Without this we can easily run into a race condition with async state changes:
- the pipeline is doing an async state change
- we set the internal bins to PLAYING but that's ignored because an
async state change is currently pending
- the async state change finishes but does not change the state of the
internal bins because of locked_state==TRUE
- the internal bins stay in PAUSED forever
Add functionality to limit the Content-Length.
API addition, Enhancement.
Define an appropriate request size limit and reject requests
exceeding the limit with response status 413 Request Entity Too Large
Related to !182
Otherwise it will never try to send us the next one: it tries to keep
exactly one message in-flight all the time.
In gst-rtsp-server this is done asynchronously via the GstRTSPWatch but
in the client sink we always write data out synchronously.
If not waiting for free thread pool before clean transport caches, there
can be a crash if a thread is executing in transport list loop in
function send_tcp_message.
Also add a check if priv->send_pool in on_message_sent to avoid that a
new thread is pushed during wait of free thread pool. This is possible
since when waiting for free thread pool mutex have to be unlocked.
Handle the situation when a call to gst_rtsp_media_set_state is done
when media status is preparing.
Also add unit test for this scenario.
The unit test simulate on a media level when two clients share a (live)
media.
Both clients have done SETUP and got responses. Now client 1 is doing
play and client 2 is just closing the connection.
Then without patch there are a problem when
client1 is calling gst_rtsp_media_unsuspend in handle_play_request.
And client2 is doing closing connection we can end up in a call
to gst_rtsp_media_set_state when
priv->status == GST_RTSP_MEDIA_STATUS_PREPARING and all the logic for
shut down media is jumped over .
With this patch and this scenario we wait until
priv->status == GST_RTSP_MEDIA_STATUS_PREPARED and then continue to
execute after that and now we will execute the logic for
shut down media.
This adds new functions for passing buffer lists through the different
layers without breaking API/ABI, and enables the appsink to actually
provide buffer lists.
This should already reduce CPU usage and potentially context switches a
bit by passing a whole buffer list from the appsink instead of
individual buffers. As a next step it would be necessary to
a) Add support for a vector of data for the GstRTSPMessage body
b) Add support for sending multiple messages at once to the
GstRTSPWatch and let it be handled internally
c) Adding API to GOutputStream that works like writev()
Fixes https://gitlab.freedesktop.org/gstreamer/gst-rtsp-server/issues/29
The close handler could trigger a crash because it invalidated the
watch_context while still leaving a source attached to it which would be
cleaned up at a later point.
The previous fix for race condition around finish_unprepare where the
function could be called twice assumed that the status wouldn't change
during execution of the function. This assumption is incorrect as the
state may change, for example if an error message arrives from the
pipeline bus.
Instead a flag keeping track on whether the finish_unprepare function
is currently executing is introduced and checked.
Fixes https://gitlab.freedesktop.org/gstreamer/gst-rtsp-server/issues/59
In plug_src we changed the element state before adding it to
the owner container. This prevented the pipeline from intercepting
a GST_STREAM_STATUS_TYPE_CREATE message from the pad in order
to assign a custom task pool.
Fixes https://gitlab.freedesktop.org/gstreamer/gst-rtsp-server/issues/53
Media is considered to be blocked when all streams that belong to
that media are blocked.
This patch solves the problem of inconsistent updates of
priv->blocked that are not synchronized with the media state.
Before the seek operation is performed on media, it's required that
its pipeline is prepared <=> the pipeline is in the PAUSED state.
At this stage, all transport parts (transport sinks) have been successfully
added to the pipeline and there is no need for blocking the streams.
The sequence number in the rtpinfo is supposed to be the first RTP
sequence number. The "seqnum" property on a payloader is supposed to be
the number from the last processed RTP packet. The sequence number for
payloaders that inherit gstrtpbasepayload will not be correct in case of
buffer lists. In order to fix the seqnum property on the payloaders
gst-rtsp-server must get the sequence number for rtpinfo elsewhere and
"seqnum-offset" from the "stats" property contains the value of the
very first RTP packet in a stream. The server will, however, try to look
at the last simple in the sink element and only use properties on the
payloader in case there no sink elements yet, and by looking at the last
sample of the sink gives the server full control of which RTP packet it
looks at. If the payloader does not have the "stats" property, "seqnum"
is still used since "seqnum-offset" is only present in as part of
"stats" and this is still an issue not solved with this patch.
Needed for gst-plugins-base!17
... by actually making it a single-include header and moving everything
related to the GstRTSPServer type to rtsp-server-object.h instead.
Otherwise there are too many circular includes.
https://bugzilla.gnome.org/show_bug.cgi?id=797361
When the underlying layers are running on_message_sent, this sometimes
causes the underlying layer to send more data, which will cause the
underlying layer to run callback on_message_sent again. This can go on
and on.
To break this chain, we introduce an idle source that takes care of
sending data if there are more to send when running callback
https://bugzilla.gnome.org/show_bug.cgi?id=797289
Avoids ending up with races where a timeout would still be around
*after* a client was gone. This could happen rather easily in
RTSP-over-HTTP mode on a local connection, where each RTSP message
would be sent as a different HTTP connection with the same tunnelid.
