161	ff_put_pixels_clamped_mmx(block, dest, line_size);
1162	1162	}
1163		−void ff_simple_idct_add_mmx(uint8_t dest, int line_size, DCTELEM block)
	1163	+void ff_simple_idct_add_mmx(uint8_t dest, int line_size, int16_t block)
1164	1164	{
1165	1165	idct(block);
1166	1166	ff_add_pixels_clamped_mmx(block, dest, line_size);

libavcodec/x86/vc1dsp_mmx.c+4−4

@@ -494,7 +494,7 @@ DECLARE_FUNCTION(3, 2) · 493 unmodified lines
494	494	DECLARE_FUNCTION(3, 3)
495	495
496	496	static void vc1_inv_trans_4x4_dc_mmxext(uint8_t *dest, int linesize,
497		− DCTELEM *block)
	497	+ int16_t *block)
498	498	{
499	499	int dc = block[0];
500	500	dc = (17 * dc + 4) >> 3;
@@ -533,7 +533,7 @@ static void vc1_inv_trans_4x4_dc_mmxext(uint8_t *dest, int linesize, · 32 unmodified lines
533	533	}
534	534
535	535	static void vc1_inv_trans_4x8_dc_mmxext(uint8_t *dest, int linesize,
536		− DCTELEM *block)
	536	+ int16_t *block)
537	537	{
538	538	int dc = block[0];
539	539	dc = (17 * dc + 4) >> 3;
@@ -595,7 +595,7 @@ static void vc1_inv_trans_4x8_dc_mmxext(uint8_t *dest, int linesize, · 55 unmodified lines
595	595	}
596	596
597	597	static void vc1_inv_trans_8x4_dc_mmxext(uint8_t *dest, int linesize,
598		− DCTELEM *block)
	598	+ int16_t *block)
599	599	{
600	600	int dc = block[0];
601	601	dc = ( 3 * dc + 1) >> 1;
@@ -634,7 +634,7 @@ static void vc1_inv_trans_8x4_dc_mmxext(uint8_t *dest, int linesize, · 32 unmodified lines
634	634	}
635	635
636	636	static void vc1_inv_trans_8x8_dc_mmxext(uint8_t *dest, int linesize,
637		− DCTELEM *block)
	637	+ int16_t *block)
638	638	{
639	639	int dc = block[0];
640	640	dc = (3 * dc + 1) >> 1;

libavcodec/x86/vp3dsp_init.c+6−5

@@ -22,17 +22,18 @@ · 21 unmodified lines
22	22	#include "libavutil/cpu.h"
23	23	#include "libavutil/x86/cpu.h"
24	24	#include "libavcodec/avcodec.h"
	25	+#include "libavcodec/dsputil.h"
25	26	#include "libavcodec/vp3dsp.h"
26	27	#include "config.h"
27	28
28		−void ff_vp3_idct_put_mmx(uint8_t dest, int line_size, DCTELEM block);
29		−void ff_vp3_idct_add_mmx(uint8_t dest, int line_size, DCTELEM block);
	29	+void ff_vp3_idct_put_mmx(uint8_t dest, int line_size, int16_t block);
	30	+void ff_vp3_idct_add_mmx(uint8_t dest, int line_size, int16_t block);
30	31
31		−void ff_vp3_idct_put_sse2(uint8_t dest, int line_size, DCTELEM block);
32		−void ff_vp3_idct_add_sse2(uint8_t dest, int line_size, DCTELEM block);
	32	+void ff_vp3_idct_put_sse2(uint8_t dest, int line_size, int16_t block);
	33	+void ff_vp3_idct_add_sse2(uint8_t dest, int line_size, int16_t block);
33	34
34	35	void ff_vp3_idct_dc_add_mmxext(uint8_t *dest, int line_size,
35		− DCTELEM *block);
	36	+ int16_t *block);
36	37
37	38	void ff_vp3_v_loop_filter_mmxext(uint8_t *src, int stride,
38	39	int *bounding_values);

