Merge remote-tracking branch 'qatar/master'

* qatar/master:
  sgidec: Use bytestream2 functions to prevent buffer overreads.
  cosmetics: Move static and inline attributes to more standard places.
  configure: provide libavfilter/version.h header to get_version()
  swscale: change yuv2yuvX code to use cpuflag().
  libx264: Don't leave max_b_frames as -1 if the user didn't set it
  FATE: convert output to rgba for the targa tests which currently output pal8
  fate: add missing reference files for targa tests in 9c2f9b0e2
  FATE: enable the 2 remaining targa conformance suite tests
  targa: add support for rgb555 palette
  FATE: fix targa tests on big-endian systems

Conflicts:
	libavcodec/sgidec.c
	libavcodec/targa.c
	libswscale/x86/output.asm
	tests/fate/image.mak

Merged-by: Michael Niedermayer <michaelni@gmx.at>
release/0.10
Michael Niedermayer 13 years ago
commit 4640da7e58
  1. 36
      libavcodec/bytestream.h
  2. 2
      libavcodec/g722.c
  3. 51
      libavcodec/h264_loopfilter.c
  4. 3
      libavcodec/libx264.c
  5. 123
      libavcodec/sgidec.c
  6. 33
      libavcodec/targa.c
  7. 2
      libavfilter/vf_fade.c
  8. 2
      libswscale/utils.c
  9. 117
      libswscale/x86/output.asm
  10. 6
      tests/fate/image.mak
  11. 1
      tests/ref/fate/targa-conformance-CCM8
  12. 1
      tests/ref/fate/targa-conformance-UCM8

@ -75,6 +75,42 @@ DEF (byte, 1, AV_RB8 , AV_WB8 )
#undef DEF64
#undef DEF_T
#if HAVE_BIGENDIAN
# define bytestream2_get_ne16 bytestream2_get_be16
# define bytestream2_get_ne24 bytestream2_get_be24
# define bytestream2_get_ne32 bytestream2_get_be32
# define bytestream2_get_ne64 bytestream2_get_be64
# define bytestream2_get_ne16u bytestream2_get_be16u
# define bytestream2_get_ne24u bytestream2_get_be24u
# define bytestream2_get_ne32u bytestream2_get_be32u
# define bytestream2_get_ne64u bytestream2_get_be64u
# define bytestream2_put_ne16 bytestream2_put_be16
# define bytestream2_put_ne24 bytestream2_put_be24
# define bytestream2_put_ne32 bytestream2_put_be32
# define bytestream2_put_ne64 bytestream2_put_be64
# define bytestream2_peek_ne16 bytestream2_peek_be16
# define bytestream2_peek_ne24 bytestream2_peek_be24
# define bytestream2_peek_ne32 bytestream2_peek_be32
# define bytestream2_peek_ne64 bytestream2_peek_be64
#else
# define bytestream2_get_ne16 bytestream2_get_le16
# define bytestream2_get_ne24 bytestream2_get_le24
# define bytestream2_get_ne32 bytestream2_get_le32
# define bytestream2_get_ne64 bytestream2_get_le64
# define bytestream2_get_ne16u bytestream2_get_le16u
# define bytestream2_get_ne24u bytestream2_get_le24u
# define bytestream2_get_ne32u bytestream2_get_le32u
# define bytestream2_get_ne64u bytestream2_get_le64u
# define bytestream2_put_ne16 bytestream2_put_le16
# define bytestream2_put_ne24 bytestream2_put_le24
# define bytestream2_put_ne32 bytestream2_put_le32
# define bytestream2_put_ne64 bytestream2_put_le64
# define bytestream2_peek_ne16 bytestream2_peek_le16
# define bytestream2_peek_ne24 bytestream2_peek_le24
# define bytestream2_peek_ne32 bytestream2_peek_le32
# define bytestream2_peek_ne64 bytestream2_peek_le64
#endif
static av_always_inline void bytestream2_init(GetByteContext *g,
const uint8_t *buf, int buf_size)
{

