The previous implementation of opengl-cb kept only latest flipped frame.
This can cause massive frame drops because rendering is done asynchronously
and only the latest frame can be rendered.
This commit introduces frame queue and releated options to opengl-cb.
frame-queue-size: the maximum size of frame queue (1-100, default: 1)
frame-drop-mode: behavior when frame queue is full (pop, clear, default: pop)
The frame queue holds delayed frames and drops frames if the frame queue is
overflowed with next method:
'pop' mode: drops all the oldest frames overflown.
'clear' mode: drops all frames in queue and clear it.
With default options(frame-queue-size=1:frame-drop-mode=pop),
opengl-cb behaves in the same way as previous implementation effectively.
For frame-queue-size > 1, opengl-cb tries to calls update() without waiting
next flip_page() in order to consume queued frames.
Signed-off-by: wm4 <wm4@nowhere>
vo_opengl was crashing since f811348d because it passed NULL for the
events parameter to vo_control. Normally the parameter should not be
NULL, so add a hack to account for this. In particular, we should
handle the events that are returned. For the call in preinit() we
skip this, but it most likely has no meaning anyway, because in this
stage no window is visible yet.
This was always supposed to work.
Just add the option declaration. Normally I'm not a fan of duplicating
such things, but in this case it's (still) harmless.
mpv can be built natively on a Windows machine using MSYS2. Add detailed
instructions on how to build and merge them with the existing
instructions for cross-compilation.
This one avoids use of a FBO. It's less flexible, because it uses works
around the whole QML rendering API. It seems to be the only way to get
OpenGL rendering without any indirections, though.
Parts of this example were insipired by Qt's "Squircle" example.
Also add a README file with a short description of each example, to
reduce the initial confusing.
This affects OSX, where memory profiles are updated e.g. on fullscreen
switches. The profile most likely doesn't change, but the LUT will
be generated and reloaded anyway.
Somewhat of a regression from commit f811348.
Fixes#1439.
If icc-path is set, but the thing is replaced with a memory profile,
then p->icc_path would point to deallocated memory.
Also, the NULL checks are unnecessary.
This used to be required to workaround PulseAudio bugs. Even later, when
the bugs were (partially?) fixed in PulseAudio, I had the feeling the
hacks gave better behavior. On the other hand, I couldn't actually
reproduce any bad behavior without the hacks lately. On top of this, it
seems our hacks sometimes perform much worse than PulseAudio's native
implementation (see #1430).
So disable the hacks by default, but still leave the code and the option
in case it still helps somewhere. Also, being able to blame PulseAudio's
code by using its native API is much easier than trying to debug our own
(mplayer2-derived) hacks.
glXGetProcAddress() is outdated, and as far as I know doesn't give all
the guarantees the "new" ARB function gives.
Probably doesn't matter too much, because until now it always appeared
to work. On the other hand, since this function is (bogusly) used only
on the gl3 code path, it could have happened that users hit this, and
just reverted to vo_opengl_old instead.
The function terminal_in_background() reports whether the player was
backgrounded. In this case, we don't want to annoy the user by still
printing the status to stderr. If no terminal interaction is assumed,
this mechanism is disabled, and stderr is always used. The read_terminal
variable signals this case.
Oddly, just redirecting stderr will disable output to stderr, because
the background check with tcgetpgrp() is done on stderr, but
read_terminal is still true (because that one depends on stdin and
stdout).
Explicitly disable this mechanism if --no-input-terminal is used by
setting read_terminal to true only if terminal input is actually
initialized.
When the given mp_image_params does not match with that of gl_video,
gl_video_config() always calls uninit_video() but calls init_video()
only if valid format is given.
Since uninit_video() does not change image_params of gl_video,
when the same params as the previous one is given to gl_video_config()
after gl_video is unitialized with invalid format, gl_video_config()
never calls init_video().
To prevent this, invalidate image_params of gl_video in uninit_video().
Previously we just forced loading a profile from file, but that has poor
integration for querying the OS / display server for an ICC profile, and
generating profiles on the fly (which we might use in the future for creating
preset 3dluts).
Also changed the previous icc-profile-auto code to use this mechanism, and
moved gl_lcms to be an opaque type with state instead of just providing pure
functions.
This causes more constant reading from the DVB device,
and seems not to hurt the stream-controls.
As the device cache seems to fill up when not
reading from it for several seconds, reading
into the cache fixes desync-issues after
the initial analyzeduration of the demuxer.
Was already possible before by injecting the magic PID
8192 into channels.conf, the flag makes this much more
useable and we also have it documented.
Useful not only for debugging, but also for incomplete
channels.conf (mplayer format...), multi-channel
recording, or channels which do dynamic PID switchng.
full-transponder is also useful for channels which switch PIDs on-the-fly.
ffmpeg can handle this, but it needs the full stream with all PIDs.
There is the STD which contains service-names
(ffmpeg can understand it),
and the EIT which contains the EPG
(ffmpeg does not parse it, but e.g. VLC does).
As we now have more PIDs in general, increase the maximum
number of demuxers we can open. ffmpeg has an internal
limit of 64 PIDs, we are still far from that.
If VDR-channel-config is used, we know the service_id.
Using that, PMT can be extracted from PAT (code for that
inspired by szap-s2 again).
For this, we need to demux PAT with a special filter-setup,
and once PMT was found it can be added to the PID-list.
PMT is only resolved the first time one tunes to a channel,
then stored in the channel-list in memory.
With PMT available, ffmpeg can safely decode the teletext-pids.