If not properly removed, that timeout would then try to free again
a client (and its contents).
By default the multicast sockets are bound to INADDR_ANY,
as it's not allowed to bind sockets to multicast addresses
in Windows. This default behaviour can be changed by setting
bind-mcast-address property on the media-factory object.
https://bugzilla.gnome.org/show_bug.cgi?id=797059
Export rtsp-server library API in headers when we're building the
library itself, otherwise import the API from the headers.
This fixes linker warnings on Windows when building with MSVC.
Fix up some missing config.h includes when building the lib which
is needed to get the export api define from config.h
https://bugzilla.gnome.org/show_bug.cgi?id=797185
If a (strange) client would reuse interleaved channel numbers in
multiple SETUP requests, we should not accept them. The channel
numbers are used for looking up stream transports in the
priv->transports hash table, and transports disappear from the table
if channel numbers are reused.
RFC 7826 (RTSP 2.0), Section 18.54, clarifies that it is OK for the
server to change the channel numbers suggested by the client.
https://bugzilla.gnome.org/show_bug.cgi?id=796988
When media is shared, the same media stream can be sent
to multiple multicast groups. Currently, there is no API
to retrieve multicast addresses from the stream.
When calling gst_rtsp_stream_get_multicast_address() function,
only the first multicast address is returned.
With this patch, each multicast destination requested in SETUP
will be stored in an internal list (call to
gst_rtsp_stream_add_multicast_client_address()).
The list of multicast groups requested by the clients can be
retrieved by calling gst_rtsp_stream_get_multicast_client_addresses().
There still exist some problems with the current implementation
in the multicast case:
1) The receiving part is currently only configured with
regard to the first multicast client (see
https://bugzilla.gnome.org/show_bug.cgi?id=796917).
2) Secondly, of security reasons, some constraints should be
put on the requested multicast destinations (see
https://bugzilla.gnome.org/show_bug.cgi?id=796916).
Change-Id: I6b060746e472a0734cc2fd828ffe4ea2956733ea
https://bugzilla.gnome.org/show_bug.cgi?id=793441
The maximum ttl value provided so far by the multicast clients
will be chosen and reported in the response to the current
client request.
Change-Id: I5408646e3b5a0a224d907ae215bdea60c4f1905f
https://bugzilla.gnome.org/show_bug.cgi?id=793441
If "transport.client-settings" parameter is set to true, the client is
allowed to specify destination, ports and ttl.
There is no need for pre-configured address pool.
Change-Id: I6ae578fb5164d78e8ec1e2ee82dc4eaacd0912d1
https://bugzilla.gnome.org/show_bug.cgi?id=793441
When two multicast clients request specific transport
configurations, and "transport.client-settings" parameter is
set to true, it's wrong to actually require that these two
clients request the same multicast group.
Removed test_client_multicast_invalid_transport_specific test
cases as they wrongly require that the requested destination
address is supposed to be present in the address pool, also in
the case when "transport.client-settings" parameter is set to true.
Change-Id: I4580182ef35996caf644686d6139f72ec599c9fa
https://bugzilla.gnome.org/show_bug.cgi?id=793441
If "transport.client-settings" parameter is set to true, the client is
allowed to specify destination, ports and ttl.
There is no need for pre-configured address pool.
https://bugzilla.gnome.org/show_bug.cgi?id=793441
multiudpsink does not support setting the socket* properties
after it has started, which meant that rtsp-server could no
longer serve on both IPV4 and IPV6 sockets since the patches
from https://bugzilla.gnome.org/show_bug.cgi?id=757488 were
merged.
When first connecting an IPV6 client then an IPV4 client,
multiudpsink fell back to using the IPV6 socket.
When first connecting an IPV4 client, then an IPV6 client,
multiudpsink errored out, released the IPV4 socket, then
crashed when trying to send a message on NULL nevertheless,
that is however a separate issue.
This could probably be fixed by handling the setting of
sockets in multiudpsink after it has started, that will
however be a much more significant effort.
For now, this commit simply partially reverts the behaviour
of rtsp-stream: it will continue to only create the udpsinks
when needed, as was the case since the patches were merged,
it will however when creating them, always allocate both
sockets and set them on the sink before it starts, as was
the case prior to the patches.
Transport configuration will only error out if the allocation
of UDP sockets fails for the actual client's family, this
also downgrades the GST_ERRORs in alloc_ports_one_family
to GST_WARNINGs, as failing to allocate is no longer
necessarily fatal.
https://bugzilla.gnome.org/show_bug.cgi?id=796875
Before, the watch backlog size in GstRTSPClient was changed
dynamically between unlimited and a fixed size, trying to avoid both
unlimited memory usage and deadlocks while waiting for place in the
queue. (Some of the deadlocks were described in a long comment in
handle_request().)
In the previous commit, we changed to a fixed backlog size of 100.