@ -129,7 +129,7 @@ static void do_adaptive_prediction(struct G722Band *band, const int cur_diff)
band->prev_qtzd_reconst = cur_qtzd_reconst;
}
static int inline linear_scale_factor(const int log_factor)
static inline int linear_scale_factor(const int log_factor)
{
const int wd1 = inv_log2_table[(log_factor >> 6) & 31];
const int shift = log_factor >> 11;

@ -101,7 +101,11 @@ static const uint8_t tc0_table[52*3][4] = {
};
/* intra: 0 if this loopfilter call is guaranteed to be inter (bS < 4), 1 if it might be intra (bS == 4) */
static void av_always_inline filter_mb_edgev( uint8_t *pix, int stride, const int16_t bS[4], unsigned int qp, int a, int b, H264Context *h, int intra ) {
static av_always_inline void filter_mb_edgev(uint8_t *pix, int stride,
const int16_t bS[4],
unsigned int qp, int a, int b,
H264Context *h, int intra)
{
const unsigned int index_a = qp + a;
const int alpha = alpha_table[index_a];
const int beta = beta_table[qp + b];
@ -118,7 +122,12 @@ static void av_always_inline filter_mb_edgev( uint8_t *pix, int stride, const in
h->h264dsp.h264_h_loop_filter_luma_intra(pix, stride, alpha, beta);
}
}
static void av_always_inline filter_mb_edgecv( uint8_t *pix, int stride, const int16_t bS[4], unsigned int qp, int a, int b, H264Context *h, int intra ) {
static av_always_inline void filter_mb_edgecv(uint8_t *pix, int stride,
const int16_t bS[4],
unsigned int qp, int a, int b,
H264Context *h, int intra)
{
const unsigned int index_a = qp + a;
const int alpha = alpha_table[index_a];
const int beta = beta_table[qp + b];
@ -136,7 +145,12 @@ static void av_always_inline filter_mb_edgecv( uint8_t *pix, int stride, const i
}
}
static void av_always_inline filter_mb_mbaff_edgev( H264Context *h, uint8_t *pix, int stride, const int16_t bS[7], int bsi, int qp, int a, int b, int intra ) {
static av_always_inline void filter_mb_mbaff_edgev(H264Context *h, uint8_t *pix,
int stride,
const int16_t bS[7], int bsi,
int qp, int a, int b,
int intra)
{
const unsigned int index_a = qp + a;
const int alpha = alpha_table[index_a];
const int beta = beta_table[qp + b];
@ -153,7 +167,13 @@ static void av_always_inline filter_mb_mbaff_edgev( H264Context *h, uint8_t *pix
h->h264dsp.h264_h_loop_filter_luma_mbaff_intra(pix, stride, alpha, beta);
}
}
static void av_always_inline filter_mb_mbaff_edgecv( H264Context *h, uint8_t *pix, int stride, const int16_t bS[7], int bsi, int qp, int a, int b, int intra ) {
static av_always_inline void filter_mb_mbaff_edgecv(H264Context *h,
uint8_t *pix, int stride,
const int16_t bS[7],
int bsi, int qp, int a,
int b, int intra)
{
const unsigned int index_a = qp + a;
const int alpha = alpha_table[index_a];
const int beta = beta_table[qp + b];
@ -171,7 +191,11 @@ static void av_always_inline filter_mb_mbaff_edgecv( H264Context *h, uint8_t *pi
}
}
static void av_always_inline filter_mb_edgeh( uint8_t *pix, int stride, const int16_t bS[4], unsigned int qp, int a, int b, H264Context *h, int intra ) {
static av_always_inline void filter_mb_edgeh(uint8_t *pix, int stride,
const int16_t bS[4],
unsigned int qp, int a, int b,
H264Context *h, int intra)
{
const unsigned int index_a = qp + a;
const int alpha = alpha_table[index_a];
const int beta = beta_table[qp + b];
@ -189,7 +213,11 @@ static void av_always_inline filter_mb_edgeh( uint8_t *pix, int stride, const in
}
}
static void av_always_inline filter_mb_edgech( uint8_t *pix, int stride, const int16_t bS[4], unsigned int qp, int a, int b, H264Context *h, int intra ) {
static av_always_inline void filter_mb_edgech(uint8_t *pix, int stride,
const int16_t bS[4],
unsigned int qp, int a, int b,
H264Context *h, int intra)
{
const unsigned int index_a = qp + a;
const int alpha = alpha_table[index_a];
const int beta = beta_table[qp + b];
@ -207,8 +235,15 @@ static void av_always_inline filter_mb_edgech( uint8_t *pix, int stride, const i
}
}
static void av_always_inline h264_filter_mb_fast_internal( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr,
unsigned int linesize, unsigned int uvlinesize, int pixel_shift) {
static av_always_inline void h264_filter_mb_fast_internal(H264Context *h,
int mb_x, int mb_y,
uint8_t *img_y,
uint8_t *img_cb,
uint8_t *img_cr,
unsigned int linesize,
unsigned int uvlinesize,
int pixel_shift)
{
MpegEncContext * const s = &h->s;
int chroma = !(CONFIG_GRAY && (s->flags&CODEC_FLAG_GRAY));
int chroma444 = CHROMA444;

