Re: [libav-devel] [PATCH] Revert "VC1: merge idct8x8, coeff adjustments and put_pixels."
On Tue, May 03, 2011 at 10:40:51PM -0400, Ronald S. Bultje wrote: > This reverts commit f8bed30d8b176fa030f6737765338bb4a2bcabc9. > --- > libavcodec/ppc/vc1dsp_altivec.c | 63 +- > libavcodec/vc1.c| 28 + > libavcodec/vc1dec.c | 58 +++ > libavcodec/vc1dsp.c | 54 ++--- > libavcodec/vc1dsp.h |6 +--- > 5 files changed, 39 insertions(+), 170 deletions(-) fine with me, just don't forget to state why it's needed in commit message ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/2] Replace strncpy() with av_strlcpy().
On Tue, May 3, 2011 at 12:45 PM, Justin Ruggles wrote: > On 05/03/2011 03:05 PM, Aℓex Converse wrote: > >> On Tue, May 3, 2011 at 11:45 AM, Justin Ruggles >> wrote: >>> On 05/03/2011 02:21 PM, Alex Converse wrote: >>> memset(compressor_name,0,32); /* FIXME not sure, ISO 14496-1 draft where it shall be set to 0 */ if (track->mode == MODE_MOV && track->enc->codec && track->enc->codec->name) - strncpy(compressor_name,track->enc->codec->name,31); + av_strlcpy(compressor_name,track->enc->codec->name,32); >>> >>> >>> Can't that memset be replaced with compressor_name[0] = '\0' now that >>> the result will be null-terminated? >>> >> >> avio_w8(pb, strlen(compressor_name)); >> avio_write(pb, compressor_name, 31); >> >> avio_write() is a mem-function not a str-function so we still want it >> zerofilled to the end. Still I can't help but notice that this looks >> like a pascal string and wonder if >> >> avio_w8(pb, strlen(compressor_name)); >> avio_write(pb, compressor_name, strlen(compressor_name)); >> >> would be better. > > ah, I didn't notice that. probably is better that way from the look of > it, but I don't know what the MOV spec says. > >> On Tue, May 3, 2011 at 11:46 AM, Justin Ruggles >> wrote: >>> On 05/03/2011 02:21 PM, Alex Converse wrote: >>> @@ -1099,4 +1100,3 @@ void pp_postprocess(const uint8_t * src[3], const int srcStride[3], } } } - >>> >>> >>> stray cosmetic change >>> >> >> Fixed > > > patch ok. > Queued ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 4/4] get_bits: make cache unsigned to eliminate undefined signed overflow.
On Sun, May 1, 2011 at 5:48 AM, Måns Rullgård wrote: > Aℓex Converse writes: > >> From 859f6de86ee7359942a9aece8d92eea141d421d0 Mon Sep 17 00:00:00 2001 >> From: Alex Converse >> Date: Tue, 26 Apr 2011 09:09:05 -0700 >> Subject: [PATCH 4/4] get_bits: make cache unsigned to eliminate undefined >> signed overflow. >> MIME-Version: 1.0 >> Content-Type: multipart/mixed; boundary="1" >> >> This is a multi-part message in MIME format. >> --1 >> Content-Type: text/plain; charset=UTF-8; format=fixed >> Content-Transfer-Encoding: 8bit >> >> --- >> libavcodec/get_bits.h | 2 +- >> 1 files changed, 1 insertions(+), 1 deletions(-) >> >> --1 >> Content-Type: text/x-patch; >> name="0004-get_bits-make-cache-unsigned-to-eliminate-undefined-.patch" >> Content-Transfer-Encoding: 8bit >> Content-Disposition: attachment; >> filename="0004-get_bits-make-cache-unsigned-to-eliminate-undefined-.patch" >> >> diff --git a/libavcodec/get_bits.h b/libavcodec/get_bits.h >> index 2d36b81..185ff31 100644 >> --- a/libavcodec/get_bits.h >> +++ b/libavcodec/get_bits.h >> @@ -127,7 +127,7 @@ for examples see get_bits, show_bits, skip_bits, get_vlc >> >> # define OPEN_READER(name, gb) \ >> unsigned int name##_index = (gb)->index; \ >> - int name##_cache = 0 >> + unsigned int name##_cache = 0 >> >> # define CLOSE_READER(name, gb) (gb)->index = name##_index > > Should be OK. > Queued ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/4] indeo3: Eliminate use of long.
On Tue, May 3, 2011 at 9:53 AM, Justin Ruggles wrote:
> On 04/27/2011 08:10 PM, Aℓex Converse wrote:
>
>> ---
>> libavcodec/indeo3.c | 22 +++---
>> 1 files changed, 11 insertions(+), 11 deletions(-)
>>
>>
>> --1
>> Content-Type: text/x-patch; name="0001-indeo3-Eliminate-use-of-long.patch"
>> Content-Transfer-Encoding: 8bit
>> Content-Disposition: attachment;
>> filename="0001-indeo3-Eliminate-use-of-long.patch"
>>
>> diff --git a/libavcodec/indeo3.c b/libavcodec/indeo3.c
>> index 29a8d4a..30b7085 100644
>> --- a/libavcodec/indeo3.c
>> +++ b/libavcodec/indeo3.c
>> @@ -149,13 +149,13 @@ static av_cold void iv_free_func(Indeo3DecodeContext
>> *s)
>> }
>>
>> struct ustr {
>> - long xpos;
>> - long ypos;
>> - long width;
>> - long height;
>> - long split_flag;
>> - long split_direction;
>> - long usl7;
>> + int xpos;
>> + int ypos;
>> + int width;
>> + int height;
>> + int split_flag;
>> + int split_direction;
>> + int usl7;
>> };
>>
>>
>> @@ -203,12 +203,12 @@ struct ustr {
>>
>> static void iv_Decode_Chunk(Indeo3DecodeContext *s,
>> uint8_t *cur, uint8_t *ref, int width, int height,
>> - const uint8_t *buf1, long cb_offset, const uint8_t *hdr,
>> + const uint8_t *buf1, int cb_offset, const uint8_t *hdr,
>> const uint8_t *buf2, int min_width_160)
>> {
>> uint8_t bit_buf;
>> - unsigned long bit_pos, lv, lv1, lv2;
>> - long *width_tbl, width_tbl_arr[10];
>> + unsigned int bit_pos, lv, lv1, lv2;
>> + int *width_tbl, width_tbl_arr[10];
>> const signed char *ref_vectors;
>> uint8_t *cur_frm_pos, *ref_frm_pos, *cp, *cp2;
>> uint32_t *cur_lp, *ref_lp;
>> @@ -982,7 +982,7 @@ static int iv_decode_frame(AVCodecContext *avctx,
>> Indeo3DecodeContext *s = avctx->priv_data;
>> unsigned int image_width, image_height,
>> chroma_width, chroma_height;
>> - unsigned long flags, cb_offset, data_size,
>> + unsigned int flags, cb_offset, data_size,
>> y_offset, v_offset, u_offset, mc_vector_count;
>> const uint8_t *hdr_pos, *buf_pos;
>
>
> Ok.
>
Queued
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] Revert "VC1: merge idct8x8, coeff adjustments and put_pixels."
Hi, On Tue, May 3, 2011 at 10:40 PM, Ronald S. Bultje wrote: > This reverts commit f8bed30d8b176fa030f6737765338bb4a2bcabc9. > --- > libavcodec/ppc/vc1dsp_altivec.c | 63 +- > libavcodec/vc1.c | 28 + > libavcodec/vc1dec.c | 58 +++ > libavcodec/vc1dsp.c | 54 ++--- > libavcodec/vc1dsp.h | 6 +--- > 5 files changed, 39 insertions(+), 170 deletions(-) Just FYI, I reverted this patch because it conflicts with bit-exactness of our VC-1 decoder. The patches that Kostya approved, and also this one, will all have a slightly negative impact on performance, at least in the short term. I intend to eventually fix that, but can't do much about it right now. We need to conform first, performance comes later. Ronald ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH] Revert "VC1: merge idct8x8, coeff adjustments and put_pixels."
This reverts commit f8bed30d8b176fa030f6737765338bb4a2bcabc9.
