> -Original Message-
> From: ffmpeg-devel On Behalf Of
> xuju...@sjtu.edu.cn
> Sent: 2020年9月19日 23:24
> To: ffmpeg-devel@ffmpeg.org
> Cc: xuju...@sjtu.edu.cn
> Subject: [FFmpeg-devel] [PATCH] dnn_backend_native_layer_conv2d.c: fix bug
> of loop boundary in single thread mode.
>
> From:
Also return early when encountering fatal error.
Signed-off-by: Paul B Mahol
---
libavcodec/cfhd.c | 50 ++-
1 file changed, 28 insertions(+), 22 deletions(-)
diff --git a/libavcodec/cfhd.c b/libavcodec/cfhd.c
index ea35f03869..7b4ba4c3dd 100644
--- a
On Sat, Sep 19, 2020 at 11:46:50PM +0200, Marton Balint wrote:
>
>
> On Sat, 19 Sep 2020, Paul B Mahol wrote:
>
> > On Thu, Sep 17, 2020 at 12:31:06PM +0200, Paul B Mahol wrote:
> > > This removes big CPU overhead for demuxing chained ogg streams.
> > >
> > > Signed-off-by: Paul B Mahol
> > >
On Sat, 19 Sep 2020, Paul B Mahol wrote:
On Thu, Sep 17, 2020 at 12:31:06PM +0200, Paul B Mahol wrote:
This removes big CPU overhead for demuxing chained ogg streams.
Signed-off-by: Paul B Mahol
---
libavformat/aviobuf.c | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
w
On Sat, Sep 19, 2020 at 07:47:49PM +0200, Paul B Mahol wrote:
> On Sat, Sep 19, 2020 at 7:45 PM Michael Niedermayer
> wrote:
>
> > On Mon, Aug 31, 2020 at 07:42:20PM +0200, Paul B Mahol wrote:
> > > On 8/31/20, Michael Niedermayer wrote:
> > > > On Mon, Aug 31, 2020 at 11:14:07AM +0200, Paul B M
On Sat, Sep 19, 2020 at 09:31:08PM +0200, Michael Niedermayer wrote:
> Fixes: Integer overflow (no testcase)
>
> Signed-off-by: Michael Niedermayer
> ---
> libavcodec/ansi.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
ok
___
ffmpeg-de
Fixes: Timeout (too long -> 1.3 sec)
Fixes:
24358/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SONIC_fuzzer-5107284099989504
Found-by: continuous fuzzing process
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer
---
libavcodec/sonic.c | 3 ++
Fixes: signed integer overflow: -190760 * 11585 cannot be represented in type
'int'
Fixes:
25471/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VP9_fuzzer-5743354917421056
Found-by: continuous fuzzing process
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Micha
Fixes: signed integer overflow: 29 + 2147483640 cannot be represented in type
'int'
Fixes:
25413/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-5697909331591168
Found-by: continuous fuzzing process
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Mich
Fixes: Timeout (minutes to less than 1sec)
Fixes:
25682/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ANSI_fuzzer-6320712032452608
Found-by: continuous fuzzing process
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer
---
libavcodec/ansi.c |
Fixes: Integer overflow (no testcase)
Signed-off-by: Michael Niedermayer
---
libavcodec/ansi.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/libavcodec/ansi.c b/libavcodec/ansi.c
index 516d07db69..272185230d 100644
--- a/libavcodec/ansi.c
+++ b/libavcodec/ansi.c
@@ -431,7
On Thu, Sep 17, 2020 at 12:31:06PM +0200, Paul B Mahol wrote:
> This removes big CPU overhead for demuxing chained ogg streams.
>
> Signed-off-by: Paul B Mahol
> ---
> libavformat/aviobuf.c | 10 +-
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
will apply soon.
__
They need it because they make use of SEI parsing code.