I also added a comment explaining the potentially occuring structures for future reference.
For tokenization, a custom strtok_r implementation is used,
inspired by strtok_r as implemented in musl and ffmpeg,
hopefully slightly more readable (av_strtok_r is not available in libav,
and strtok_r is not available everywhere).
Move PID-string parsing into a separate function, reduces code-duplication.
Drop TPID-parsing again since teletext-stream seems not really correctly identified (ffmpeg guessed mp3 in my test...).
Parsing now done using strtok + sscanf instead of sscanfs with fixed numbers of expected fields -
gives more correct checking of PID-overflow, will allow VDR-style PID-field parsing.
The PID-parsing is not really nicely done and also does not yet handle the special VDR-style syntax.
A later commit will move the parsing to a separate function handling that which also checks that the maximum PID-count is not exceeded.
In mplayer-style channels-config, we had a LNB-field used for that.
In old VDR times, the location-field was also containing DISEQc information,
now it does that only indirectly (location => LNB => vdr knows from lnb-config).
We only accept it as this if the field is fully numeric.
Now also "stream ID" (for DVB-S2) and "inversion" are understood.
The parameter-string can also provide information on FEC, rolloff etc. For DVB-S, "auto" which mpv uses by default should be fine, I can also confirm it works. For non-DVB-S cards, it might be useful to also parse this information in case of a vdr-channel list.
As I have no such hardware and thus would have to do it blindly, I added a FIXME.
Mostly complete vdr-channels.conf format documentation is at http://www.vdr-wiki.de/wiki/index.php/Channels.conf (german only).
Still incomplete. Initialize is_dvb_s2 boolean in channel-struct.
We first check whether the channels.conf-line at hand is sscanf'able with a vdr-style pattern. If yes, we assume it is a vdr-channel-config (we check whether sscanf consumed the full line).
The vdr-style config also contains a parameter-string which contains information about polarization + delivery type (e.g. DVB-S2).
With this change, S2-tuning works with a VDR-channel list.
Missing (later commits):
- vdr-parameter-string also contains other information, e.g. invert-flag, needs to be parsed.
- Diseqc-lnb-number is not present in VDR-config (I believe it is handled via the location-parameter + lnb-config there). For backwards compatibility, the location-parameter can be the lnb-number - we should test whether it is an int and assume this in this case.
- VID, AID and TID-lists are extremely ugly in their syntax for VDR. At the moment, only the first number is parsed (and TID fully ignored), needs to be fixed.
It contains now also a boolean which tells whether this is an DVB-S2 channel (not initialized yet - tbd in next commit).
We could also pass through the type of the delivery system (would be more flexible), but the delivery-system-enum is only available with S2API, so this is more backwards-compatible.
If someone has DVB-T2 / DVB-C2 hardware and extends the code to deal with it, this should be changed.
Also allows demuxers to buffer, and we explicitly discard stale QPSK events.
Inspiration taken from the szap-s2 implementation.
S2API is only used if available and for DVB-S cards - it might also be useful to have that for DVB-T2 / DVB-C2 tuning later, but I do not have the HW / no DVB-T2 broadcasting station nearby to test functionality.
This should do no functional changes, only usage of the different API. The S2API is more extensible and a requirement for all the HD-deliveries (e.g. DVB-S2) and since 2.6.28 is the successor of the old API.
Tuning to DVB-S2 channels actually already "works" like this if the delivery is hardcoded in - for a fully working implementation, the channels.conf.sat would need an additional field indicating the delivery type (VDR-type channels.conf have that).
This commit also:
- Cleans up some debug output.
- Moves an unneeded usleep(100000) (only needed for diseqc switching for SAT) into the DVB-S-only part of the implementation.
Define a new macro "DVB_USE_S2API" to be used to switch to S2API if available,
also document what is the main difference in the new S2API.
StreamID-handling is only supported since API 5.2, and some macros
were renamed or only added later with 5.8 (somewhere around kernel 3.6),
so some define-fixups are needed to support these intermediate
kernel header versions.
Commit 87c13de6 added a fallback to video resolution if the subtitle
resolution is unknown. Apparently this fixed some broken files with
vobsubs.
This broke some DVB subtitles. Apparently .ts captures with 1920x1080
video resolution and 720x576 subtitles do exist. The sample at hand had
some streams with 720x576 resolution and no sub resolution set, and some
streams with 1920x1080 resolution and sub resolution set (both against
the same 1920x1080 video). My conclusion is that 720x576 is the only
reasonable fallback for DVB (but I can't be sure).
The fallback is removed for PGS too. I don't know about the PGS case; it
seems the sub resolution must always be set, so it shouldn't matter.
Fixes#1425.
This makes vo_opengl_cb respond to controls like "gamma" and
"brightness". The commit includes an awkward refactor for vo_opengl to
make it easier for vo_opengl_cb.
One problem is a logical race condition. The set of supported controls
depends on the pixelformat, which in turn is set by reconfig(). But the
actual reconfig() call (on the renderer) happens asynchronously on the
renderer thread. At the time it happens, the player most likely already
tried to set some controls for command line options (see init_vo() in
video.c). So setting this command line options will fail most of the
time, though it could randomly succeed. This can't be fixed directly,
because the player can't wait on the renderer thread, because the
renderer thread might already wait on the player.
Although the line count increases, this is better for making sure
everything is handled consistently for all users of the mp_csp_params
stuff.
This also makes sure mp_csp_params is always initialized with
MP_CSP_PARAMS_DEFAULTS (for consistency).
