Directly setting rtspsrc to the NULL state before putting the pipeline
in the NULL state usually works, but it can cause a deadlock in some
cases, so it's not a reliable mechanism to fix this.

This reverts commit f37afdaf:
"rtspsrc: Fix state changes from PAUSED to PLAYING"

and commit 76d624b2:
"rtspsrc: Do not send PAUSE command when going to GST_STATE_NULL"

Part-of: <!908>

Sebastian Dröge authored 4 years ago and GStreamer Marge Bot committed 4 years ago

Due to an off-by-one when parsing the string, the most significant digit
or the clock offset was skipped when parsing the offset.

Part-of: <!907>

This was accidentally broken in the last commit that touched this
because I missed the fall-through in the case immediately above this.

Part-of: <!906>

Sebastian Dröge authored 4 years ago and Tim-Philipp Müller committed 4 years ago

The old code had a couple of issues that all lead to potential memory
safety bugs.

  - Use a constant for the Wavpack4Header size instead of using sizeof.
    It's written out into the data and not from the struct and who knows
    what special alignment/padding requirements some C compilers have.
  - gst_buffer_set_size() does not realloc the buffer when setting a
    bigger size than allocated, it only allows growing up to the maximum
    allocated size. Instead use a GstAdapter to collect all the blocks
    and take out everything at once in the end.
  - Check that enough data is actually available in the input and
    otherwise handle it an error in all cases instead of silently
    ignoring it.

Among other things this fixes out of bounds writes because the code
assumed gst_buffer_set_size() can grow the buffer and simply wrote after
the end of the buffer.

Thanks to Natalie Silvanovich for reporting.

Fixes #859

Part-of: <!902>

Sebastian Dröge authored 4 years ago and Tim-Philipp Müller committed 4 years ago

Various error return paths don't set it to NULL and callers are only
checking if the pointer is NULL. As it's allocated on the stack this
usually contains random stack memory, and more often than not the memory
of a previously parsed track.

This then causes all kinds of memory corruptions further down the line.

Thanks to Natalie Silvanovich for reporting.

Fixes #858

Part-of: <!902>

This usually doesn't matter, but it is disruptive when streaming from
a shared media since it will pause all other clients when any client
exits.

This new behaviour is opt-in and should be safe because you need to
set the NULL state on rtspsrc directly, instead of just on the
pipeline. See the updated documentation for an explanation.

Part-of: <!901>

Philipp Zabel authored 4 years ago and GStreamer Marge Bot committed 4 years ago

V4L2 makes no difference between the BT.601 and BT.709 transfer
functions [1], but GStreamer does since 1.18 [2].

Adapt gst_v4l2_object_get_colorspace() and
gst_v4l2_object_set_format_full().

[1] https://linuxtv.org/downloads/v4l-dvb-apis-new/userspace-api/v4l/colorspaces-details.html#colorspace-smpte-170m-v4l2-colorspace-smpte170m
[2] gst-plugins-base!724

Part-of: <!856>

GstRtspSrc -> GstRTSPSrc

This would have been noticed by the since checker, but those
properties were introduced prior to that.

Part-of: <!899>

Vladimir Menshakov authored 4 years ago and Sebastian Dröge committed 4 years ago

This commit negotiate F32 audio format if MODE_FLOAT used in wavpack file.
Wavpack float mode is always in 32-bit IEEE format.

The following pipeline plays distorted audio if source file is encoded in float mode:
gst-launch-1.0 filesrc ... ! wavpackparse ! wavpackdec ! pulsesink

Part-of: <!894>

Matthew Waters authored 4 years ago and GStreamer Marge Bot committed 4 years ago

Otherwise sending seek events would fail to actually seek.

Part-of: <!891>

Some cameras (e.g. HikVision DS-2CD2732F-IS) return "551 Option
not supported" when a command is sent that is not implemented
(e.g. PAUSE). Instead; it should return "501 Not Implemented".

This is wrong, as previously, the camera did announce support for PAUSE
in the OPTIONS.

In this case, handle the 551 as if it was 501 to avoid throwing errors
to application level. */

Part-of: <!885>

Qi Hou authored 4 years ago and GStreamer Marge Bot committed 4 years ago

Vp8/vp9 supported profiles/levels are listed in decoder sink caps, but
there is no parser for these two formats and the demuxers also don't have
these information. It causes negotiation fail between demuxers and decoder
when check caps "accept = gst_caps_is_subset (caps, template_caps);".
To fix this, need to remove profiles/levels for vp8/vp9 formats in decoder
sink caps.