@ -532,6 +532,9 @@ static av_cold int X264_init(AVCodecContext *avctx)
// update AVCodecContext with x264 parameters
avctx->has_b_frames = x4->params.i_bframe ?
x4->params.i_bframe_pyramid ? 2 : 1 : 0;
if (avctx->max_b_frames < 0)
avctx->max_b_frames = 0;
avctx->bit_rate = x4->params.rc.i_bitrate*1000;
#if FF_API_X264_GLOBAL_OPTS
avctx->crf = x4->params.rc.f_rf_constant;

@ -32,26 +32,27 @@ typedef struct SgiState {
unsigned int depth;
unsigned int bytes_per_channel;
int linesize;
GetByteContext g;
} SgiState;
/**
* Expand an RLE row into a channel.
* @param in_buf input buffer
* @param in_end end of input buffer
* @param s the current image state
* @param out_buf Points to one line after the output buffer.
* @param out_end end of line in output buffer
* @param pixelstride pixel stride of input buffer
* @return size of output in bytes, -1 if buffer overflows
*/
static int expand_rle_row(const uint8_t *in_buf, const uint8_t* in_end,
unsigned char *out_buf, uint8_t* out_end, int pixelstride)
static int expand_rle_row(SgiState *s, uint8_t *out_buf,
uint8_t *out_end, int pixelstride)
{
unsigned char pixel, count;
unsigned char *orig = out_buf;
while (1) {
if(in_buf + 1 > in_end) return -1;
pixel = bytestream_get_byte(&in_buf);
if (bytestream2_get_bytes_left(&s->g) < 1)
return AVERROR_INVALIDDATA;
pixel = bytestream2_get_byteu(&s->g);
if (!(count = (pixel & 0x7f))) {
return (out_buf - orig) / pixelstride;
}
@ -61,11 +62,11 @@ static int expand_rle_row(const uint8_t *in_buf, const uint8_t* in_end,
if (pixel & 0x80) {
while (count--) {
*out_buf = bytestream_get_byte(&in_buf);
*out_buf = bytestream2_get_byte(&s->g);
out_buf += pixelstride;
}
} else {
pixel = bytestream_get_byte(&in_buf);
pixel = bytestream2_get_byte(&s->g);
while (count--) {
*out_buf = pixel;
@ -78,85 +79,73 @@ static int expand_rle_row(const uint8_t *in_buf, const uint8_t* in_end,
/**
* Read a run length encoded SGI image.
* @param out_buf output buffer
* @param in_buf input buffer
* @param in_end end of input buffer
* @param s the current image state
* @return 0 if no error, else return error number.
*/
static int read_rle_sgi(unsigned char* out_buf, const uint8_t *in_buf,
const uint8_t *in_end, SgiState* s)
static int read_rle_sgi(uint8_t *out_buf, SgiState *s)
{
uint8_t *dest_row;
unsigned int len = s->height * s->depth * 4;
const uint8_t *start_table = in_buf;
GetByteContext g_table = s->g;
unsigned int y, z;
unsigned int start_offset;
/* size of RLE offset and length tables */
if(len * 2 > in_end - in_buf) {
if (len * 2 > bytestream2_get_bytes_left(&s->g)) {
return AVERROR_INVALIDDATA;
}
in_buf -= SGI_HEADER_SIZE;
for (z = 0; z < s->depth; z++) {
dest_row = out_buf;
for (y = 0; y < s->height; y++) {
dest_row -= s->linesize;
start_offset = bytestream_get_be32(&start_table);
if(start_offset > in_end - in_buf) {
start_offset = bytestream2_get_be32(&g_table);
bytestream2_seek(&s->g, start_offset, SEEK_SET);
if (expand_rle_row(s, dest_row + z, dest_row + FFABS(s->linesize),
s->depth) != s->width) {
return AVERROR_INVALIDDATA;
}
if (expand_rle_row(in_buf + start_offset, in_end, dest_row + z,
dest_row + FFABS(s->linesize), s->depth) != s->width)
return AVERROR_INVALIDDATA;
}
}
return 0;
}
static av_always_inline void copy_loop(uint8_t *out_buf, const uint8_t *in_buf,
unsigned offset, unsigned bytes_per_channel,
SgiState *s)
{
int x, y, z;
for (y = s->height - 1; y >= 0; y--) {
uint8_t *line = out_buf + (y * s->linesize);
for (x = s->width; x > 0; x--) {
const uint8_t *ptr = in_buf;
in_buf += bytes_per_channel;
for(z = 0; z < s->depth; z ++) {
