[FFmpeg-devel] [PATCH 1/2] tools/target_dec_fuzzer: Adjust threshold for smacker audio

[FFmpeg-devel] [PATCH 1/2] tools/target_dec_fuzzer: Adjust threshold for smacker audio

[Michael Niedermayer]

Fixes: Timeout
Fixes: 47043/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SMACKAUD_fuzzer-4824799337119744

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
---
 tools/target_dec_fuzzer.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tools/target_dec_fuzzer.c b/tools/target_dec_fuzzer.c
index 288aa63313..0eddeaf69a 100644
--- a/tools/target_dec_fuzzer.c
+++ b/tools/target_dec_fuzzer.c
@@ -202,6 +202,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
     case AV_CODEC_ID_SANM:        maxpixels  /= 16;    break;
     case AV_CODEC_ID_SCPR:        maxpixels  /= 32;    break;
     case AV_CODEC_ID_SCREENPRESSO:maxpixels  /= 64;    break;
+    case AV_CODEC_ID_SMACKAUDIO:  maxsamples /= 4096;  break;
     case AV_CODEC_ID_SMACKVIDEO:  maxpixels  /= 64;    break;
     case AV_CODEC_ID_SNOW:        maxpixels  /= 128;   break;
     case AV_CODEC_ID_TARGA:       maxpixels  /= 128;   break;
-- 
2.17.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

[FFmpeg-devel] [PATCH 2/2] avcodec/smacker: Optimize constant 16bit audio output

[Michael Niedermayer]

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
---
 libavcodec/smacker.c | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/libavcodec/smacker.c b/libavcodec/smacker.c
index 179c70f1ee..5d94a54179 100644
--- a/libavcodec/smacker.c
+++ b/libavcodec/smacker.c
@@ -601,7 +601,7 @@ static int smka_decode_frame(AVCodecContext *avctx, AVFrame *frame,
     int i, res, ret;
     int unp_size;
     int bits, stereo;
-    unsigned pred[2], val;
+    unsigned pred[2], val, val2;
 
     if (buf_size <= 4) {
         av_log(avctx, AV_LOG_ERROR, "packet is too small\n");
@@ -672,7 +672,11 @@ static int smka_decode_frame(AVCodecContext *avctx, AVFrame *frame,
             pred[i] = av_bswap16(get_bits(&gb, 16));
         for(i = 0; i <= stereo; i++)
             *samples++ = pred[i];
-        for(; i < unp_size / 2; i++) {
+        unp_size /= 2;
+
+        if (vlc[0       ].table || vlc[         1].table ||
+            vlc[2*stereo].table || vlc[2*stereo+1].table) {
+        for(; i < unp_size ; i++) {
             unsigned idx = 2 * (i & stereo);
             if (get_bits_left(&gb) < 0) {
                 ret = AVERROR_INVALIDDATA;
@@ -691,6 +695,22 @@ static int smka_decode_frame(AVCodecContext *avctx, AVFrame *frame,
             pred[idx / 2] += val;
             *samples++ = pred[idx / 2];
         }
+        } else if (stereo) {
+            val  = 256*values[1] + values[0];
+            val2 = 256*values[3] + values[2];
+            for(; i < unp_size; i+=2) {
+                pred[0] += val;
+                pred[1] += val2;
+                *samples++ = pred[0];
+                *samples++ = pred[1];
+            }
+        } else {
+            val = 256*values[1] + values[0];
+            for(; i < unp_size; i++) {
+                pred[0] += val;
+                *samples++ = pred[0];
+            }
+        }
     } else { //8-bit data
         for(i = stereo; i >= 0; i--)
             pred[i] = get_bits(&gb, 8);
-- 
2.17.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [PATCH 2/2] avcodec/smacker: Optimize constant 16bit audio output

[Tomas Härdin]

tis 2022-05-03 klockan 18:30 +0200 skrev Michael Niedermayer:
> 
> +        } else if (stereo) {
> +            val  = 256*values[1] + values[0];
> +            val2 = 256*values[3] + values[2];
> +            for(; i < unp_size; i+=2) {
> +                pred[0] += val;
> +                pred[1] += val2;
> +                *samples++ = pred[0];
> +                *samples++ = pred[1];
> +            }
> +        } else {
> +            val = 256*values[1] + values[0];
> +            for(; i < unp_size; i++) {
> +                pred[0] += val;
> +                *samples++ = pred[0];
> +            }
> +        }

