If we are given a non-render node, try to find the matching render node and
fail if that isn't possible.
libva will not accept a non-render device which is not DRM master, because
it requires legacy DRM authentication to succeed in that case:
<https://github.com/intel/libva/blob/master/va/drm/va_drm.c#L68-L75>. This
is annoying for kmsgrab because in most recording situations DRM master is
already held by something else (such as a windowing system), leading to
device derivation not working and forcing the user to create the target
VAAPI device separately.
Add function pointer field in vaapi_profile_map[], set profile_parser
for HEVC_REXT to find the exact va_profile.
Also add format map support.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Add vaapi_parse_rext_profile and use profile constraint flags to
determine the exact va_profile for HEVC_REXT.
If profile mismatch is allowed, select Main profile by default.
Add build object in Makefile for h265_profile_level dependency.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Parse all the constraint flags according to ITU-T Rec. H.265 (02/2018).
They have to be passed to hw decoders to determine the exact profile for Range
Extension HEVC.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Extend ParameterBufferHEVC to ParameterBufferHEVCExtension for both
VAPicture and VASlice.
Pass Range Extension flags to support the decode for HEVC REXT.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
VA_RT_FORMAT describes the desired sampling format for surface.
When creating surface, VA_RT_FORMAT will be used firstly to choose
the expected fourcc/media_format for the surface. And the fourcc
will be revised by the value of VASurfaceAttribPixelFormat.
Add vaapi_format_map support for new pixel_format Y210.
This is fundamental for both VA-API and QSV.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Add swscale input support for Y210LE, output support and fate
test could be added later if there is requirement for software
CSC to this packed format.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Otherwise the user is not able to override the SRT API connect timeout above 5
sec without also setting the timeout option.
Signed-off-by: Marton Balint <cus@passwd.hu>
Return os error code if available, check for both SRT_EASYNCRCV and
SRT_EASYNCSND when transforming them to EAGAIN and only display error if it is
not EAGAIN.
Signed-off-by: Marton Balint <cus@passwd.hu>
After f8990c5f41 we properly set non-blocking
mode which makes the connect() call return always 0 even if no connection can
be established.
Fix this by always doing a poll after calling connect(). Also there was some
leftover copy paste code which checks for various errors which are simply not
possible with SRT.
Signed-off-by: Marton Balint <cus@passwd.hu>
rw_timeout is the generic URLcontext option, not the protocol specific timeout
option, also ?rw_timeout never worked because ?timeout was parsed instead.
Signed-off-by: Marton Balint <cus@passwd.hu>
This ensures it's written at the beginning of a segment in non streaming mode
when segment duration differs from fragment duration.
Signed-off-by: James Almer <jamrial@gmail.com>
Such errors may make sense for specific formats, but general parsing
logic shouldn't be treating these as errors regardless of the error
recognition mode.
Fixes loading of the following wave when using -err_detect explode:
https://cs.chromium.org/chromium/src/third_party/blink/web_tests/external/wpt/webaudio/resources/4ch-440.wav
Signed-off-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If an audio stream is present, the Sega FILM muxer checks for its
compability with the container during init, so that the very same check
needn't be repeated during writing the trailer.
Essentially the same is true for the presence of a video stream: It has
already been checked during init. Furthermore, after the check for the
presence of a video stream succeeded, a pointer is set to point to the
video stream. Yet said pointer (which was NULL before) will be
derefenced anyway regardless of the result of the check. Coverity thus
complained about this in CID 1434155 and removing this pointless check
will also fix this issue.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: left shift of 233 by 24 places cannot be represented in type 'int'
Fixes: 20736/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PCM_LXF_fuzzer-4829212685107200
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array read
Fixes: 20742/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CDTOONS_fuzzer-5738148607033344
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
No testcases, found by code review when debuging issue found by oss-fuzz
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Commit 36e156bef0 ("avformat/spdifenc: fix handling of large TrueHD
frame") added an obviously incorrect bitshift that caused incorrect
samples-per-frame calculation for TrueHD streams over 48kHz.
Fix that.
The TrueHD IEC 61937 encapsulation code uses a very naive method of
always inserting 24 TrueHD frames evenly in a MAT frame. This does not
work for larger frames as they may exceed the size of 1/24th of a MAT
frame.
To fix that, use the input_timing field in the TrueHD frame to determine
the proper position of the TrueHD frame in the MAT frame. That field is
basically a dts field, telling the time to feed this frame to the
decoder in sample count units.
This can cause a TrueHD frame to be split between two MAT frames, so a
second concatenation hd_buf is added, alternating with the first buffer.
Large frames are preceded by smaller frames that have input_timing
values that cause the frames to be sent out faster than the nominal rate
(i.e. increasing decoder latency, long decoder buffer), allowing the
larger frames to then be sent out slower than the nominal rate as the
decoder has enough data buffered to keep it busy.
This is preparation for adding a second hd_buf in a followup commit.
Also, slightly improve the comments for hd_buf_x members to clarify
which ones are actually used and kept up-to-date depending on which
codec is being muxed.