memcpy(line, ptr, bytes_per_channel);
line += bytes_per_channel;
ptr += offset;
}
}
}
}
/**
* Read an uncompressed SGI image.
* @param out_buf output buffer
* @param out_end end ofoutput buffer
* @param in_buf input buffer
* @param in_end end of input buffer
* @param s the current image state
* @return 0 if read success, otherwise return -1.
*/
static int read_uncompressed_sgi(unsigned char* out_buf, uint8_t* out_end,
const uint8_t *in_buf, const uint8_t *in_end, SgiState* s)
SgiState *s)
{
int x, y, z;
unsigned int offset = s->height * s->width * s->bytes_per_channel;
GetByteContext gp[4];
/* Test buffer size. */
if (offset * s->depth > in_end - in_buf) {
return -1;
if (offset * s->depth > bytestream2_get_bytes_left(&s->g))
return AVERROR_INVALIDDATA;
/* Create a reader for each plane */
for (z = 0; z < s->depth; z++) {
gp[z] = s->g;
bytestream2_skip(&gp[z], z * offset);
}
if (s->bytes_per_channel == 2) {
copy_loop(out_buf, in_buf, offset, 2, s);
} else {
av_assert1(s->bytes_per_channel == 1);
copy_loop(out_buf, in_buf, offset, 1, s);
for (y = s->height - 1; y >= 0; y--) {
out_end = out_buf + (y * s->linesize);
if (s->bytes_per_channel == 1) {
for (x = s->width; x > 0; x--)
for (z = 0; z < s->depth; z++)
*out_end++ = bytestream2_get_byteu(&gp[z]);
} else {
uint16_t *out16 = (uint16_t *)out_end;
for (x = s->width; x > 0; x--)
for (z = 0; z < s->depth; z++)
*out16++ = bytestream2_get_ne16u(&gp[z]);
}
}
return 0;
}
@ -165,33 +154,31 @@ static int decode_frame(AVCodecContext *avctx,
void *data, int *data_size,
AVPacket *avpkt)
{
const uint8_t *in_buf = avpkt->data;
int buf_size = avpkt->size;
SgiState *s = avctx->priv_data;
AVFrame *picture = data;
AVFrame *p = &s->picture;
const uint8_t *in_end = in_buf + buf_size;
unsigned int dimension, rle;
int ret = 0;
uint8_t *out_buf, *out_end;
if (buf_size < SGI_HEADER_SIZE){
av_log(avctx, AV_LOG_ERROR, "buf_size too small (%d)\n", buf_size);
return -1;
bytestream2_init(&s->g, avpkt->data, avpkt->size);
if (bytestream2_get_bytes_left(&s->g) < SGI_HEADER_SIZE) {
av_log(avctx, AV_LOG_ERROR, "buf_size too small (%d)\n", avpkt->size);
return AVERROR_INVALIDDATA;
}
/* Test for SGI magic. */
if (bytestream_get_be16(&in_buf) != SGI_MAGIC) {
if (bytestream2_get_be16(&s->g) != SGI_MAGIC) {
av_log(avctx, AV_LOG_ERROR, "bad magic number\n");
return -1;
return AVERROR_INVALIDDATA;
}
rle = bytestream_get_byte(&in_buf);
s->bytes_per_channel = bytestream_get_byte(&in_buf);
dimension = bytestream_get_be16(&in_buf);
s->width = bytestream_get_be16(&in_buf);
s->height = bytestream_get_be16(&in_buf);
s->depth = bytestream_get_be16(&in_buf);
rle = bytestream2_get_byte(&s->g);
s->bytes_per_channel = bytestream2_get_byte(&s->g);
dimension = bytestream2_get_be16(&s->g);
s->width = bytestream2_get_be16(&s->g);
s->height = bytestream2_get_be16(&s->g);
s->depth = bytestream2_get_be16(&s->g);
if (s->bytes_per_channel != 1 && (s->bytes_per_channel != 2 || rle)) {
av_log(avctx, AV_LOG_ERROR, "wrong channel number\n");
@ -237,19 +224,19 @@ static int decode_frame(AVCodecContext *avctx,
s->linesize = p->linesize[0];
/* Skip header. */
in_buf += SGI_HEADER_SIZE - 12;
bytestream2_seek(&s->g, SGI_HEADER_SIZE, SEEK_SET);
if (rle) {
ret = read_rle_sgi(out_end, in_buf, in_end, s);
ret = read_rle_sgi(out_end, s);
} else {
ret = read_uncompressed_sgi(out_buf, out_end, in_buf, in_end, s);
ret = read_uncompressed_sgi(out_buf, out_end, s);
}
if (ret == 0) {
*picture = s->picture;
*data_size = sizeof(AVPicture);
return buf_size;
return avpkt->size;
} else {
return -1;
return ret;
}
}

