prctl is supposed to take 5 arguments. It used to work with 2 arguments on some
versions of libc because it is defined as a varags function there.
See #611911
This either must never happen (which makes sense in this case) and thus should
use assert() or we should use a traditional if (poll_data->message) return;
to avoid differnet behaviour of intenal api when compiling with
G_DISABLE_CHECKS.
This avoids creating empty caps and destroying them in the case of an error. We
also avoid double checking in other code path where we call the internal api.
See 8fe63000de for why having a TRUE/FALSE
return value is a bad idea.
I've scanned a few plugins and they generally get it wrong and aren't
unloadable when they return FALSE.
This is what can happen in a plugin_init function:
- An element based on GstBaseSink is registered
- Other elements fail to register
- The plugin_init function returns FALSE
Now if this the plugin is the first plugin to link against
libgstbase.so, it will have caused libgstbase.so to be loaded and static
strings from that library will have been added to gobject while
registering GstBaseSink.
So unloading the plugin will cause those strings to go stale and the
next plugin using GstBaseSink will crash. So we must not unload modules
after calling into them ever.
https://bugzilla.redhat.com/show_bug.cgi?id=572800
This brings total call speedups between 5% and 25%.
gst_caps_set_simple_valist: +5%
gst_structure_set_valist: + 10%
gst_structure_id_set_valist: +25%
gst_tag_list_add_valist: +5%
Measured using valgrind when run over the discovery of 200 media files.
Fixes#610256
The alignment guaranteed by malloc is not always sufficient. E.g. vector
instructions or hardware subsystems want specifically aligned buffers. The
attached patch will use posix_memalign if available to allocate buffers.
The desired alignment can be set when running configure using the new
--with-buffer-alignment option.
This changes some APIs in compatible ways:
- Some functions now take "const char *" arguments, not "char *"
- Some structs now have "conts char *" members, not "char *"
The changes may cause warnings when compiling with the right warning
flags. You've been warned.
Also adds -Wwrite-strings as a warning flag in configure.ac.
https://bugzilla.gnome.org/show_bug.cgi?id=611692
Adds that warning to configure.ac
Includes a tiny change of the GST_BOILERPLATE_FULL() macro:
The get_type() function is no longer declared before being defined.
https://bugzilla.gnome.org/show_bug.cgi?id=611692
Previous code treated "1/1yourmom" the same as "1/1" and "1wimsmom" the
same as "1". Now the code is stricter and will fail to convert a
fraction when followed by garbage text.
This may cause crashes when logging is enabled, especially on windows.
It's not safe to pass random pointers to g_type_check_instance_is_a().
Fixes#611719.
Add check to make sure gst-plugin-scanner really gets installed where
we will look for it later, ie. paths and prefixes are set at configure
time and not specified via make.
Fixes#609941.
Late g_thread_init() is fine with newer GLib versions and done automatically
from g_type_init() there, so don't warn if the application hasn't called
g_thread_init() yet when gst_init() is called with new GLib versions.
Fixes#608398.
Refactor the code to take the current start_time when going to PAUSED.
Make sure we also call the start_time update code after we chained up to the
parent bin.
Fixes#607842
If the GST_PLUGIN_SCANNER environment variable is set, we should try
the scanner specified there first, to make sure the right scanner binary
is used for uninstalled setups and builds from source when there's
already an installed version.
We require threads to be supported in any case and defining this
will simplify the mutex, condition variable, etc. macros from gthread
to not always check if threads are really supported.
Fixes bug #607481.
For the reason outlined at the beginning of gst_private.h (inline
functions in glib may need the g_log_domain variable). Also include
gst_private.h before using any G_OS_* defines, esp. in plugin loader.
This check is not necesarry as we are not negotiating anymore. And it can
be wrong if upstream can't produce this caps anymore, but downstream can
process them fine.
All callers of this static function already check for NULL-ness
themselves, so no need to do it again (and if we do it, we should
probably do so before dereferencing the pointer for the first time).
Avoid checking the GType of the value twice (once on our side and
once in g_value_get_*()) by by-passing g_value_get() and accessing
the GValue structure directly.