Fix #854

Part-of: <!887>

We don't expect this object is a part of public library.

gstrtphdrext-twcc.c(45): warning C4273: 'gst_rtp_header_extension_twcc_get_type': inconsistent dll linkage

Part-of: <!889>

Philipp Zabel authored 4 years ago and Philipp Zabel committed 4 years ago

Negotiating v4l2h264dec ! v4l2h264enc transcoding pipelines fails in
case the encoder does not accept framerate=(fraction)0/1.
The acquired caps used for downstream negotiation are determined from
gst_v4l2_object_acquire_format(), which sets the GstVideoInfo::fps_n
and ::fps_d fields to 0.
To fix this, copy the frame rate from the sink side.

Part-of: <!882>

Jordan Petridіs authored 4 years ago and GStreamer Marge Bot committed 4 years ago

Could only test on rpi 3b+

Close #839

Part-of: <!875>

Don't be too spamy about unsupported source change flags as these will be
commonly extended in the future.

Part-of: <!870>

As we lock the DV_TIMINGS (and standards in the future), we need to probe the
caps after, otherwise, we may endup fixating to an unsupported resolution,
which would lead to a not-negotiated error.

Part-of: <!870>

And use this framerate in our preference. Note that we also flush
the probed caps as it seems that the format enumeration may change
when a new source change event get triggered.

Part-of: <!870>

This adds support to DV_TIMINGS query and locking. The timing width and
height is then used as a preference.

Part-of: <!870>

As mandated by the specification, make sure to cycle through streamoff
/ streamon regardless if the caps have changed or not.

Part-of: <!870>

Part-of: <!870>

Avoid passing around a bare structure for the preference, this removes
the need to copy and free that structure and simplify the code. Also
fix a type in the structure name, Prefered -> Preferred.

Part-of: <!870>

This stubs the ability to use preferred resolution from digital
video timings, analog TV standards or driver reported native
resolution.

Part-of: <!870>

When we subscribe for source-change event, we need to specify for which
input. Make sure we subscribe for the current input.

Part-of: <!870>

As part of the support to select a preferred size, we can also
detect the signal status. This is a split patch so that feature
is separated to ease review.

Part-of: <!870>

This is a wrapper around ENUM_INPUT renamed for readability.

Part-of: <!870>

This is an unsigned integer in the kernel API.

Part-of: <!870>

This patch adds support for source resolution change detection.
Resolution change is signaled by drivers when a change in the detected
signal have been detected. This is notably seen on HDMI receivers.

Part-of: <!870>

This patch adds the detection, dequeuing and reporting of the SOURCE_CHANGE
event when the CH_RESOLUTION flag is set. The acquire function will now return
a new custom success called GST_V4L2_FLOW_RESOLUTION_CHANGE. In order to use
this new feature, elements must enable it by calling:

  gst_v4l2_buffer_pool_enable_resolution_change (pool);

Part-of: <!870>

Part-of: <!870>

This uses the GST_V4L2_FLOW_LAST_BUFFER alias instead of
GST_FLOW_CUSTOM_SUCCESS to make the code more readable.

Part-of: <!870>

Lucas Stach authored 6 years ago and Nicolas Dufresne committed 4 years ago

Considering NV12 an 'odd' format is a historical artifact. This format
is now quite common, and usually preferable to I420 due to more memory
friendly access patterns.

Part-of: <!857>

wavparse claims to be able to support seeking in the READY state by
saving the pending seek event and actually seeking later after having parsed the
header.
Problem was that this seek event was reset on the READY to PAUSED
transition, making all this code useless. Fixing it by stop resetting
on READY to PAUSED transition as we already reset on PAUSED to READY
and when initiating the element.

Note that DTS marker detection isn't support in such scenario as
gst_type_find_helper_for_buffer() needs a buffer containing the
beginning of the stream.

Part-of: <!879>

No semantic change.

Part-of: <!879>

Mathieu Duponchelle authored 4 years ago and GStreamer Marge Bot committed 4 years ago

Part-of: <!834>

Mathieu Duponchelle authored 4 years ago and GStreamer Marge Bot committed 4 years ago

When set, the depayloader will request new keyframes on packet
loss

Part-of: <!834>

Mathieu Duponchelle authored 4 years ago and GStreamer Marge Bot committed 4 years ago

When set, the depayloader will request new keyframes on packet
loss

Part-of: <!834>

Mathieu Duponchelle authored 4 years ago and GStreamer Marge Bot committed 4 years ago

Similar to rtpvp8depay, when packet loss occurs, the depayloader
starts waiting for a keyframe.

We try to only stop waiting when all the packets for the new keyframe
have been received, by only resetting waiting_for_keyframe when
encountering the first packet of a keyframe, this is slightly
fragile because there is no bit that explicitly marks the start
of an access unit, so we rely on the existing picture_start
detection code.

As a consequence, the property is only meaningful when outputting
access units, and is ignored when outputting NALs directly.

Part-of: <!834>

Part-of: <!878>

Ashley B authored 4 years ago and Sebastian Dröge committed 4 years ago

wavenc: Fixed INFO chunk corruption, caused by odd sized data not being padded. Code style was updated.

Part-of: <!873>