@ -178,24 +178,45 @@ static int decode_frame(AVCodecContext *avctx,
}
if(colors){
size_t pal_size;
int pal_size, pal_sample_size;
if((colors + first_clr) > 256){
av_log(avctx, AV_LOG_ERROR, "Incorrect palette: %i colors with offset %i\n", colors, first_clr);
return -1;
}
if(csize != 24){
switch (csize) {
case 24: pal_sample_size = 3; break;
case 16:
case 15: pal_sample_size = 2; break;
default:
av_log(avctx, AV_LOG_ERROR, "Palette entry size %i bits is not supported\n", csize);
return -1;
}
pal_size = colors * ((csize + 1) >> 3);
pal_size = colors * pal_sample_size;
CHECK_BUFFER_SIZE(buf, buf_end, pal_size, "color table");
if(avctx->pix_fmt != PIX_FMT_PAL8)//should not occur but skip palette anyway
buf += pal_size;
else{
int t;
int32_t *pal = ((int32_t*)p->data[1]) + first_clr;
for(t = 0; t < colors; t++){
*pal++ = (0xff<<24) | bytestream_get_le24(&buf);
uint32_t *pal = ((uint32_t *)p->data[1]) + first_clr;
switch (pal_sample_size) {
case 3:
/* RGB24 */
for (t = 0; t < colors; t++)
*pal++ = (0xffU<<24) | bytestream_get_le24(&buf);
break;
case 2:
/* RGB555 */
for (t = 0; t < colors; t++) {
uint32_t v = bytestream_get_le16(&buf);
v = ((v & 0x7C00) << 9) |
((v & 0x03E0) << 6) |
((v & 0x001F) << 3);
/* left bit replication */
v |= (v & 0xE0E0E0U) >> 5;
*pal++ = (0xffU<<24) | v;
}
break;
}
p->palette_has_changed = 1;
}