Strings in the binary registry are NUL-terminated, so we can just use them
directly if we only need them temporarily, and avoid unnecessary mallocs
and frees.
No need to call gst_element_link_pads_filtered with filter=NULL, which would
call gst_element_link_pads() in that way. Call it directly to save a call and
expensive gobject type checks.
Fix up translated strings for some recently-added tags to match the
existing strings: we want short mnemonic-like strings here that start
with a lower case letter.
The only reason these two functions are still around is that at some
point in the past they were in a public header, so we can't really
remove them now even though they should have been private all along
(and aren't really particularly useful). Since these are just empty
stubs now that do nothing but return FALSE and will be removed in
0.11 anyway, we may just as well deprecate them formally.
Move the parent buffer pointer into the GstBuffer struct so that we can
remove the subbuffer class and type. This is interesting because it allows us to
more naturally implement methods to get the real type and parent
of a subbuffer (See #545501).
It should also be slightly faster because there is no extra object hierarchy to
initialize and free.
When we unblock a pad with the same user_data, the destroy callback is not
called. This leads to refcounting leaks that cannot be avoided. Instead always
call the destroy notify whenever we install a new pad block.
In particular, this fixes a nasty pad leak in decodebin2.
Also update the unit test to have more accurate comments and test the required
behaviour.
Adds the following new tags:
GST_TAG_SHOW_NAME
GST_TAG_SHOW_SORTNAME
GST_TAG_SHOW_EPISODE_NUMBER
GST_TAG_SHOW_SEASON_NUMBER
GST_TAG_LYRICS
GST_TAG_COMPOSER_SORTNAME
GST_TAG_GROUPING
Fixes#599759
Setting an object name is nice for proper debug logging. Ideally this would
still happens earlier (.e.g when pads are added to an element, its not yet set).
Never skip the state change to playing, even if the element is already in the
right state. We need this because we also distribute the base_time while doing
the state change and skipping this step would leave some elements without a new
base_time.
Fixes#600313
gst_value_list_size and gst_value_list_get_value will do a series of
extra checks due to being public methods.
When we use them from within gstvalue.c we can directly use them without
the extra checks.
This makes ghostpad/proxypad creation 5 times faster and avoids contention
over the global funcptr lock.
I also moved the two class init down in the code to avoid having to forward
declare all the various functions.
gst_event_new_sink_message()
gst_event_parse_sink_message()
This event is used for sending a GstMessage downstream and synchronized
with the stream, to be posted by the sink once it reaches the sink.
Fixes bug #602275.
and install into a different directory $(libexecdir/gstreamer-0.10) so that
everything is versioned properly.
NOTE: run 'make clean' after updating; if you are running an uninstalled setup,
you will need to update your gst-uninstalled script (unless it's symlinked
to gstreamer core master) and exit/enter your uninstalled environment to get
the updated environment. If you are running an installed setup, you should
run 'make uninstall' before merging this change or remove the old
plugin-scanner binary manually.
Fixes#601698.
This reverts commit f864187bf5.
Reverting this as it changes behaviour and the documentation is
ambiguous about whether the caller must check the type first or
not (call must check type vs. returns NULL if not a string). If
GLib has been compiled with G_DISABLE_CHECKS then g_value_get_string()
may return complete garbage even if the value does not contain
a string. Better play it safe, esp. since the extra check is just
an integer comparison. For fundamental types we could return values
from the GValue structure directly if we really wanted to bypass
the extra check.
gst_util_greatest_common_divisor()
gst_util_double_to_fraction()
gst_util_fraction_to_double()
Using these instead of going over GValue has much lower overhead.
Also add float<->fraction transform functions for GValue.
Avoid a race where a miniobject is recycled and quickly freed, which causes the
g_type_free_instance() to be called on the same object twice.
Ref the object before calling the finalize method and check if we still need to
free it afterward.
Also add a unit test for this case.
Fixes#601587
This returns the internal feature list cookie, which changes every
time a feature is added or removed. This can be used by elements
to check if they should update their cached feature lists.
Force printf format checking for debug messages if GST_DISABLE_PRINTF_EXTENSION
is defined. This is useful to quickly check code for printf format mismatches
in debugging messages that would usually not be caught (with glibc+gcc and
printf extensions being used).