---
libavcodec/ppc/vc1dsp_altivec.c | 63 +-
libavcodec/vc1.c| 28 +
libavcodec/vc1dec.c | 58 +++
libavcodec/vc1dsp.c | 54 ++---
libavcodec/vc1dsp.h |6 +---
5 files changed, 39 insertions(+), 170 deletions(-)
diff --git a/libavcodec/ppc/vc1dsp_altivec.c b/libavcodec/ppc/vc1dsp_altivec.c
index c52b5a9..307e0e9 100644
--- a/libavcodec/ppc/vc1dsp_altivec.c
+++ b/libavcodec/ppc/vc1dsp_altivec.c
@@ -130,8 +130,7 @@ do { \
/** Do inverse transform on 8x8 block
*/
-static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
- int sign, int rangered)
+static void vc1_inv_trans_8x8_altivec(DCTELEM block[64])
{
vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
vector signed int s0, s1, s2, s3, s4, s5, s6, s7;
@@ -145,9 +144,6 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
const vector unsigned int vec_2 = vec_splat_u32(2);
const vector signed int vec_1s = vec_splat_s32(1);
const vector unsigned int vec_1 = vec_splat_u32(1);
-const vector unsigned short rangered_shift = vec_splat_u16(1);
-const vector signed short signed_bias = vec_sl(vec_splat_s16(4),
- vec_splat_u16(4));
src0 = vec_ld( 0, block);
src1 = vec_ld( 16, block);
@@ -217,27 +213,6 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
src6 = vec_pack(sE, s6);
src7 = vec_pack(sF, s7);
-if (rangered) {
-if (!sign) {
-src0 = vec_sub(src0, signed_bias);
-src1 = vec_sub(src1, signed_bias);
-src2 = vec_sub(src2, signed_bias);
-src3 = vec_sub(src3, signed_bias);
-src4 = vec_sub(src4, signed_bias);
-src5 = vec_sub(src5, signed_bias);
-src6 = vec_sub(src6, signed_bias);
-src7 = vec_sub(src7, signed_bias);
-}
-src0 = vec_sl(src0, rangered_shift);
-src1 = vec_sl(src1, rangered_shift);
-src2 = vec_sl(src2, rangered_shift);
-src3 = vec_sl(src3, rangered_shift);
-src4 = vec_sl(src4, rangered_shift);
-src5 = vec_sl(src5, rangered_shift);
-src6 = vec_sl(src6, rangered_shift);
-src7 = vec_sl(src7, rangered_shift);
-}
-
vec_st(src0, 0, block);
vec_st(src1, 16, block);
vec_st(src2, 32, block);
@@ -248,36 +223,6 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
vec_st(src7,112, block);
}
-static void vc1_inv_trans_8x8_add_altivec(uint8_t *dest, int stride, DCTELEM
*b)
-{
-vc1_inv_trans_8x8_altivec(b, 0, 0);
-ff_add_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_signed_altivec(uint8_t *dest, int stride,
DCTELEM *b)
-{
-vc1_inv_trans_8x8_altivec(b, 1, 0);
-ff_put_signed_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_signed_rangered_altivec(uint8_t *dest, int
stride, DCTELEM *b)
-{
-vc1_inv_trans_8x8_altivec(b, 1, 1);
-ff_put_signed_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_altivec(uint8_t *dest, int stride, DCTELEM
*b)
-{
-vc1_inv_trans_8x8_altivec(b, 0, 0);
-ff_put_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_rangered_altivec(uint8_t *dest, int stride,
DCTELEM *b)
-{
-vc1_inv_trans_8x8_altivec(b, 0, 1);
-ff_put_pixels_clamped_c(b, dest, stride);
-}
-
/** Do inverse transform on 8x4 part of block
*/
static void vc1_inv_trans_8x4_altivec(uint8_t *dest, int stride, DCTELEM
*block)
@@ -396,11 +341,7 @@ void ff_vc1dsp_init_altivec(VC1DSPContext* dsp)
if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
return;
-dsp->vc1_inv_trans_8x8_add = vc1_inv_trans_8x8_add_altivec;
-dsp->vc1_inv_trans_8x8_put_signed[0] =
vc1_inv_trans_8x8_put_signed_altivec;
-dsp->vc1_inv_trans_8x8_put_signed[1] =
vc1_inv_trans_8x8_put_signed_rangered_altivec;
-dsp->vc1_inv_trans_8x8_put[0] = vc1_inv_trans_8x8_put_altivec;
-dsp->vc1_inv_trans_8x8_put[1] = vc1_inv_trans_8x8_put_rangered_altivec;
+dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_altivec;
dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_altivec;
dsp->put_no_rnd_vc1_chroma_pixels_tab[0] =
put_no_rnd_vc1_chroma_mc8_altivec;
dsp->avg_no_rnd_vc1_chroma_pixels_tab[0] =
avg_no_rnd_vc1_chroma_mc8_altivec;
diff --git a/libavcodec/vc1.c b/libavcodec/vc1.c
index 536320a..7b878c1 100644
--- a/libavcodec/vc1.c
+++ b/libavcodec/vc1.c
@@ -280,28 +280,6 @@ static int vop_dquant_decoding(VC1Context *v)
static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb);
-static void simple_idct_put_rangered(uint8_t *dest, int line_size, DCTELEM
*block)
-{
-int i;
-ff_simple_idct(block);
-for (i =
Re: [libav-devel] [PATCH] asfdec: fix assert failure on invalid files
Hi, On Tue, May 3, 2011 at 10:34 PM, Luca Barbato wrote: > On 5/4/11 4:18 AM, Ronald S. Bultje wrote: >> >> From: Uoti Urpala >> >> Add an extra size validity check in asf_read_frame_header(). Without >> this asf->packet_size_left may become negative, which triggers an >> assertion failure later. >> >> Signed-off-by: Ronald S. Bultje >> --- >> libavformat/asfdec.c | 5 + >> 1 files changed, 5 insertions(+), 0 deletions(-) > > Seems fine, not sure about the new line added. Yeah I noticed after sending it, removed locally. Ronald ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] asfdec: fix assert failure on invalid files
On 5/4/11 4:18 AM, Ronald S. Bultje wrote: From: Uoti Urpala Add an extra size validity check in asf_read_frame_header(). Without this asf->packet_size_left may become negative, which triggers an assertion failure later. Signed-off-by: Ronald S. Bultje --- libavformat/asfdec.c |5 + 1 files changed, 5 insertions(+), 0 deletions(-) Seems fine, not sure about the new line added. ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 3/3] mpegvideo: reindent.
On 5/3/11 2:23 PM, Ronald S. Bultje wrote: --- libavcodec/mpegvideo.c | 227 +++ 1 files changed, 111 insertions(+), 116 deletions(-) I guess ok, maybe long lines could be broken as well. ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] avfilter: check malloc return values.
On 5/4/11 4:06 AM, Ronald S. Bultje wrote: --- libavfilter/avfilter.c | 26 +- 1 files changed, 25 insertions(+), 1 deletions(-) Looks ok. ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 3/3] mpegvideo: reindent.
Hi, On Tue, May 3, 2011 at 8:23 AM, Ronald S. Bultje wrote: > --- > libavcodec/mpegvideo.c | 227 +++ > 1 files changed, 111 insertions(+), 116 deletions(-) Ping. Ronald ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH] asfdec: fix assert failure on invalid files
From: Uoti Urpala
Add an extra size validity check in asf_read_frame_header(). Without
this asf->packet_size_left may become negative, which triggers an
assertion failure later.
Signed-off-by: Ronald S. Bultje
---
libavformat/asfdec.c |5 +
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/libavformat/asfdec.c b/libavformat/asfdec.c
index 637ceed..76cd7a7 100644
--- a/libavformat/asfdec.c
+++ b/libavformat/asfdec.c
@@ -852,6 +852,7 @@ static int asf_read_frame_header(AVFormatContext *s,
AVIOContext *pb){
}
if (asf->packet_flags & 0x01) {
DO_2BITS(asf->packet_segsizetype >> 6, asf->packet_frag_size, 0); // 0
is illegal
+
if(asf->packet_frag_size > asf->packet_size_left - rsize){
if (asf->packet_frag_size > asf->packet_size_left - rsize +
asf->packet_padsize) {
av_log(s, AV_LOG_ERROR, "packet_frag_size is invalid
(%d-%d)\n", asf->packet_size_left, rsize);
@@ -864,6 +865,10 @@ static int asf_read_frame_header(AVFormatContext *s,
AVIOContext *pb){
}
//printf("Fragsize %d\n", asf->packet_frag_size);
} else {
+if (rsize > asf->packet_size_left) {
+av_log(s, AV_LOG_ERROR, "packet_replic_size is invalid\n");
+return -1;
+}
asf->packet_frag_size = asf->packet_size_left - rsize;
//printf("Using rest %d %d %d\n", asf->packet_frag_size,
asf->packet_size_left, rsize);
}
--
1.7.2.1
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH] avfilter: check malloc return values.