@ -149,7 +149,7 @@ static av_cold void uninit(AVFilterContext *ctx)
static int query_formats(AVFilterContext *ctx)
{
const static enum PixelFormat pix_fmts[] = {
static const enum PixelFormat pix_fmts[] = {
PIX_FMT_YUV444P, PIX_FMT_YUV422P, PIX_FMT_YUV420P,
PIX_FMT_YUV411P, PIX_FMT_YUV410P,
PIX_FMT_YUVJ444P, PIX_FMT_YUVJ422P, PIX_FMT_YUVJ420P,

@ -72,7 +72,7 @@ typedef struct FormatEntry {
int is_supported_in, is_supported_out;
} FormatEntry;
const static FormatEntry format_entries[PIX_FMT_NB] = {
static const FormatEntry format_entries[PIX_FMT_NB] = {
[PIX_FMT_YUV420P] = { 1 , 1 },
[PIX_FMT_YUYV422] = { 1 , 1 },
[PIX_FMT_RGB24] = { 1 , 1 },

@ -56,7 +56,7 @@ SECTION .text
; of 2. $offset is either 0 or 3. $dither holds 8 values.
;-----------------------------------------------------------------------------
%macro yuv2planeX_fn 4
%macro yuv2planeX_fn 3
%ifdef ARCH_X86_32
%define cntr_reg r1
@ -66,12 +66,12 @@ SECTION .text
%define movsx movsxd
%endif
cglobal yuv2planeX_%2_%1, %4, 7, %3
%if %2 == 8 || %2 == 9 || %2 == 10
cglobal yuv2planeX_%1, %3, 7, %2
%if %1 == 8 || %1 == 9 || %1 == 10
pxor m6, m6
%endif ; %2 == 8/9/10
%endif ; %1 == 8/9/10
%if %2 == 8
%if %1 == 8
%ifdef ARCH_X86_32
%assign pad 0x2c - (stack_offset & 15)
SUB rsp, pad
@ -120,7 +120,7 @@ cglobal yuv2planeX_%2_%1, %4, 7, %3
mova [rsp+16], m3
mova [rsp+24], m_dith
%endif ; mmsize == 8/16
%endif ; %2 == 8
%endif ; %1 == 8
xor r5, r5
@ -130,11 +130,11 @@ cglobal yuv2planeX_%2_%1, %4, 7, %3
; 8 pixels but we can only handle 2 pixels per register, and thus 4
; pixels per iteration. In order to not have to keep track of where
; we are w.r.t. dithering, we unroll the mmx/8bit loop x2.
%if %2 == 8
%if %1 == 8
%rep 16/mmsize
%endif ; %2 == 8
%endif ; %1 == 8
%if %2 == 8
%if %1 == 8
%ifdef ARCH_X86_32
mova m2, [rsp+mmsize*(0+%%i)]
mova m1, [rsp+mmsize*(1+%%i)]
@ -142,31 +142,31 @@ cglobal yuv2planeX_%2_%1, %4, 7, %3
mova m2, m8
mova m1, m_dith
%endif ; x86-32/64
%else ; %2 == 9/10/16
mova m1, [yuv2yuvX_%2_start]
%else ; %1 == 9/10/16
mova m1, [yuv2yuvX_%1_start]
mova m2, m1
%endif ; %2 == 8/9/10/16
%endif ; %1 == 8/9/10/16
movsx cntr_reg, r1m
.filterloop_ %+ %%i:
; input pixels
mov r6, [r2+gprsize*cntr_reg-2*gprsize]
%if %2 == 16
%if %1 == 16
mova m3, [r6+r5*4]
mova m5, [r6+r5*4+mmsize]
%else ; %2 == 8/9/10
%else ; %1 == 8/9/10
mova m3, [r6+r5*2]
%endif ; %2 == 8/9/10/16
%endif ; %1 == 8/9/10/16
mov r6, [r2+gprsize*cntr_reg-gprsize]
%if %2 == 16
%if %1 == 16
mova m4, [r6+r5*4]
mova m6, [r6+r5*4+mmsize]
%else ; %2 == 8/9/10
%else ; %1 == 8/9/10
mova m4, [r6+r5*2]
%endif ; %2 == 8/9/10/16
%endif ; %1 == 8/9/10/16