To use: make clean; make CFLAGS='-g -O2 -DGST_DISABLE_PRINTF_EXTENSION'
The logic to handle short reads/writes was incorrect, causing the
packet handler to attempt to handle incomplete packets.
Grow the packet transmit buffer in proportion to observed usage,
causing fewer reallocs.
Add some more debug in the registry chunks code.
When invalid registry chunks are received from the child, and parsing
fails, don't access an invalid plugin pointer. Instead attempt to
figure out which plugin caused the problem and blacklist it.
There's not much point in using GST_DEBUG_FUNCPTR with GObject
virtual functions such as get_property, set_propery, finalize and
dispose, since they'll never be used by anyone anyway. Saves a
few bytes and possibly a sixteenth of a polar bear.
If a downstream element returns an error while upstream has already
put all data into queue2 (including EOS), upstream will no longer
chain into queue2, so it is up to queue2 to perform some
EOS handling / message posting in such cases. See #589991.
Keep track of the max requested position and compare this to the write position
in the temp file to get the current amount of buffered data.
Fix memleak of all incomming buffers.
Fixes#588551
Don't flush the file by closing and opening it but instead use g_freopen. This
avoids a deadlock in shutdown because we emit the temp-location property change
with the wrong lock held.
Add a new temp-template property so that queue2 can securely allocate a
temporary filename. Deprecate the temp-location property for setting the
location but still use it to notify the allocated temp file.
Original commit message from CVS:
* docs/design/draft-keyframe-force.txt:
Fix typo.
* gst/playback/gstqueue2.c: (update_buffering),
(gst_queue_handle_src_query):
Set buffering mode in the messages.
Set buffering percent in the query.
* tests/examples/seek/seek.c: (update_fill), (msg_state_changed),
(do_stream_buffering), (do_download_buffering), (msg_buffering):
Do some more fancy things based on the buffering method in use.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (update_buffering),
(gst_queue_close_temp_location_file), (gst_queue_handle_src_query),
(gst_queue_src_checkgetrange_function):
Include extra buffering stats in the buffering message.
Implement BUFFERING query.
* gst/playback/gsturidecodebin.c: (do_async_start),
(do_async_done), (type_found), (setup_streaming), (setup_source),
(gst_uri_decode_bin_change_state):
Only add decodebin2 when the type is found in streaming mode.
Make uridecodebin async to PAUSED even when we don't have decodebin2
added yet.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (update_out_rates),
(gst_queue_open_temp_location_file),
(gst_queue_close_temp_location_file), (gst_queue_handle_src_event),
(gst_queue_handle_src_query), (gst_queue_set_property):
Update the estimated input data when we push out a buffer.
Add some debug info about the temp file.
Only forward src events when we are not using a temp file.
Don't block the duration query, we need to find something better.
Don't leak the temp filename.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (gst_queue_is_filled):
The queue is never filled when there are no buffers in the queue at all.
Fixes#523993.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (gst_queue_init), (gst_queue_finalize),
(reset_rate_timer), (update_in_rates), (update_out_rates),
(gst_queue_locked_enqueue), (gst_queue_locked_dequeue),
(gst_queue_chain), (gst_queue_loop):
Use separate timers for input and output rates.
Pause measuring the output rate when we block for more data.
See #503262.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (gst_queue_chain):
Pause the timer to measure the input rate when we block because the
queue is filled. See #503262.
Original commit message from CVS:
* ext/pango/gsttextoverlay.c: (gst_text_overlay_init),
(gst_text_overlay_set_property):
* ext/vorbis/vorbisdec.c: (vorbis_handle_data_packet):
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_render):
* gst-libs/gst/rtp/gstrtcpbuffer.c: (gst_rtcp_ntp_to_unix),
(gst_rtcp_unix_to_ntp):
* gst-libs/gst/rtsp/gstrtspmessage.c: (gst_rtsp_message_get_type):
* gst/playback/gstqueue2.c:
* tests/examples/seek/seek.c: (set_scale):
Fix a bunch of compile warnings shown with Forte.