Signed-off-by: Andreas Rheinhardt
---
configure | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/configure b/configure
index 5d68695192..70740a7bc7 100755
--- a/configure
+++ b/configure
@@ -3143,8 +3143,8 @@ wmv3_cry
On Sat, Sep 19, 2020 at 7:45 PM Michael Niedermayer
wrote:
> On Mon, Aug 31, 2020 at 07:42:20PM +0200, Paul B Mahol wrote:
> > On 8/31/20, Michael Niedermayer wrote:
> > > On Mon, Aug 31, 2020 at 11:14:07AM +0200, Paul B Mahol wrote:
> > >> On 8/31/20, Michael Niedermayer wrote:
> > >> > This i
On Mon, Aug 31, 2020 at 07:42:20PM +0200, Paul B Mahol wrote:
> On 8/31/20, Michael Niedermayer wrote:
> > On Mon, Aug 31, 2020 at 11:14:07AM +0200, Paul B Mahol wrote:
> >> On 8/31/20, Michael Niedermayer wrote:
> >> > This is based on the encoder and a small number of CFHD sample files
> >> > I
get_content_url() allocates two buffers for temporary strings and when
one of them couldn't be allocated, it simply returns, although one of
the two allocations could have succeeded and would leak in this
scenario. This can be fixed by avoiding one of the temporary buffers.
Signed-off-by: Andreas
Signed-off-by: Andreas Rheinhardt
---
There is still stuff left to fix in this demuxer after this patchset:
1. resolve_content_path() is in bad shape; i.e. it can try to read
str[-1].
2. One can get a crash in get_current_fragment() (or rather in
ff_dash_fill_tmpl_params()) when one has a represen
1. Perform the necessary reindentations after the last few commits.
2. Adapt switches to the ordinary indentation style.
3. Now that the effective lifetimes of the variables containing
the freshly allocated strings used when parsing the representation
are disjoint, the variables can be replaced by
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 90 +--
1 file changed, 45 insertions(+), 45 deletions(-)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index 3b6ae29b1b..71c6b5cc1a 100644
--- a/libavformat/dashdec.c
+++ b/libavfo
Up until now, the DASH demuxer used av_dynarray_add() to add
audio/video/subtitles representations to arrays. Yet av_dynarray_add()
frees the array upon failure, leading to leaks of its elements;
furthermore, the element to be added leaks, too.
This has been fixed by using av_dynarray_add_nofree()
This commit removes two always-true checks as well as a dead default
case of a switch. The check when parsing manifests is always true,
because we now jump to the cleaning code in case the format of the
representation is unknown. The default case of the switch is dead,
because the type of the repre
This allows to reduce the level of indentation for parsing the supported
representations (audio, video and subtitles). It also allows to avoid
some allocations and frees for unsupported representations.
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 30 +++-
open_url() in the DASH as well in the hls demuxer share a common bug:
They modify an AVDictionary (i.e. set a new entry) given to them as
AVDictionary *, yet if this new entry leads to reallocation and
relocation of the AVDictionary, the caller's pointer will become
dangling, leading to use-after-f
Signed-off-by: Andreas Rheinhardt
---
Making sure that it's safe to call this demuxer's read_close function
upon read_header failure was of course the initial rationale behind this
patchset.
libavformat/dashdec.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/libavformat/dashdec.c b/libav
These languages are normally freed after having been added as metadata
to their respective AVStreams. Yet if one never reaches said point, they
leak. This can happen as a result of an error when reading the header or
as a result of refreshing the manifests.
Signed-off-by: Andreas Rheinhardt
---
Signed-off-by: Andreas Rheinhardt
---
What do we need this program for anyway? What sense does it make to
create a single program containing all streams?
libavformat/dashdec.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index 55212661be..4f87
The DASH demuxer currently extracts several strings at once from an xml
document before processing them one by one; these strings are allocated,
stored in local variables and need to be freed by the demuxer itself.
So if an error happens when processing one of them, all strings need to
be freed bef
It is always zero. Also remove other unused elements.