---
libavfilter/avfilter.c | 26 +-
1 files changed, 25 insertions(+), 1 deletions(-)
diff --git a/libavfilter/avfilter.c b/libavfilter/avfilter.c
index c71c046..82350d1 100644
--- a/libavfilter/avfilter.c
+++ b/libavfilter/avfilter.c
@@ -583,29 +583,53 @@ int avfilter_open(AVFilterContext **filter_ctx, AVFilter
*filter, const char *in
return AVERROR(EINVAL);
ret = av_mallocz(sizeof(AVFilterContext));
+if (!ret)
+return AVERROR(ENOMEM);
ret->av_class = &avfilter_class;
ret->filter = filter;
ret->name = inst_name ? av_strdup(inst_name) : NULL;
-if (filter->priv_size)
+if (filter->priv_size) {
ret->priv = av_mallocz(filter->priv_size);
+if (!ret->priv)
+goto err;
+}
ret->input_count = pad_count(filter->inputs);
if (ret->input_count) {
ret->input_pads = av_malloc(sizeof(AVFilterPad) * ret->input_count);
+if (!ret->input_pads)
+goto err;
memcpy(ret->input_pads, filter->inputs, sizeof(AVFilterPad) *
ret->input_count);
ret->inputs = av_mallocz(sizeof(AVFilterLink*) *
ret->input_count);
+if (!ret->inputs)
+goto err;
}
ret->output_count = pad_count(filter->outputs);
if (ret->output_count) {
ret->output_pads = av_malloc(sizeof(AVFilterPad) * ret->output_count);
+if (!ret->output_pads)
+goto err;
memcpy(ret->output_pads, filter->outputs, sizeof(AVFilterPad) *
ret->output_count);
ret->outputs = av_mallocz(sizeof(AVFilterLink*) *
ret->output_count);
+if (!ret->outputs)
+goto err;
}
*filter_ctx = ret;
return 0;
+
+err:
+av_freep(&ret->inputs);
+av_freep(&ret->input_pads);
+ret->input_count = 0;
+av_freep(&ret->outputs);
+av_freep(&ret->output_pads);
+ret->output_count = 0;
+av_freep(&ret->priv);
+av_free(ret);
+return AVERROR(ENOMEM);
}
void avfilter_free(AVFilterContext *filter)
--
1.7.2.1
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] Port drawtext filter by Hemanth from the libavfilter soc repo
On 05/04/2011 02:05 AM, Stefano Sabatini wrote: > The changes look fine, but I consider the idea of picking a patch and > randomly fixing it very bad, indeed this is making very hard to spot > the differences with the original and spoiling authorship/blame > information. The signed off tag means that the first person to be contacted about a commit is the undersigned, then the author. > I suggest to apply the original patch, and apply the fixes on top of > that, this also helps the original contributor to review and > eventually port the various fixes. I'd rather not have multiple commits just because the patch imported wasn't reviewed when it hit some foreign tree... lu -- Luca Barbato Gentoo/linux http://dev.gentoo.org/~lu_zero ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] Port drawtext filter by Hemanth from the libavfilter soc repo
Hi, On Tue, May 3, 2011 at 5:05 PM, Stefano Sabatini wrote: > I suggest to apply the original patch, and apply the fixes on top of > that, this also helps the original contributor to review and > eventually port the various fixes. If our goal was to make merging and upstreaming to FFmpeg easy, I would agree with you. But I don't believe that is our goal. Ronald ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] ac3enc: add support for transient detection and block switching
On 05/03/2011 08:19 PM, Justin Ruggles wrote:
> ---
> This is an updated patch which removes some unrelated hunks and also makes it
> easier to add block switching support to the ac3_fixed encoder once we have
> a fixed-point iir filter.
>
> Still depends on the channel coupling patch.
>
> doc/encoders.texi |7 +
> libavcodec/ac3enc.c | 297 +++-
> libavcodec/ac3enc_fixed.c |4 +-
> libavcodec/ac3enc_float.c | 24 +++-
> libavcodec/iirfilter.c|6 +
> libavcodec/iirfilter.h| 10 ++
> 6 files changed, 336 insertions(+), 12 deletions(-)
>
>
> @@ -628,6 +877,11 @@ static void apply_channel_coupling(AC3EncodeContext *s)
> new_coords = 1;
> last_coord_blk = blk;
> }
> +/* for stereo, encode coordinates more often */
> +if (s->fbw_channels == 2 && blk > 2 && last_coord_blk < (blk-2))
> {
> +new_coords = 1;
> +last_coord_blk = blk;
> +}
> }
> s->blocks[blk].new_cpl_coords = new_coords;
> }
*grumble* I don't know how that hunk slipped back into this patch...
it's supposed to be part of the coupling patch not this one.
-Justin
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH] ac3enc: add support for transient detection and block switching
---
This is an updated patch which removes some unrelated hunks and also makes it
easier to add block switching support to the ac3_fixed encoder once we have
a fixed-point iir filter.
Still depends on the channel coupling patch.
doc/encoders.texi |7 +
libavcodec/ac3enc.c | 297 +++-
libavcodec/ac3enc_fixed.c |4 +-
libavcodec/ac3enc_float.c | 24 +++-
libavcodec/iirfilter.c|6 +
libavcodec/iirfilter.h| 10 ++
6 files changed, 336 insertions(+), 12 deletions(-)
diff --git a/doc/encoders.texi b/doc/encoders.texi
index 3f6b4c9..aa3871a 100644
--- a/doc/encoders.texi
+++ b/doc/encoders.texi
@@ -357,6 +357,13 @@ HDCD A/D Converter
@table @option
+@item -block_switching @var{boolean}
+Block Switching. Enables/Disables switching from long blocks to short blocks
+based on transient detection. This allows the encoder to determine when it is
+best to use short blocks for transient signals. This option is enabled by
+default, and it is recommended that it be left as enabled except for testing
+purposes or to increase encoding speed.