* gst/audiorate/gstaudiorate.c:
Always pull in config.h before including any system headers.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (update_buffering),
(gst_queue_locked_flush), (gst_queue_locked_enqueue),
(gst_queue_handle_sink_event), (gst_queue_chain),
(gst_queue_push_one), (gst_queue_sink_activate_push),
(gst_queue_src_activate_push), (gst_queue_src_activate_pull):
Also fix#476514 for queue2.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (apply_segment), (update_buffering):
Use other metrics as well when estimating the buffer level.
Original commit message from CVS:
* gst/playback/gstplaybasebin.c: (make_decoder), (setup_source):
Small debug improvement.
* gst/playback/gstqueue2.c: (apply_segment), (update_buffering),
(plugin_init):
Tweak the rate estimation period.
When calculating the buffer filledness in rate estimation mode, don't
mix it with other metrics.
Original commit message from CVS:
Patches by: Thiago Sousa Santos <thiagossantos at gmail dot com>
* gst/playback/gstqueue2.c: (update_buffering),
(gst_queue_locked_enqueue):
Fix a division by zero when the max percent is <= 0. Fixes#446572.
also update the buffering status when receiving events. Fixes#446551.
Original commit message from CVS:
Based on patch by: Thiago Sousa Santos <thiagossantos at gmail dot com>
* gst/playback/gstqueue2.c: (gst_queue_peer_query),
(gst_queue_handle_src_query):
Wait for preroll before attempting to forward a duration query upstream.
Fixes#445505.
Original commit message from CVS:
Patch by: Thiago Sousa Santos <thiagossantos at gmail dot com>
* gst/playback/gstqueue2.c: (gst_queue_init),
(gst_queue_handle_sink_event), (gst_queue_chain),
(gst_queue_get_range), (gst_queue_src_checkgetrange_function),
(gst_queue_sink_activate_push), (gst_queue_src_activate_push),
(gst_queue_src_activate_pull):
Add pull based scheduling and fix some deadlocks. Fixes#444523.
Does not yet completely work because duration queries upstream won't
block yet.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (gst_queue_have_data),
(gst_queue_create_read), (gst_queue_read_item_from_file),
(gst_queue_open_temp_location_file), (gst_queue_locked_enqueue):
Include stdio to define fseeko.
Original commit message from CVS:
* gst/playback/gstqueue2.c: (update_rates):
Tweak the buffering thresholds a little.
Update the buffer size with the previously calculate rate instead of
only when we calculate a new rate so that we get smoother buffering
updates.
* gst/playback/Makefile.am:
* gst/playback/gsturidecodebin.c: (gst_uri_decode_bin_base_init),
(gst_uri_decode_bin_class_init), (gst_uri_decode_bin_init),
(gst_uri_decode_bin_finalize), (gst_uri_decode_bin_set_property),
(gst_uri_decode_bin_get_property), (unknown_type),
(add_element_stream), (no_more_pads_full), (no_more_pads),
(source_no_more_pads), (new_decoded_pad), (array_has_value),
(gen_source_element), (has_all_raw_caps), (analyse_source),
(remove_decoders), (make_decoder), (remove_source),
(source_new_pad), (setup_source), (decoder_query_init),
(decoder_query_duration_fold), (decoder_query_duration_done),
(decoder_query_position_fold), (decoder_query_position_done),
(decoder_query_latency_fold), (decoder_query_latency_done),
(decoder_query_seeking_fold), (decoder_query_seeking_done),
(decoder_query_generic_fold), (gst_uri_decode_bin_query),
(gst_uri_decode_bin_change_state), (plugin_init):
New element that intergrates a source, optional buffering element and
decodebin.
Original commit message from CVS:
* gst/playback/Makefile.am:
* gst/playback/gstqueue2.c: (gst_queue_get_type),
(gst_queue_class_init), (gst_queue_init), (gst_queue_finalize),
(gst_queue_getcaps), (gst_queue_bufferalloc),
(gst_queue_acceptcaps), (update_time_level), (apply_segment),
(apply_buffer), (update_buffering), (reset_rate_timer),
(update_rates), (gst_queue_locked_flush),
(gst_queue_locked_enqueue), (gst_queue_locked_dequeue),
(gst_queue_handle_sink_event), (gst_queue_is_empty),
(gst_queue_is_filled), (gst_queue_chain), (gst_queue_push_one),
(gst_queue_loop), (gst_queue_handle_src_event),
(gst_queue_handle_src_query), (gst_queue_sink_activate_push),
(gst_queue_src_activate_push), (gst_queue_change_state),
(gst_queue_set_property), (gst_queue_get_property), (plugin_init):
On our way to playbin2 this is the new network queue that does buffering
all by itself using high and low watermarks. It can also measure up and
downstream bandwidth to optimally size the queue.