libavcodec/x86/vp8dsp.asm+5−5

@@ -899,7 +899,7 @@ cglobal put_vp8_pixels16, 5, 5, 2, dst, dststride, src, srcstride, height · 898 unmodified lines
899	899	REP_RET
900	900
901	901	;-----------------------------------------------------------------------------
902		−; void vp8_idct_dc_add_<opt>(uint8_t *dst, DCTELEM block[16], int stride);
	902	+; void vp8_idct_dc_add_<opt>(uint8_t *dst, int16_t block[16], int stride);
903	903	;-----------------------------------------------------------------------------
904	904
905	905	%macro ADD_DC 4
@@ -977,7 +977,7 @@ cglobal vp8_idct_dc_add, 3, 3, 6, dst, block, stride · 71 unmodified lines
977	977	RET
978	978
979	979	;-----------------------------------------------------------------------------
980		−; void vp8_idct_dc_add4y_<opt>(uint8_t *dst, DCTELEM block[4][16], int stride);
	980	+; void vp8_idct_dc_add4y_<opt>(uint8_t *dst, int16_t block[4][16], int stride);
981	981	;-----------------------------------------------------------------------------
982	982
983	983	%if ARCH_X86_32
@@ -1050,7 +1050,7 @@ cglobal vp8_idct_dc_add4y, 3, 3, 6, dst, block, stride · 66 unmodified lines
1050	1050	RET
1051	1051
1052	1052	;-----------------------------------------------------------------------------
1053		−; void vp8_idct_dc_add4uv_<opt>(uint8_t *dst, DCTELEM block[4][16], int stride);
	1053	+; void vp8_idct_dc_add4uv_<opt>(uint8_t *dst, int16_t block[4][16], int stride);
1054	1054	;-----------------------------------------------------------------------------
1055	1055
1056	1056	INIT_MMX mmx
@@ -1092,7 +1092,7 @@ cglobal vp8_idct_dc_add4uv, 3, 3, 0, dst, block, stride · 35 unmodified lines
1092	1092	RET
1093	1093
1094	1094	;-----------------------------------------------------------------------------
1095		−; void vp8_idct_add_<opt>(uint8_t *dst, DCTELEM block[16], int stride);
	1095	+; void vp8_idct_add_<opt>(uint8_t *dst, int16_t block[16], int stride);
1096	1096	;-----------------------------------------------------------------------------
1097	1097
1098	1098	; calculate %1=mul_35468(%1)-mul_20091(%2); %2=mul_20091(%1)+mul_35468(%2)
@@ -1172,7 +1172,7 @@ INIT_MMX sse · 73 unmodified lines
1172	1172	VP8_IDCT_ADD
1173	1173
1174	1174	;-----------------------------------------------------------------------------
1175		−; void vp8_luma_dc_wht_mmxext(DCTELEM block[4][4][16], DCTELEM dc[16])
	1175	+; void vp8_luma_dc_wht_mmxext(int16_t block[4][4][16], int16_t dc[16])
1176	1176	;-----------------------------------------------------------------------------
1177	1177
1178	1178	%macro SCATTER_WHT 3

libavcodec/x86/vp8dsp_init.c+9−9

@@ -230,21 +230,21 @@ HVBILIN(ssse3, 8, 4, 8) · 229 unmodified lines
230	230	HVBILIN(ssse3, 8, 8, 16)
231	231	HVBILIN(ssse3, 8, 16, 16)
232	232
233		−extern void ff_vp8_idct_dc_add_mmx(uint8_t *dst, DCTELEM block[16],
	233	+extern void ff_vp8_idct_dc_add_mmx(uint8_t *dst, int16_t block[16],
234	234	ptrdiff_t stride);
235		−extern void ff_vp8_idct_dc_add_sse4(uint8_t *dst, DCTELEM block[16],
	235	+extern void ff_vp8_idct_dc_add_sse4(uint8_t *dst, int16_t block[16],
236	236	ptrdiff_t stride);
237		−extern void ff_vp8_idct_dc_add4y_mmx(uint8_t *dst, DCTELEM block[4][16],
	237	+extern void ff_vp8_idct_dc_add4y_mmx(uint8_t *dst, int16_t block[4][16],
238	238	ptrdiff_t stride);
239		−extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, DCTELEM block[4][16],
	239	+extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, int16_t block[4][16],
240	240	ptrdiff_t stride);
241		−extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, DCTELEM block[2][16],
	241	+extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, int16_t block[2][16],
242	242	ptrdiff_t stride);
243		−extern void ff_vp8_luma_dc_wht_mmx(DCTELEM block[4][4][16], DCTELEM dc[16]);
244		−extern void ff_vp8_luma_dc_wht_sse(DCTELEM block[4][4][16], DCTELEM dc[16]);
245		−extern void ff_vp8_idct_add_mmx(uint8_t *dst, DCTELEM block[16],
	243	+extern void ff_vp8_luma_dc_wht_mmx(int16_t block[4][4][16], int16_t dc[16]);
	244	+extern void ff_vp8_luma_dc_wht_sse(int16_t block[4][4][16], int16_t dc[16]);
	245	+extern void ff_vp8_idct_add_mmx(uint8_t *dst, int16_t block[16],
246	246	ptrdiff_t stride);
247		−extern void ff_vp8_idct_add_sse(uint8_t *dst, DCTELEM block[16],
	247	+extern void ff_vp8_idct_add_sse(uint8_t *dst, int16_t block[16],
248	248	ptrdiff_t stride);
249	249
250	250	#define DECLARE_LOOP_FILTER(NAME)\

tools/patcheck+1−1

@@ -173,7 +173,7 @@ vertical align = · 172 unmodified lines
173	173	/* and * align
174	174	arrays fitting in smaller types
175	175	variables written to twice with no interspaced read
176		−memset(block, 0, 664sizeof(DCTELEM)); -> clear_blocks
	176	+memset(block, 0, 664sizeof(int16_t)); -> clear_blocks
177	177	check existence of long_name in AVCodec
178	178	check that the patch does not touch codec & (de)muxer layer at the same time ->split
179	179

Drop DCTELEM typedef

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code