Got any numbers on how much faster this is? Just out of curiosity

Probably want to follow this up with a reindent patch

/Tomas

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [PATCH 2/2] avcodec/smacker: Optimize constant 16bit audio output

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 1227 bytes --]

On Wed, May 04, 2022 at 11:39:54AM +0200, Tomas Härdin wrote:
> tis 2022-05-03 klockan 18:30 +0200 skrev Michael Niedermayer:
> > 
> > +        } else if (stereo) {
> > +            val  = 256*values[1] + values[0];
> > +            val2 = 256*values[3] + values[2];
> > +            for(; i < unp_size; i+=2) {
> > +                pred[0] += val;
> > +                pred[1] += val2;
> > +                *samples++ = pred[0];
> > +                *samples++ = pred[1];
> > +            }
> > +        } else {
> > +            val = 256*values[1] + values[0];
> > +            for(; i < unp_size; i++) {
> > +                pred[0] += val;
> > +                *samples++ = pred[0];
> > +            }
> > +        }
> 
> Got any numbers on how much faster this is? Just out of curiosity

With the fuzzed sample:
before:
3263902379 decicycles in ABBB,     128 runs,      0 skips

after:
398977744 decicycles in ABBB,    1024 runs,      0 skips

the first times out after 128 runs which is why the runs differ

thx

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Opposition brings concord. Out of discord comes the fairest harmony.
-- Heraclitus

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [PATCH 2/2] avcodec/smacker: Optimize constant 16bit audio output

[Paul B Mahol]

lgtm
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [PATCH 2/2] avcodec/smacker: Optimize constant 16bit audio output

[Tomas Härdin]

ons 2022-07-06 klockan 19:11 +0200 skrev Michael Niedermayer:
> On Wed, May 04, 2022 at 11:39:54AM +0200, Tomas Härdin wrote:
> > tis 2022-05-03 klockan 18:30 +0200 skrev Michael Niedermayer:
> > > 
> > > +        } else if (stereo) {
> > > +            val  = 256*values[1] + values[0];
> > > +            val2 = 256*values[3] + values[2];
> > > +            for(; i < unp_size; i+=2) {
> > > +                pred[0] += val;
> > > +                pred[1] += val2;
> > > +                *samples++ = pred[0];
> > > +                *samples++ = pred[1];
> > > +            }
> > > +        } else {
> > > +            val = 256*values[1] + values[0];
> > > +            for(; i < unp_size; i++) {
> > > +                pred[0] += val;
> > > +                *samples++ = pred[0];
> > > +            }
> > > +        }
> > 
> > Got any numbers on how much faster this is? Just out of curiosity
> 
> With the fuzzed sample:
> before:
> 3263902379 decicycles in ABBB,     128 runs,      0 skips
> 
> after:
> 398977744 decicycles in ABBB,    1024 runs,      0 skips
> 
> the first times out after 128 runs which is why the runs differ

Cool. Well, looks good

/Tomas

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [PATCH 2/2] avcodec/smacker: Optimize constant 16bit audio output

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 261 bytes --]

On Wed, Jul 06, 2022 at 07:18:45PM +0200, Paul B Mahol wrote:
> lgtm

will apply

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I have often repented speaking, but never of holding my tongue.
-- Xenocrates

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [PATCH 1/2] tools/target_dec_fuzzer: Adjust threshold for smacker audio

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 656 bytes --]

On Tue, May 03, 2022 at 06:30:58PM +0200, Michael Niedermayer wrote:
> Fixes: Timeout
> Fixes: 47043/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SMACKAUD_fuzzer-4824799337119744
> 
> Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
> ---
>  tools/target_dec_fuzzer.c | 1 +
>  1 file changed, 1 insertion(+)

will apply

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Elect your leaders based on what they did after the last election, not
based on what they say before an election.


[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".