+
@item -stereo_rematrixing @var{boolean}
Stereo Rematrixing. Enables/Disables use of rematrixing for stereo input. This
is an optional AC-3 feature that increases quality by selectively encoding
diff --git a/libavcodec/ac3enc.c b/libavcodec/ac3enc.c
index f09fdf6..c65d74b 100644
--- a/libavcodec/ac3enc.c
+++ b/libavcodec/ac3enc.c
@@ -42,6 +42,7 @@
#include "ac3.h"
#include "audioconvert.h"
#include "fft.h"
+#include "iirfilter.h"
#ifndef CONFIG_AC3ENC_FLOAT
@@ -66,7 +67,9 @@ typedef int64_t CoefSumType;
typedef struct AC3MDCTContext {
const SampleType *window; ///< MDCT window function
-FFTContext fft; ///< FFT context for MDCT calculation
+SampleType *rot_tmp;///< temp buffer for pre-rotated samples
+CoefType *coef_tmp; ///< temp buffer for 256-point MDCT coefs
+FFTContext fft[2]; ///< FFT context for MDCT calculation
} AC3MDCTContext;
/**
@@ -97,6 +100,7 @@ typedef struct AC3EncOptions {
/* other encoding options */
int allow_per_frame_metadata;
+int block_switching;
int stereo_rematrixing;
int channel_coupling;
int cpl_start;
@@ -117,6 +121,7 @@ typedef struct AC3Block {
uint16_t **qmant; ///< quantized mantissas
uint8_t **cpl_coord_exp; ///< coupling coord exponents (cplcoexp)
uint8_t **cpl_coord_mant; ///< coupling coord mantissas (cplcomant)
+uint8_t blk_switch[AC3_MAX_CHANNELS][2]; ///< block switch flags [0]=current [1]=previous
uint8_t coeff_shift[AC3_MAX_CHANNELS]; ///< fixed-point coefficient shift values
uint8_t new_rematrixing_strategy; ///< send new rematrixing flags in this block
int num_rematrixing_bands; ///< number of rematrixing bands
@@ -180,6 +185,11 @@ typedef struct AC3EncodeContext {
int start_freq[AC3_MAX_CHANNELS]; ///< start frequency bin(strtmant)
int cpl_end_freq; ///< coupling channel end frequency bin
+int blksw_enabled; ///< block switching enabled
+struct FFIIRFilterCoeffs *blksw_coefs; ///< block switch filter coefficients
+struct FFIIRFilterState *blksw_state[2][AC3_MAX_CHANNELS]; /// block switch filter states
+struct FFIIRFilterState *temp_state;///< block switch filter temp state
+
int cpl_on; ///< coupling turned on for this frame
int cpl_enabled;///< coupling enabled for all frames
int num_cpl_subbands; ///< number of coupling subbands(ncplsubnd)
@@ -202,6 +212,7 @@ typedef struct AC3EncodeContext {
int exponent_bits; ///< number of bits used for exponents
SampleType **planar_samples;
+SampleType *filtered_samples;
uint8_t *bap_buffer;
uint8_t *bap1_buffer;
CoefType *mdct_coef_buffer;
@@ -286,6 +297,7 @@ static const AVOption options[] = {
{"standard", "Standard (default)", 0, FF_OPT_TYPE_CONST, 0, INT_MIN, INT_MAX, AC3ENC_PARAM, "ad_conv_type"},
{"hdcd", "HDCD", 0, FF_OPT_TYPE_CONST, 1, INT_MIN, INT_MAX, AC3ENC_PARAM, "ad_conv_type"},
/* Other Encoding Options */
+{"block_switching", "Block Switching", OFFSET(block_switching), FF_OPT_TYPE_INT, 1, 0, 1, AC3ENC_PARAM},
{"stereo_rematrixing", "Stereo Rematrixing", OFFSET(stereo_rematrixing), FF_OPT_TYPE_INT, 1, 0, 1, AC3ENC_PARAM},
{"channel_coupling", "Channel Coupling", OFFSET(channel_coupling), FF_OPT_TYPE_INT, 1, 0, 1, AC3ENC_PARAM, "channel_coupling"},
{"auto", "Selected by the Encoder", 0, FF_OPT_TYPE_CONST, -1, INT_MIN, INT_MAX, AC3ENC_PARAM, "channel_coupling"},
@@ -471,6 +483,237 @@ static void deinterleave_input_samples(
Re: [libav-devel] [PATCH] Port drawtext filter by Hemanth from the libavfilter soc repo
On date Wednesday 2011-04-13 08:20:30 +0200, Anton Khirnov encoded:
> From: Stefano Sabatini
>
> With the following additions:
> * support to anti-aliased glyph rendering
> * support to UTF-8 text and Unicode chars rendering
> * support for RGB packed formats
> * fix minor errors and typos in the filter description
> * extend/clarify examples in the filter description
>
> Signed-off-by: Anton Khirnov
[...]
> +static int glyph_cmp(void *key, const void *b)
> +{
> +const Glyph *a = key, *bb = b;
> +int64_t diff = (int64_t)a->code - (int64_t)bb->code;
> +return diff > 0 ? 1 : diff < 0 ? -1 : 0;
> +}
Nit: key->a, b->bb is a little schizofrenic
...
The changes look fine, but I consider the idea of picking a patch and
randomly fixing it very bad, indeed this is making very hard to spot
the differences with the original and spoiling authorship/blame
information.
I suggest to apply the original patch, and apply the fixes on top of
that, this also helps the original contributor to review and
eventually port the various fixes.
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/2] Replace strncpy() with av_strlcpy().
On 05/03/2011 03:05 PM, Aℓex Converse wrote: > On Tue, May 3, 2011 at 11:45 AM, Justin Ruggles > wrote: >> On 05/03/2011 02:21 PM, Alex Converse wrote: >> >>> memset(compressor_name,0,32); >>> /* FIXME not sure, ISO 14496-1 draft where it shall be set to 0 */ >>> if (track->mode == MODE_MOV && track->enc->codec && >>> track->enc->codec->name) >>> -strncpy(compressor_name,track->enc->codec->name,31); >>> +av_strlcpy(compressor_name,track->enc->codec->name,32); >> >> >> Can't that memset be replaced with compressor_name[0] = '\0' now that >> the result will be null-terminated? >> > > avio_w8(pb, strlen(compressor_name)); > avio_write(pb, compressor_name, 31); > > avio_write() is a mem-function not a str-function so we still want it > zerofilled to the end. Still I can't help but notice that this looks > like a pascal string and wonder if > > avio_w8(pb, strlen(compressor_name)); > avio_write(pb, compressor_name, strlen(compressor_name)); > > would be better. ah, I didn't notice that. probably is better that way from the look of it, but I don't know what the MOV spec says. > On Tue, May 3, 2011 at 11:46 AM, Justin Ruggles > wrote: >> On 05/03/2011 02:21 PM, Alex Converse wrote: >> >>> @@ -1099,4 +1100,3 @@ void pp_postprocess(const uint8_t * src[3], const >>> int srcStride[3], >>> } >>> } >>> } >>> - >> >> >> stray cosmetic change >> > > Fixed patch ok. -Justin ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/2] Replace strncpy() with av_strlcpy().
On Tue, May 3, 2011 at 11:45 AM, Justin Ruggles
wrote:
> On 05/03/2011 02:21 PM, Alex Converse wrote:
>
>> memset(compressor_name,0,32);
>> /* FIXME not sure, ISO 14496-1 draft where it shall be set to 0 */
>> if (track->mode == MODE_MOV && track->enc->codec &&
>> track->enc->codec->name)
>> - strncpy(compressor_name,track->enc->codec->name,31);
>> + av_strlcpy(compressor_name,track->enc->codec->name,32);
>
>
> Can't that memset be replaced with compressor_name[0] = '\0' now that
> the result will be null-terminated?
>
avio_w8(pb, strlen(compressor_name));
avio_write(pb, compressor_name, 31);
avio_write() is a mem-function not a str-function so we still want it
zerofilled to the end. Still I can't help but notice that this looks
like a pascal string and wonder if
avio_w8(pb, strlen(compressor_name));
avio_write(pb, compressor_name, strlen(compressor_name));
would be better.
On Tue, May 3, 2011 at 11:46 AM, Justin Ruggles
wrote:
> On 05/03/2011 02:21 PM, Alex Converse wrote:
>
>> @@ -1099,4 +1100,3 @@ void pp_postprocess(const uint8_t * src[3], const int
>> srcStride[3],
>> }
>> }
>> }
>> -
>
>
> stray cosmetic change
>
Fixed
From 116210e7040cfaaba58234945263468e3801e48d Mon Sep 17 00:00:00 2001
From: Alex Converse
Date: Tue, 3 May 2011 11:19:31 -0700
Subject: [PATCH 1/2] Replace strncpy() with av_strlcpy().
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="1"
This is a multi-part message in MIME format.