; coefficients
movd m0, [r0+2*cntr_reg-4]; coeff[0], coeff[1]
%if %2 == 16
%if %1 == 16
pshuflw m7, m0, 0 ; coeff[0]
pshuflw m0, m0, 0x55 ; coeff[1]
pmovsxwd m7, m7 ; word -> dword
@ -181,7 +181,7 @@ cglobal yuv2planeX_%2_%1, %4, 7, %3
paddd m1, m5
paddd m2, m4
paddd m1, m6
%else ; %2 == 10/9/8
%else ; %1 == 10/9/8
punpcklwd m5, m3, m4
punpckhwd m3, m4
SPLATD m0, m0
@ -191,85 +191,84 @@ cglobal yuv2planeX_%2_%1, %4, 7, %3
paddd m2, m5
paddd m1, m3
%endif ; %2 == 8/9/10/16
%endif ; %1 == 8/9/10/16
sub cntr_reg, 2
jg .filterloop_ %+ %%i
%if %2 == 16
psrad m2, 31 - %2
psrad m1, 31 - %2
%else ; %2 == 10/9/8
psrad m2, 27 - %2
psrad m1, 27 - %2
%endif ; %2 == 8/9/10/16
%if %1 == 16
psrad m2, 31 - %1
psrad m1, 31 - %1
%else ; %1 == 10/9/8
psrad m2, 27 - %1
psrad m1, 27 - %1
%endif ; %1 == 8/9/10/16
%if %2 == 8
%if %1 == 8
packssdw m2, m1
packuswb m2, m2
movh [r3+r5*1], m2
%else ; %2 == 9/10/16
%if %2 == 16
%else ; %1 == 9/10/16
%if %1 == 16
packssdw m2, m1
paddw m2, [minshort]
%else ; %2 == 9/10
%ifidn %1, sse4
packusdw m2, m1
%elifidn %1, avx
%else ; %1 == 9/10
%if cpuflag(sse4)
packusdw m2, m1
%else ; mmx2/sse2
packssdw m2, m1
pmaxsw m2, m6
%endif ; mmx2/sse2/sse4/avx
pminsw m2, [yuv2yuvX_%2_upper]
%endif ; %2 == 9/10/16
pminsw m2, [yuv2yuvX_%1_upper]
%endif ; %1 == 9/10/16
mova [r3+r5*2], m2
%endif ; %2 == 8/9/10/16
%endif ; %1 == 8/9/10/16
add r5, mmsize/2
sub r4d, mmsize/2
%if %2 == 8
%if %1 == 8
%assign %%i %%i+2
%endrep
%endif ; %2 == 8
%endif ; %1 == 8
jg .pixelloop
%if %2 == 8
%if %1 == 8
%ifdef ARCH_X86_32
ADD rsp, pad
RET
%else ; x86-64
REP_RET
%endif ; x86-32/64
%else ; %2 == 9/10/16
%else ; %1 == 9/10/16
REP_RET
%endif ; %2 == 8/9/10/16
%endif ; %1 == 8/9/10/16
%endmacro
%define PALIGNR PALIGNR_MMX
%ifdef ARCH_X86_32
INIT_MMX
yuv2planeX_fn mmx2, 8, 0, 7
yuv2planeX_fn mmx2, 9, 0, 5
yuv2planeX_fn mmx2, 10, 0, 5
INIT_MMX mmx2
yuv2planeX_fn 8, 0, 7
yuv2planeX_fn 9, 0, 5
yuv2planeX_fn 10, 0, 5
%endif
INIT_XMM
yuv2planeX_fn sse2, 8, 10, 7
yuv2planeX_fn sse2, 9, 7, 5
yuv2planeX_fn sse2, 10, 7, 5
INIT_XMM sse2
yuv2planeX_fn 8, 10, 7
yuv2planeX_fn 9, 7, 5
yuv2planeX_fn 10, 7, 5
%define PALIGNR PALIGNR_SSSE3
yuv2planeX_fn sse4, 8, 10, 7
yuv2planeX_fn sse4, 9, 7, 5
yuv2planeX_fn sse4, 10, 7, 5
yuv2planeX_fn sse4, 16, 8, 5
INIT_XMM sse4
yuv2planeX_fn 8, 10, 7
yuv2planeX_fn 9, 7, 5
yuv2planeX_fn 10, 7, 5
yuv2planeX_fn 16, 8, 5
%ifdef HAVE_AVX
INIT_AVX
yuv2planeX_fn avx, 8, 10, 7
yuv2planeX_fn avx, 9, 7, 5
yuv2planeX_fn avx, 10, 7, 5
INIT_XMM avx
yuv2planeX_fn 8, 10, 7
yuv2planeX_fn 9, 7, 5
yuv2planeX_fn 10, 7, 5
%endif
; %1=outout-bpc, %2=alignment (u/a)