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 41 +
1 file changed, 13 insertions(+), 28 deletions(-)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index 311a19e425..82cb33
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index 8b421a4493..3b83075e7a 100644
--- a/libavformat/dashdec.c
+++ b/libavformat/dashdec.c
@@ -2274,6 +2274,7 @@ static int dash_close(AVF
If parsing a representation fails, it is not added to the list of
representations and is therefore not freed in dash_close(); it therefore
leaked in most error paths in parse_manifest_representation() (some
error paths had (incomplete) code for freeing). This commit fixes
freeing the representation
Just postpone the allocation of the dict until it is really needed
(after the checks that can fail).
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index 3b83075e7a
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 15 ---
1 file changed, 4 insertions(+), 11 deletions(-)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index 3f51e079cf..8b421a4493 100644
--- a/libavformat/dashdec.c
+++ b/libavformat/dashdec.c
@@ -616,9 +616
The code in question seems to have been copied from about 70 lines
above; yet the code here is only executed if some of the variables
(namely representation_segmenttemplate_node and fragment_template_node)
are NULL, so it makes no sense to check them for a child element.
Also remove a redundant re
Signed-off-by: Andreas Rheinhardt
---
One could also error out instead.
libavformat/dashdec.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavformat/dashdec.c b/libavformat/dashdec.c
index d84da16dbd..3f51e079cf 100644
--- a/libavformat/dashdec.c
+++ b/libavformat/dashdec.c
@@ -614,6 +6
This currently doesn't cause any trouble, because the only caller did
not clean up the representation upon error at all; but fixing this is
a prerequisite for doing so.
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a
When using one of the AV_DICT_DONT_STRDUP_KEY/VAL flags, av_dict_set()
already frees the key/value on error, so that freeing it again would
lead to a double free.
Signed-off-by: Andreas Rheinhardt
---
libavformat/dashdec.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/li
From: Xu Jun
Before patch, fate test for dnn may fail in some Windows environment
while succeed in my Linux. The bug was caused by a wrong loop boundary.
Win10 and Linux seems to have different explaination for that.After
patch, fate test succeed in my windows mingw 64-bit.
Signed-off-by: Xu Jun
Hello -
I'm looking to *hire a skilled developer* to aid in some advanced FFmpeg
commands pertaining to encoding and broadcasting. Please shoot me an email
and I'd be happy to go into details.
My company, *Inlet Media*, is a US based company.
Thank you and look forward to hearing from someone.
Andreas Rheinhardt:
> The RealMedia demuxer's read_header function initially initializes ret,
> the variable designated for the return variable to -1. Afterwards, chunks
> of the file are parsed in a loop until an error happens or until the actual
> frame data is encountered. If the first function
As it was brought up that the current documentation leaves things
as specific to YCbCr only, ICtCp and RGB are now mentioned.
Additionally, the specifications on which these definitions of
narrow and full range are defined are mentioned.
This way, the documentation of AVColorRange should now match
On Sat, Sep 19, 2020 at 12:47 AM James Almer wrote:
>
> On 9/18/2020 6:05 PM, Jan Ekström wrote:
> > As it was brought up that the current documentation leaves things
> > as specific to YCbCr only, ICtCp and RGB are now mentioned.
> > Additionally, the specifications on which these definitions of
Michael Niedermayer:
> On Sun, Jul 19, 2020 at 07:55:24PM +0200, Andreas Rheinhardt wrote:
>> James Almer:
>>> On 7/19/2020 2:42 PM, Michael Niedermayer wrote:
Fixes: memleak
Fixes:
23660/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-6007508031504384
Found-by: conti
On Sun, Jul 19, 2020 at 07:55:24PM +0200, Andreas Rheinhardt wrote:
> James Almer:
> > On 7/19/2020 2:42 PM, Michael Niedermayer wrote:
> >> Fixes: memleak
> >> Fixes:
> >> 23660/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-6007508031504384
> >>
> >> Found-by: continuous fuzzing process
>
43 matches
Mail list logo