--1
Content-Type: text/plain; charset=UTF-8; format=fixed
Content-Transfer-Encoding: 8bit
---
ffmpeg.c |3 +--
libavcodec/ac3enc.c | 43 ++-
libavcodec/ass.c |4 ++--
libavformat/movenc.c |2 +-
libavformat/mp3enc.c |3 ++-
libavutil/log.c |3 ++-
libpostproc/postprocess.c |3 ++-
7 files changed, 32 insertions(+), 29 deletions(-)
--1
Content-Type: text/x-patch; name="0001-Replace-strncpy-with-av_strlcpy.patch"
Content-Transfer-Encoding: 8bit
Content-Disposition: attachment; filename="0001-Replace-strncpy-with-av_strlcpy.patch"
diff --git a/ffmpeg.c b/ffmpeg.c
index d728f14..c586811 100644
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -3695,8 +3695,7 @@ static int opt_streamid(const char *opt, const char *arg)
char *p;
char idx_str[16];
-strncpy(idx_str, arg, sizeof(idx_str));
-idx_str[sizeof(idx_str)-1] = '\0';
+av_strlcpy(idx_str, arg, sizeof(idx_str));
p = strchr(idx_str, ':');
if (!p) {
fprintf(stderr,
diff --git a/libavcodec/ac3enc.c b/libavcodec/ac3enc.c
index bf106eb..a35ff29 100644
--- a/libavcodec/ac3enc.c
+++ b/libavcodec/ac3enc.c
@@ -33,6 +33,7 @@
#include "libavutil/audioconvert.h"
#include "libavutil/avassert.h"
+#include "libavutil/avstring.h"
#include "libavutil/crc.h"
#include "libavutil/opt.h"
#include "avcodec.h"
@@ -1578,10 +1579,10 @@ static void dprint_options(AVCodecContext *avctx)
char strbuf[32];
switch (s->bitstream_id) {
-case 6: strncpy(strbuf, "AC-3 (alt syntax)", 32); break;
-case 8: strncpy(strbuf, "AC-3 (standard)", 32);break;
-case 9: strncpy(strbuf, "AC-3 (dnet half-rate)", 32); break;
-case 10: strncpy(strbuf, "AC-3 (dnet quater-rate", 32); break;
+case 6: av_strlcpy(strbuf, "AC-3 (alt syntax)", 32); break;
+case 8: av_strlcpy(strbuf, "AC-3 (standard)", 32);break;
+case 9: av_strlcpy(strbuf, "AC-3 (dnet half-rate)", 32); break;
+case 10: av_strlcpy(strbuf, "AC-3 (dnet quater-rate", 32); break;
default: snprintf(strbuf, 32, "ERROR");
}
av_dlog(avctx, "bitstream_id: %s (%d)\n", strbuf, s->bitstream_id);
@@ -1608,9 +1609,9 @@ static void dprint_options(AVCodecContext *avctx)
if (opt->audio_production_info) {
av_dlog(avctx, "mixing_level: %ddB\n", opt->mixing_level);
switch (opt->room_type) {
-case 0: strncpy(strbuf, "notindicated", 32); break;
-case 1: strncpy(strbuf, "large", 32);break;
-case 2: strncpy(strbuf, "small", 32);break;
+case 0: av_strlcpy(strbuf, "notindicated", 32); break;
+case 1: av_strlcpy(strbuf, "large", 32);break;
+case 2: av_strlcpy(strbuf, "small", 32);break;
default: snprintf(strbuf, 32, "ERROR (%d)", opt->room_type);
}
av_dlog(avctx, "room_type: %s\n", strbuf);
@@ -1622,9 +1623,9 @@ static void dprint_options(AVCodecContext *avctx)
av_dlog(avctx, "dialnorm: %ddB\n", opt->dialogue_level);
if (s->channel_mode == AC3_CHMODE_STEREO) {
switch (opt->dolby_surround_mode) {
-case 0: strncpy(strbuf, "notindicated", 32); break;
-case 1: strncpy(strbuf, "on", 32); break;
-case 2: strncpy(strbuf, "off", 32); break;
+case 0: av_strlcpy(strbuf, "notindicated", 32); break;
+case 1:
Re: [libav-devel] [PATCH 2/2] Ban strncpy() it's too easy to misuse.
On 05/03/2011 02:21 PM, Alex Converse wrote: > --- > libavutil/internal.h |2 ++ > 1 files changed, 2 insertions(+), 0 deletions(-) > > > > 0002-Ban-strncpy-it-s-too-easy-to-misuse.patch > > > diff --git a/libavutil/internal.h b/libavutil/internal.h > index 37bc681..e890ae7 100644 > --- a/libavutil/internal.h > +++ b/libavutil/internal.h > @@ -137,6 +137,8 @@ > #define sprintf sprintf_is_forbidden_due_to_security_issues_use_snprintf > #undef strcat > #define strcat strcat_is_forbidden_due_to_security_issues_use_av_strlcat > +#undef strncpy > +#define strncpy strncpy_is_forbidden_due_to_security_issues_use_av_strlcpy > #undef exit > #define exit exit_is_forbidden > #ifndef LIBAVFORMAT_BUILD Looks like a good idea to me. -Justin ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/2] Replace strncpy() with av_strlcpy().
On 05/03/2011 02:21 PM, Alex Converse wrote: > @@ -1099,4 +1100,3 @@ void pp_postprocess(const uint8_t * src[3], const int > srcStride[3], > } > } > } > - stray cosmetic change ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/2] Replace strncpy() with av_strlcpy().
On 05/03/2011 02:21 PM, Alex Converse wrote: > memset(compressor_name,0,32); > /* FIXME not sure, ISO 14496-1 draft where it shall be set to 0 */ > if (track->mode == MODE_MOV && track->enc->codec && > track->enc->codec->name) > -strncpy(compressor_name,track->enc->codec->name,31); > +av_strlcpy(compressor_name,track->enc->codec->name,32); Can't that memset be replaced with compressor_name[0] = '\0' now that the result will be null-terminated? -Justin ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH 2/2] Ban strncpy() it's too easy to misuse.
--- libavutil/internal.h |2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/libavutil/internal.h b/libavutil/internal.h index 37bc681..e890ae7 100644 --- a/libavutil/internal.h +++ b/libavutil/internal.h @@ -137,6 +137,8 @@ #define sprintf sprintf_is_forbidden_due_to_security_issues_use_snprintf #undef strcat #define strcat strcat_is_forbidden_due_to_security_issues_use_av_strlcat +#undef strncpy +#define strncpy strncpy_is_forbidden_due_to_security_issues_use_av_strlcpy #undef exit #define exit exit_is_forbidden #ifndef LIBAVFORMAT_BUILD ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH 1/2] Replace strncpy() with av_strlcpy().
---
ffmpeg.c |3 +--
libavcodec/ac3enc.c | 43 ++-
libavcodec/ass.c |4 ++--
libavformat/movenc.c |2 +-
libavformat/mp3enc.c |3 ++-
libavutil/log.c |3 ++-
libpostproc/postprocess.c |4 ++--
7 files changed, 32 insertions(+), 30 deletions(-)
diff --git a/ffmpeg.c b/ffmpeg.c
index d728f14..c586811 100644
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -3695,8 +3695,7 @@ static int opt_streamid(const char *opt, const char *arg)
char *p;
char idx_str[16];
-strncpy(idx_str, arg, sizeof(idx_str));
-idx_str[sizeof(idx_str)-1] = '\0';
+av_strlcpy(idx_str, arg, sizeof(idx_str));
p = strchr(idx_str, ':');
if (!p) {
fprintf(stderr,
diff --git a/libavcodec/ac3enc.c b/libavcodec/ac3enc.c
index bf106eb..a35ff29 100644
--- a/libavcodec/ac3enc.c
+++ b/libavcodec/ac3enc.c
@@ -33,6 +33,7 @@
#include "libavutil/audioconvert.h"
#include "libavutil/avassert.h"
+#include "libavutil/avstring.h"
#include "libavutil/crc.h"
#include "libavutil/opt.h"
#include "avcodec.h"
@@ -1578,10 +1579,10 @@ static void dprint_options(AVCodecContext *avctx)
char strbuf[32];
switch (s->bitstream_id) {
-case 6: strncpy(strbuf, "AC-3 (alt syntax)", 32); break;
-case 8: strncpy(strbuf, "AC-3 (standard)", 32);break;
-case 9: strncpy(strbuf, "AC-3 (dnet half-rate)", 32); break;
-case 10: strncpy(strbuf, "AC-3 (dnet quater-rate", 32); break;
+case 6: av_strlcpy(strbuf, "AC-3 (alt syntax)", 32); break;
+case 8: av_strlcpy(strbuf, "AC-3 (standard)", 32);break;
+case 9: av_strlcpy(strbuf, "AC-3 (dnet half-rate)", 32); break;
+case 10: av_strlcpy(strbuf, "AC-3 (dnet quater-rate", 32); break;
default: snprintf(strbuf, 32, "ERROR");
}
av_dlog(avctx, "bitstream_id: %s (%d)\n", strbuf, s->bitstream_id);