@ -34,10 +34,12 @@ FATE_TESTS += $(FATE_IMAGE)
fate-image: $(FATE_IMAGE)
FATE_TARGA = CBW8 \
CCM8 \
CTC16 \
CTC24 \
CTC32 \
UBW8 \
UCM8 \
UTC16 \
UTC24 \
UTC32
@ -49,12 +51,12 @@ FATE_TESTS += $(FATE_TARGA)
fate-targa: $(FATE_TARGA)
fate-targa-conformance-CBW8: CMD = framecrc -i $(SAMPLES)/targa-conformance/CBW8.TGA
# fate-targa-conformance-CCM8: CMD = framecrc -i $(SAMPLES)/targa-conformance/CCM8.TGA
fate-targa-conformance-CCM8: CMD = framecrc -i $(SAMPLES)/targa-conformance/CCM8.TGA -pix_fmt rgba
fate-targa-conformance-CTC16: CMD = framecrc -i $(SAMPLES)/targa-conformance/CTC16.TGA -pix_fmt rgb555le
fate-targa-conformance-CTC24: CMD = framecrc -i $(SAMPLES)/targa-conformance/CTC24.TGA
fate-targa-conformance-CTC32: CMD = framecrc -i $(SAMPLES)/targa-conformance/CTC32.TGA -pix_fmt bgra
fate-targa-conformance-UBW8: CMD = framecrc -i $(SAMPLES)/targa-conformance/UBW8.TGA
# fate-targa-conformance-UCM8: CMD = framecrc -i $(SAMPLES)/targa-conformance/UCM8.TGA
fate-targa-conformance-UCM8: CMD = framecrc -i $(SAMPLES)/targa-conformance/UCM8.TGA -pix_fmt rgba
fate-targa-conformance-UTC16: CMD = framecrc -i $(SAMPLES)/targa-conformance/UTC16.TGA -pix_fmt rgb555le
fate-targa-conformance-UTC24: CMD = framecrc -i $(SAMPLES)/targa-conformance/UTC24.TGA
fate-targa-conformance-UTC32: CMD = framecrc -i $(SAMPLES)/targa-conformance/UTC32.TGA -pix_fmt bgra

@ -0,0 +1 @@
0, 0, 65536, 0x47e97fe9

@ -0,0 +1 @@
0, 0, 65536, 0x47e97fe9
Loading…
Cancel
Save