The default gst_pad_fixate_caps() previously would only fixate each individual
struct. In case there are multiple structs, the resulting caps would still not
be fixed. In the spirit of how individual structs are fixated, this patch
changes gst_pad_fixate_caps() to remove all but the first struct.
Fixes#595886
This avoids:
* triple-checking for the GType when type-checking is enabled (see #597260)
* Avoids going through an expensive no-argument checking which landed in
glib-2.22
* Avoids going through 2 extrac functions (g_object_new -> g_object_new_valist)
The OGI links are dead, so remove them. Also remove the paragraph that pointed
to OGI and DS. Only mentioning DS there made it a but pointless. Add a generic
paragraph instead that tells a bit about the usecases gstreamer covers.
There's not much point in using GST_DEBUG_FUNCPTR with GObject
virtual functions such as get_property, set_propery, finalize and
dispose, since they'll never be used by anyone anyway. Saves a
few bytes and possibly a tenth of a polar bear.
This avoids having to do the sorting everytime we use typefind
The behaviour of gst_type_find_factory_get_list has subtlely changed
in the sense that the order was previously undefined, whereas now
it returns them sorted by rank and then by name.
Simply use casting macros for accessing the message fields like we do for
buffers and events. Avoids some costly typechecking that does not really buy us
much.
CAPS_IS_ANY and CAPS_IS_EMPTY are the equivalent of their gst_caps_*
counterpart except that they avoid the typechecking and are inlined.
CAPS_IS_EMPTY_SIMPLE only checks for empty caps (without checking if
the caps is ANY).
Maintain a hashtable of the plugin basename. We can then use this
hashtable to speedup the search for an existing plugin and avoid
a whole lot of strcmp calls.
This patch adds gst_caps_set_value() and allows gst_caps_set_simple() to
work on non-simple caps. See the API documentation for the functions
about what they do.
The intention of these changes is to ease working with caps in caps
transform functions. An example for this would be ffmpegcolorspace,
where the caps transform function could be changed to look roughly like
this (pseudocode ahead):
result = gst_caps_copy (template_caps);
value = gst_structure_get_value (gst_caps_get_structure (caps, 0),
"widh");
gst_caps_set_value (result, value);
/* same for height, framerate and par */
return caps;
which is much cleaner and easier to understand than the current code.
https://bugzilla.gnome.org/show_bug.cgi?id=597690
This is slightly more efficient because the compiler can't do tail
recursion here and has to keep all stack frames.
Not that efficiency is that important here but I already had
the iterative version somewhere else and both are easy to read.
This allows the macosx versions to properly error out when fds are closed.
This is only a temporary fix until the pluginloader is switched to not
use GstPoll but GIOChannels.
In the plugin loader subprocess, move stdin and stdout to new fd's
so that plugins printing things during plugin init or (*gasp*)
possibly reading from stdin don't interfere with the data sent to
and from the parent.
When the plugin-scanner load fails (because the helper can't be
spawned), make sure to load the plugin that failed in-process, so
that all plugins do get loaded.
Fix an off-by-one error in the size guard for unpack_element, and
improve various debug statements in the failure paths.
Also, swap some g_new0 to g_malloc0 for the fun of it.
In most places in core and baseclasses we just need the caps to do caps-
intersections. In that case ref'ed caps are enough (no need to copy).
This patch also switches the code to use the new functions.
API: gst_pad_get_caps_refed(), gst_pad_peer_get_caps_refed()
Everyone running an uninstalled git setup is going to wonder about
this failure next time they update, so let's mention the solution
in the error message.
Reflow the code to move error handling to the end of the functions. Initialize
gvalue like we do in the setter. Add a unit-test module with two simple tests
the catche this bug.