@@ -1608,9 +1609,9 @@ static void dprint_options(AVCodecContext *avctx)
if (opt->audio_production_info) {
av_dlog(avctx, "mixing_level: %ddB\n", opt->mixing_level);
switch (opt->room_type) {
-case 0: strncpy(strbuf, "notindicated", 32); break;
-case 1: strncpy(strbuf, "large", 32);break;
-case 2: strncpy(strbuf, "small", 32);break;
+case 0: av_strlcpy(strbuf, "notindicated", 32); break;
+case 1: av_strlcpy(strbuf, "large", 32);break;
+case 2: av_strlcpy(strbuf, "small", 32);break;
default: snprintf(strbuf, 32, "ERROR (%d)", opt->room_type);
}
av_dlog(avctx, "room_type: %s\n", strbuf);
@@ -1622,9 +1623,9 @@ static void dprint_options(AVCodecContext *avctx)
av_dlog(avctx, "dialnorm: %ddB\n", opt->dialogue_level);
if (s->channel_mode == AC3_CHMODE_STEREO) {
switch (opt->dolby_surround_mode) {
-case 0: strncpy(strbuf, "notindicated", 32); break;
-case 1: strncpy(strbuf, "on", 32); break;
-case 2: strncpy(strbuf, "off", 32); break;
+case 0: av_strlcpy(strbuf, "notindicated", 32); break;
+case 1: av_strlcpy(strbuf, "on", 32); break;
+case 2: av_strlcpy(strbuf, "off", 32); break;
default: snprintf(strbuf, 32, "ERROR (%d)", opt->dolby_surround_mode);
}
av_dlog(avctx, "dsur_mode: %s\n", strbuf);
@@ -1636,9 +1637,9 @@ static void dprint_options(AVCodecContext *avctx)
if (s->bitstream_id == 6) {
if (opt->extended_bsi_1) {
switch (opt->preferred_stereo_downmix) {
-case 0: strncpy(strbuf, "notindicated", 32); break;
-case 1: strncpy(strbuf, "ltrt", 32); break;
-case 2: strncpy(strbuf, "loro", 32); break;
+case 0: av_strlcpy(strbuf, "notindicated", 32); break;
+case 1: av_strlcpy(strbuf, "ltrt", 32); break;
+case 2: av_strlcpy(strbuf, "loro", 32); break;
default: snprintf(strbuf, 32, "ERROR (%d)", opt->preferred_stereo_downmix);
}
av_dlog(avctx, "dmix_mode: %s\n", strbuf);
@@ -1655,23 +1656,23 @@ static void dprint_options(AVCodecContext *avctx)
}
if (opt->extended_bsi_2) {
switch (opt->dolby_surround_ex_mode) {
-case 0: strncpy(strbuf, "notindicated", 32); break;
-case 1: strncpy(strbuf, "on", 32); break;
-case 2: strncpy(strbuf, "off", 32); break;
+case 0: av_strlcpy(strbuf, "notindicated", 32); break;
+case 1: av_strlcpy(strbuf, "on", 32); break;
+case 2: av_strlcpy(strbuf, "off", 32); break;
default: snprintf(strbuf, 32, "ERROR (%d)", opt->dolby_surround_ex_mode);
}
av_dlog(avctx, "dsurex_mode: %s\n", strbuf);
switch (opt->dolby_headphone_mode) {
-case 0: strncpy(strb
Re: [libav-devel] [PATCH] Implement fate-rsync target
On Tue, May 03, 2011 at 17:59:53 (CEST), Måns Rullgård wrote: > Reinhard Tartler writes: > >> +fate-rsync: >> +ifdef SAMPLES > > There is already a section conditional on SAMPLES. Including this there > might look nicer. The duplicated target line doesn't matter IMO. Changed > >> +rsync -vaLW rsync://fate-suite.libav.org:/fate-suite/ $(SAMPLES) > > Why the extra : ? removed >> +else >> +@echo "use 'make fate-rsync SAMPLES=/path/to/samples' to sync the fate >> suite" >> +endif >> + >> .PHONY: documentation *test regtest-* alltools check config >> -- >> 1.7.4.1 >> patch pushed -- Gruesse/greetings, Reinhard Tartler, KeyID 945348A4 ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/4] indeo3: Eliminate use of long.
On 04/27/2011 08:10 PM, Aℓex Converse wrote:
> ---
> libavcodec/indeo3.c | 22 +++---
> 1 files changed, 11 insertions(+), 11 deletions(-)
>
>
> --1
> Content-Type: text/x-patch; name="0001-indeo3-Eliminate-use-of-long.patch"
> Content-Transfer-Encoding: 8bit
> Content-Disposition: attachment;
> filename="0001-indeo3-Eliminate-use-of-long.patch"
>
> diff --git a/libavcodec/indeo3.c b/libavcodec/indeo3.c
> index 29a8d4a..30b7085 100644
> --- a/libavcodec/indeo3.c
> +++ b/libavcodec/indeo3.c
> @@ -149,13 +149,13 @@ static av_cold void iv_free_func(Indeo3DecodeContext *s)
> }
>
> struct ustr {
> -long xpos;
> -long ypos;
> -long width;
> -long height;
> -long split_flag;
> -long split_direction;
> -long usl7;
> +int xpos;
> +int ypos;
> +int width;
> +int height;
> +int split_flag;
> +int split_direction;
> +int usl7;
> };
>
>
> @@ -203,12 +203,12 @@ struct ustr {
>
> static void iv_Decode_Chunk(Indeo3DecodeContext *s,
> uint8_t *cur, uint8_t *ref, int width, int height,
> -const uint8_t *buf1, long cb_offset, const uint8_t *hdr,
> +const uint8_t *buf1, int cb_offset, const uint8_t *hdr,
> const uint8_t *buf2, int min_width_160)
> {
> uint8_t bit_buf;
> -unsigned long bit_pos, lv, lv1, lv2;
> -long *width_tbl, width_tbl_arr[10];
> +unsigned int bit_pos, lv, lv1, lv2;
> +int *width_tbl, width_tbl_arr[10];
> const signed char *ref_vectors;
> uint8_t *cur_frm_pos, *ref_frm_pos, *cp, *cp2;
> uint32_t *cur_lp, *ref_lp;
> @@ -982,7 +982,7 @@ static int iv_decode_frame(AVCodecContext *avctx,
> Indeo3DecodeContext *s = avctx->priv_data;
> unsigned int image_width, image_height,
> chroma_width, chroma_height;
> -unsigned long flags, cb_offset, data_size,
> +unsigned int flags, cb_offset, data_size,
>y_offset, v_offset, u_offset, mc_vector_count;
> const uint8_t *hdr_pos, *buf_pos;
Ok.
-Justin
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] Implement fate-rsync target
Reinhard Tartler writes: > +fate-rsync: > +ifdef SAMPLES There is already a section conditional on SAMPLES. Including this there might look nicer. The duplicated target line doesn't matter IMO. > + rsync -vaLW rsync://fate-suite.libav.org:/fate-suite/ $(SAMPLES) Why the extra : ? > +else > + @echo "use 'make fate-rsync SAMPLES=/path/to/samples' to sync the fate > suite" > +endif > + > .PHONY: documentation *test regtest-* alltools check config > -- > 1.7.4.1 > -- Måns Rullgård m...@mansr.com ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] Implement fate-rsync target
Hi, On Tue, May 3, 2011 at 8:40 AM, Reinhard Tartler wrote: > --- > Makefile | 7 +++ > 1 files changed, 7 insertions(+), 0 deletions(-) > > diff --git a/Makefile b/Makefile > index 0849443..8370adc 100644 > --- a/Makefile > +++ b/Makefile > @@ -290,4 +290,11 @@ $(FATE): ffmpeg$(EXESUF) > $(FATE_UTILS:%=tests/%$(HOSTEXESUF)) > fate-list: > @printf '%s\n' $(sort $(FATE)) > > +fate-rsync: > +ifdef SAMPLES > + rsync -vaLW rsync://fate-suite.libav.org:/fate-suite/ $(SAMPLES) > +else > + @echo "use 'make fate-rsync SAMPLES=/path/to/samples' to sync the > fate suite" > +endif > + > .PHONY: documentation *test regtest-* alltools check config Looks good, please apply. Ronald ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH] Implement fate-rsync target
--- Makefile |7 +++ 1 files changed, 7 insertions(+), 0 deletions(-) diff --git a/Makefile b/Makefile index 0849443..8370adc 100644 --- a/Makefile +++ b/Makefile @@ -290,4 +290,11 @@ $(FATE): ffmpeg$(EXESUF) $(FATE_UTILS:%=tests/%$(HOSTEXESUF)) fate-list: @printf '%s\n' $(sort $(FATE)) +fate-rsync: +ifdef SAMPLES + rsync -vaLW rsync://fate-suite.libav.org:/fate-suite/ $(SAMPLES) +else + @echo "use 'make fate-rsync SAMPLES=/path/to/samples' to sync the fate suite" +endif + .PHONY: documentation *test regtest-* alltools check config -- 1.7.4.1 ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 1/3] avfilter: check malloc return values.