We don't need the hold the proxy mutex locked for getting the internal pad and
for linking the new target pad when we retarget. So take the lock a little later
and release it earlier.
Fixes#596366
Guard against a hostile child process that sends bogus data
due to memory corruption by adding a magic number to each packet,
and limit the maximum size of any message to 32MB
When trying to find a function plugin-scanner, include a check on the
version of the binary registry chunks it sends, to make sure it's
what we understand.
Add a simple version check when starting the plugin-scanner so we can
verify we're talking to one that talks the same language.
First try a plugin-scanner in the installed path, then try one via the
GST_PLUGIN_SCANNER env var if that doesn't work.
Update the uninstalled script.
Install the plugin-scanner to the libexec dir
When updating the registry, we don't need to re-read the registry cache
and waste time replacing all our existing, hopefully identical, plugins
and features that we're about to re-scan anyway.
phase 2 - make the plugin loader receive the list of plugins to load and
send back the results asynchronously, so we don't context switch back
and forth so much.
The structure_change message was originally emitted on source pads and
then recently changed to be sink pads. This causes a failure in the
gst-python testsuite. Disable the restriction so that the published
behaviour is still allowed.
Also parse the nano of the version and assume that X.Y.Z-1.1 >= X.Y.Z
With this change we can also check development versions against the version of
the upcomming release.
This is available in newer gcc releases and it should only exist
on platforms that provide some native 128bit integer arithmetic
instructions.
The x86-64 assembly for this is still kept for non-gcc compilers
that don't provide __uint128_t magic.
Post the structure change messages on the sinkpads of the elements. This allows
us to catch unlinked pads earlier without ending up with inconsistent element
degrees.
When we detect a pad unlink in progress, we will not be updating the degree of
the parent element. This can cause false loop detected warnings because the
degree counter is invalid. Handle this case by marking the iterator as 'dirty'
when we detect a pad unlink and avoid emiting the warning in this case. We have
to continue our state change as good as we can, we will eventually resync when
the pad unlink completed.
When an element is added to the bin, only set the index if we have a
cached index, rather than setting a NULL index on elements that might
have a default index object of their own.
elementfactory field is filled in by gst_element_base_class_init,
but it needs some info set on the element's type, so have it
available prior to class structure creation spinning up.
This affects elements that have a well-known/public type (e.g. pipeline)
and can be created by other means than gst_element_factory_make
(which will also fill in the element's factory).
Cache the last index that was set with _set_index() and return this in the
_get_index() call.
Set the cached index on newly added elements.
Fixes#566881
Based on clock implementation by Håvard Graff <havard.graff@tandberg.com>
Try to get the time on windows using the performance counters. These have a much
higher resolution and accuracy than the regular getcurrenttime(). Be careful to
fall back to regular getcurrenttime() or posix clocks when performance counters
are not available.
We can use a shift for scaling the denominator instead of a divide since the
denom is always positive. This avoids having the compiler generate code for the
different rounding rules when scaling negative values.
64bit x86 has native 64x64->128 bit multiply that we can use with some inline
assembler to speed up large multiplications.
Use bsr to find the number of leading zeros more efficiently.
The internal links function is deprecated since some time and
there already were GST_REMOVE_DEPRECATED markers in the source file,
now add them to the header too.
Fixes bug #592209.
Before the signal handler would get the ghostpad passed as second
argument but it could've already been unreffed and destroyed.
This would then lead to crashes and all that.
Now we get the ghostpad from the proxy pad, which we get from the
target pad as it's peer.
Fixes bug #591318.
The new functions are
gst_util_uint64_scale_int_round()
gst_util_uint64_scale_int_ceil()
gst_util_uint64_scale_round()
gst_util_uint64_scale_ceil()
Fixes bug #590919.
Before it returned that [start,stop] is inside the segment and that the
difference between segment_start and start needs to be clipped. If the
clipping is done on a buffer (like in baseaudiosink) this will result
in the data pointer being at a invalid memory position.
Fixes bug #589849.
Check for metadata writability when setting caps on buffer or when copying
metadata flags. Only enable these extra assertions in git versions.
This should help us find bad elements.