On 5/3/11 2:23 PM, Ronald S. Bultje wrote: ret->input_pads = av_malloc(sizeof(AVFilterPad) * ret->input_count); memcpy(ret->input_pads, filter->inputs, sizeof(AVFilterPad) * ret->input_count); ret->inputs = av_mallocz(sizeof(AVFilterLink*) * ret->input_count); +if (!ret->inputs || !ret->input_pads) +goto err; Wrong I guess... lu ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/3] nutenc: check malloc return values.
On 5/3/11 2:23 PM, Ronald S. Bultje wrote:
---
libavformat/nutenc.c |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/libavformat/nutenc.c b/libavformat/nutenc.c
index 4ca761a..85340b1 100644
--- a/libavformat/nutenc.c
+++ b/libavformat/nutenc.c
@@ -588,6 +588,12 @@ static int write_header(AVFormatContext *s){
nut->chapter = av_mallocz(sizeof(ChapterContext)*s->nb_chapters);
nut->time_base= av_mallocz(sizeof(AVRational )*(s->nb_streams +
s->nb_chapters));
+if (!nut->stream || (s->nb_chapters&& !nut->chapter) || !nut->time_base) {
+av_freep(&nut->stream);
+av_freep(&nut->chapter);
+av_freep(&nut->time_base);
+return AVERROR(ENOMEM);
+}
for(i=0; inb_streams; i++){
AVStream *st= s->streams[i];
Seems ok
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH 3/3] mpegvideo: reindent.
---
libavcodec/mpegvideo.c | 227 +++
1 files changed, 111 insertions(+), 116 deletions(-)
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index 6ccc50f..f4de8dd 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -521,100 +521,97 @@ av_cold int MPV_common_init(MpegEncContext *s)
s->flags2= s->avctx->flags2;
if (s->width && s->height) {
+s->mb_width = (s->width + 15) / 16;
+s->mb_stride = s->mb_width + 1;
+s->b8_stride = s->mb_width*2 + 1;
+s->b4_stride = s->mb_width*4 + 1;
+mb_array_size= s->mb_height * s->mb_stride;
+mv_table_size= (s->mb_height+2) * s->mb_stride + 1;
-s->mb_width = (s->width + 15) / 16;
-s->mb_stride = s->mb_width + 1;
-s->b8_stride = s->mb_width*2 + 1;
-s->b4_stride = s->mb_width*4 + 1;
-mb_array_size= s->mb_height * s->mb_stride;
-mv_table_size= (s->mb_height+2) * s->mb_stride + 1;
+/* set chroma shifts */
+avcodec_get_chroma_sub_sample(s->avctx->pix_fmt,&(s->chroma_x_shift),
+ &(s->chroma_y_shift) );
-/* set chroma shifts */
-avcodec_get_chroma_sub_sample(s->avctx->pix_fmt,&(s->chroma_x_shift),
-&(s->chroma_y_shift) );
+/* set default edge pos, will be overriden in decode_header if needed
*/
+s->h_edge_pos= s->mb_width*16;
+s->v_edge_pos= s->mb_height*16;
-/* set default edge pos, will be overriden in decode_header if needed */
-s->h_edge_pos= s->mb_width*16;
-s->v_edge_pos= s->mb_height*16;
+s->mb_num = s->mb_width * s->mb_height;
-s->mb_num = s->mb_width * s->mb_height;
+s->block_wrap[0]=
+s->block_wrap[1]=
+s->block_wrap[2]=
+s->block_wrap[3]= s->b8_stride;
+s->block_wrap[4]=
+s->block_wrap[5]= s->mb_stride;
-s->block_wrap[0]=
-s->block_wrap[1]=
-s->block_wrap[2]=
-s->block_wrap[3]= s->b8_stride;
-s->block_wrap[4]=
-s->block_wrap[5]= s->mb_stride;
+y_size = s->b8_stride * (2 * s->mb_height + 1);
+c_size = s->mb_stride * (s->mb_height + 1);
+yc_size = y_size + 2 * c_size;
-y_size = s->b8_stride * (2 * s->mb_height + 1);
-c_size = s->mb_stride * (s->mb_height + 1);
-yc_size = y_size + 2 * c_size;
+/* convert fourcc to upper case */
+s->codec_tag = ff_toupper4(s->avctx->codec_tag);
-/* convert fourcc to upper case */
-s->codec_tag = ff_toupper4(s->avctx->codec_tag);
+s->stream_codec_tag = ff_toupper4(s->avctx->stream_codec_tag);
-s->stream_codec_tag = ff_toupper4(s->avctx->stream_codec_tag);
+s->avctx->coded_frame= (AVFrame*)&s->current_picture;
-s->avctx->coded_frame= (AVFrame*)&s->current_picture;
-
-FF_ALLOCZ_OR_GOTO(s->avctx, s->mb_index2xy, (s->mb_num+1)*sizeof(int),
fail) //error ressilience code looks cleaner with this
-for(y=0; ymb_height; y++){
-for(x=0; xmb_width; x++){
-s->mb_index2xy[ x + y*s->mb_width ] = x + y*s->mb_stride;
+FF_ALLOCZ_OR_GOTO(s->avctx, s->mb_index2xy, (s->mb_num+1)*sizeof(int),
fail) //error ressilience code looks cleaner with this
+for(y=0; ymb_height; y++){
+for(x=0; xmb_width; x++){
+s->mb_index2xy[ x + y*s->mb_width ] = x + y*s->mb_stride;
+}
}
-}
-s->mb_index2xy[ s->mb_height*s->mb_width ] = (s->mb_height-1)*s->mb_stride
+ s->mb_width; //FIXME really needed?
+s->mb_index2xy[ s->mb_height*s->mb_width ] =
(s->mb_height-1)*s->mb_stride + s->mb_width; //FIXME really needed?
-if (s->encoding) {
-/* Allocate MV tables */
-FF_ALLOCZ_OR_GOTO(s->avctx, s->p_mv_table_base,
mv_table_size * 2 * sizeof(int16_t), fail)
-FF_ALLOCZ_OR_GOTO(s->avctx, s->b_forw_mv_table_base ,
mv_table_size * 2 * sizeof(int16_t), fail)
-FF_ALLOCZ_OR_GOTO(s->avctx, s->b_back_mv_table_base ,
mv_table_size * 2 * sizeof(int16_t), fail)
-FF_ALLOCZ_OR_GOTO(s->avctx, s->b_bidir_forw_mv_table_base ,
mv_table_size * 2 * sizeof(int16_t), fail)
-FF_ALLOCZ_OR_GOTO(s->avctx, s->b_bidir_back_mv_table_base ,
mv_table_size * 2 * sizeof(int16_t), fail)
-FF_ALLOCZ_OR_GOTO(s->avctx, s->b_direct_mv_table_base ,
mv_table_size * 2 * sizeof(int16_t), fail)
-s->p_mv_table = s->p_mv_table_base+ s->mb_stride
+ 1;
-s->b_forw_mv_table = s->b_forw_mv_table_base + s->mb_stride
+ 1;
-s->b_back_mv_table = s->b_back_mv_table_base + s->mb_stride
+ 1;
-s->b_bidir_forw_mv_table= s->b_bidir_forw_mv_table_base + s->mb_stride
+ 1;
-s->b_bidir_back_mv_table= s->b_bidir_back_mv_table_base + s->mb_stride
+ 1;
-s->b_direct_mv_table= s->b_direct_mv_table_base + s->mb_stride
+ 1;
-
-if(s->msmpeg4_version){
-
[libav-devel] [PATCH 1/3] avfilter: check malloc return values.
---
libavfilter/avfilter.c | 22 +-
1 files changed, 21 insertions(+), 1 deletions(-)
diff --git a/libavfilter/avfilter.c b/libavfilter/avfilter.c
index c71c046..98fd2f7 100644
--- a/libavfilter/avfilter.c
+++ b/libavfilter/avfilter.c
@@ -583,18 +583,25 @@ int avfilter_open(AVFilterContext **filter_ctx, AVFilter
*filter, const char *in
return AVERROR(EINVAL);
ret = av_mallocz(sizeof(AVFilterContext));
+if (!ret)
+return AVERROR(ENOMEM);
ret->av_class = &avfilter_class;
ret->filter = filter;
ret->name = inst_name ? av_strdup(inst_name) : NULL;
-if (filter->priv_size)
+if (filter->priv_size) {
ret->priv = av_mallocz(filter->priv_size);
+if (!ret->priv)
+goto err;
+}
ret->input_count = pad_count(filter->inputs);
if (ret->input_count) {
ret->input_pads = av_malloc(sizeof(AVFilterPad) * ret->input_count);
memcpy(ret->input_pads, filter->inputs, sizeof(AVFilterPad) *
ret->input_count);
ret->inputs = av_mallocz(sizeof(AVFilterLink*) *
ret->input_count);
+if (!ret->inputs || !ret->input_pads)
+goto err;
}
ret->output_count = pad_count(filter->outputs);
@@ -602,10 +609,23 @@ int avfilter_open(AVFilterContext **filter_ctx, AVFilter
*filter, const char *in
ret->output_pads = av_malloc(sizeof(AVFilterPad) * ret->output_count);
memcpy(ret->output_pads, filter->outputs, sizeof(AVFilterPad) *
ret->output_count);
ret->outputs = av_mallocz(sizeof(AVFilterLink*) *
ret->output_count);
+if (!ret->outputs || !ret->output_pads)
+goto err;
}
*filter_ctx = ret;
return 0;
+
+err:
+av_freep(&ret->inputs);
+av_freep(&ret->input_pads);
+ret->input_count = 0;
+av_freep(&ret->outputs);
+av_freep(&ret->output_pads);
+ret->output_count = 0;
+av_freep(&ret->priv);
+av_free(ret);
+return AVERROR(ENOMEM);
}
void avfilter_free(AVFilterContext *filter)
--
1.7.2.1
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH 2/3] nutenc: check malloc return values.
---
libavformat/nutenc.c |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/libavformat/nutenc.c b/libavformat/nutenc.c
index 4ca761a..85340b1 100644
--- a/libavformat/nutenc.c
+++ b/libavformat/nutenc.c
@@ -588,6 +588,12 @@ static int write_header(AVFormatContext *s){
nut->chapter = av_mallocz(sizeof(ChapterContext)*s->nb_chapters);
nut->time_base= av_mallocz(sizeof(AVRational )*(s->nb_streams +
s->nb_chapters));
+if (!nut->stream || (s->nb_chapters && !nut->chapter) || !nut->time_base) {
+av_freep(&nut->stream);
+av_freep(&nut->chapter);
+av_freep(&nut->time_base);
+return AVERROR(ENOMEM);
+}
for(i=0; inb_streams; i++){
AVStream *st= s->streams[i];
--
1.7.2.1
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/2] vc1: make overlap filter for I-frames bit-exact.
On Tue, May 03, 2011 at 04:52:18AM -0700, Ronald S. Bultje wrote:
> Hi,
>
> On Mon, May 2, 2011 at 9:36 PM, Kostya wrote:
> > On Mon, May 02, 2011 at 10:12:34PM -0400, Ronald S. Bultje wrote:
> >> ---
> >> libavcodec/vc1.h | 2 +
> >> libavcodec/vc1dec.c | 279
> >> +++
> >> libavcodec/vc1dsp.c | 54 ++
> >> libavcodec/vc1dsp.h | 6 +-
> >> 4 files changed, 296 insertions(+), 45 deletions(-)
> >>
> > [...]
> >> @@ -672,6 +724,8 @@ av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
> >> dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_c;
> >> dsp->vc1_h_overlap = vc1_h_overlap_c;
> >> dsp->vc1_v_overlap = vc1_v_overlap_c;
> >> + dsp->vc1_h_s_overlap = vc1_h_s_overlap_c;
> >> + dsp->vc1_v_s_overlap = vc1_v_s_overlap_c;
> >> dsp->vc1_v_loop_filter4 = vc1_v_loop_filter4_c;
> >> dsp->vc1_h_loop_filter4 = vc1_h_loop_filter4_c;
> >> dsp->vc1_v_loop_filter8 = vc1_v_loop_filter8_c;
> >> diff --git a/libavcodec/vc1dsp.h b/libavcodec/vc1dsp.h
> >> index 7b1ae10..e1b6ba0 100644
> >> --- a/libavcodec/vc1dsp.h
> >> +++ b/libavcodec/vc1dsp.h
> >> @@ -40,8 +40,10 @@ typedef struct VC1DSPContext {
> >> void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM
> >> *block);
> >> void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM
> >> *block);
> >> void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM
> >> *block);
> >> - void (*vc1_v_overlap)(uint8_t* src, int stride);
> >> - void (*vc1_h_overlap)(uint8_t* src, int stride);
> >> + void (*vc1_v_overlap)(uint8_t *src, int stride);
> >> + void (*vc1_h_overlap)(uint8_t *src, int stride);
> >> + void (*vc1_v_s_overlap)(DCTELEM *top, DCTELEM *bottom);
> >> + void (*vc1_h_s_overlap)(DCTELEM *left, DCTELEM *right);
> >> void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
> >> void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
> >> void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);
> >> --
> >
> > Almost ok but I have one question - is vc1_[hv]_overlap used anymore (I
> > don't
> > think they should)? If not then just replace old functions with new ones but
> > leave old names.
>
> For archive purposes: no, I only changed adv. profile I frames.
> main/simple profile I frames and P frames still use the old overlap
> filter. Once they're all bitexact/changed/fixed, I'll remove the 8bit
> versions in a separate patch.
patch ok then
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/2] vc1: make overlap filter for I-frames bit-exact.
Hi,
On Mon, May 2, 2011 at 9:36 PM, Kostya wrote:
> On Mon, May 02, 2011 at 10:12:34PM -0400, Ronald S. Bultje wrote:
>> ---
>> libavcodec/vc1.h | 2 +
>> libavcodec/vc1dec.c | 279
>> +++
>> libavcodec/vc1dsp.c | 54 ++
>> libavcodec/vc1dsp.h | 6 +-
>> 4 files changed, 296 insertions(+), 45 deletions(-)
>>
> [...]
>> @@ -672,6 +724,8 @@ av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
>> dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_c;
>> dsp->vc1_h_overlap = vc1_h_overlap_c;
>> dsp->vc1_v_overlap = vc1_v_overlap_c;
>> + dsp->vc1_h_s_overlap = vc1_h_s_overlap_c;
>> + dsp->vc1_v_s_overlap = vc1_v_s_overlap_c;
>> dsp->vc1_v_loop_filter4 = vc1_v_loop_filter4_c;
>> dsp->vc1_h_loop_filter4 = vc1_h_loop_filter4_c;
>> dsp->vc1_v_loop_filter8 = vc1_v_loop_filter8_c;
>> diff --git a/libavcodec/vc1dsp.h b/libavcodec/vc1dsp.h
>> index 7b1ae10..e1b6ba0 100644
>> --- a/libavcodec/vc1dsp.h
>> +++ b/libavcodec/vc1dsp.h
>> @@ -40,8 +40,10 @@ typedef struct VC1DSPContext {
>> void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM
>> *block);
>> void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM
>> *block);
>> void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM
>> *block);
>> - void (*vc1_v_overlap)(uint8_t* src, int stride);
>> - void (*vc1_h_overlap)(uint8_t* src, int stride);
>> + void (*vc1_v_overlap)(uint8_t *src, int stride);
>> + void (*vc1_h_overlap)(uint8_t *src, int stride);
>> + void (*vc1_v_s_overlap)(DCTELEM *top, DCTELEM *bottom);
>> + void (*vc1_h_s_overlap)(DCTELEM *left, DCTELEM *right);
>> void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
>> void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
>> void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);
>> --
>
> Almost ok but I have one question - is vc1_[hv]_overlap used anymore (I don't
> think they should)? If not then just replace old functions with new ones but
> leave old names.
For archive purposes: no, I only changed adv. profile I frames.
main/simple profile I frames and P frames still use the old overlap
filter. Once they're all bitexact/changed/fixed, I'll remove the 8bit
versions in a separate patch.
Ronald
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] Make ffmpeg support generic data stream
On Thu, Apr 14, 2011 at 01:32:36PM +0200, Luca Barbato wrote: > The patch is the first step to support -dcodec copy > --- > cmdutils.h |1 + > ffmpeg.c | 85 +++ > libavformat/riff.c |1 + > 3 files changed, 80 insertions(+), 7 deletions(-) yes, looks fine to me ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