This is possible, because we now handle RTP/RTCP data messages differently
from RTSP request/response messages.
The data messages are messages tunneled over TCP. We allow at most one
queued data message per stream in GstRTSPClient at a time, and
successfully sent data messages are acked by sending a "message-sent"
callback from the GstStreamTransport. Until that ack comes, the
GstRTSPStream does not call pull_sample() on its appsink, and
therefore the streaming thread in the pipeline will not be blocked
inside GstRTSPClient, waiting for a place in the queue.
pull_sample() is called when we have both an ack and a "new-sample"
signal from the appsink. Then, we know there is a buffer to write.
RTSP request/response messages are not acked in the same way as data
messages. The rest of the 100 places in the queue are used for
them. If the queue becomes full of request/response messages, we
return an error and close the connection to the client.
Change-Id: I275310bc90a219ceb2473c098261acc78be84c97
Change to using a fixed backlog size WATCH_BACKLOG_SIZE.
Preparation for the next commit, which changes to a different way of
avoiding both deadlocks and unlimited memory usage with the watch
backlog.
Fix race when setting up source elements.
Since we set the source element(s) to PLAYING state before hooking
them up to the downstream funnel, it's possible for the source element
to receive packets before we actually get to linking it to the funnel,
in which case buffers would be pushed out on an unlinked pad, causing
it to error out and stop receiving more data.
We fix this by blocking the source's srcpad until we have linked it.
https://bugzilla.gnome.org/show_bug.cgi?id=796160
Transport specific sink elements are added to the pipeline
in PLAY request and sockets are already created in SETUP so
it's actually wrong to require the presence of sinks in
_get_*_socket() functions.
https://bugzilla.gnome.org/show_bug.cgi?id=793441
If a multicast client requests different transport settings
than the existing one make sure that this new transport
configuruation is propagated to the multicast udp sink.
https://bugzilla.gnome.org/show_bug.cgi?id=793441
Passwords are usually not stored in clear text, but instead
stored already hashed in a .htdigest file.
Add support for parsing such files, add API to allow setting
a custom realm in RTSPAuth, and update the digest example.
https://bugzilla.gnome.org/show_bug.cgi?id=796637
We were previously only ever waiting for a single stream to notify it's
blocked status through GstRTSPStreamBlocking. Actually count streams to
wait for.
Fixes rtspclientsink sending SDP's without out some of the input
streams.
https://bugzilla.gnome.org/show_bug.cgi?id=796624
When streaming data over TCP then is not the keep-alive
functionality working.
The reason is that the function do_send_data have changed
to boolean but the code is still checking the received result
from send_func with GST_RTSP_OK.
The result is that a successful send_func will always lead to
that do_send_data is returning false and the keep-alive will
not be updated.
https://bugzilla.gnome.org/show_bug.cgi?id=795321
This reverts commit 3d275b1345.
While RFC 3264 (SDP) says that sendonly/recvonly are from the point of view of
the requester, the actual RTSP RFCs (RFC 2326 / 7826) disagree and say
the opposite, just like the ONVIF standard.
Let's follow those RFCs as we're doing RTSP here, and add a property at
a later time if needed to switch to the SDP RFC behaviour.
https://bugzilla.gnome.org/show_bug.cgi?id=793964
If the media is complete, i.e. one or more streams have been configured
with sinks, then we want to query the position on those streams only.
A query on an incomplete stream may return a position that originates from
an earlier preroll.
https://bugzilla.gnome.org/show_bug.cgi?id=794964
Set transport string to NULL after freeing it, so that
at worst we get a NULL pointer if constructing a new
transport string fails (which shouldn't really fail here).
Also check return value of that, just in case.
CID 1433768.
Added it right before pushing the previous commit, it is
incorrect and deadlocks because this function gets called
from the join_bin thread, which already holds the lock,
that's the reason why request_aux_sender didn't take the
lock either.
"do-retransmission" was previously set when rtx-time != 0,
which made no sense as do-retransmission is used to enable
the sending of retransmission requests, where as rtx-time
is used by the peer to enable storing of buffers in order
to respond to retransmission requests.
rtsp-media now also provides a callback for the
request-aux-receiver signal.
https://bugzilla.gnome.org/show_bug.cgi?id=794822
This was broken since the work for delayed transport creation
was merged: the creation of the transports string depends on
calling stream_get_server_port, which only starts returning
something meaningful after a call to stream_allocate_udp_sockets
has been made, this function expects a transport that we parse
from the transport string ...
Significant refactoring is in order, but does not look entirely
trivial, for now we put a band aid on and create a second transport
string after the stream has been completed, to pass it in
the request headers instead of the previous, incomplete one.
https://bugzilla.gnome.org/show_bug.cgi?id=794789
rtsp-media-factory-uri.c: In function ‘rtsp_media_factory_uri_create_element’:
rtsp-media-factory-uri.c:621:17: error: assignment from incompatible pointer type [-Werror=incompatible-pointer-types]
data->factory = g_object_ref (factory);
^
We need different export decorators for the different libs.
For now no actual change though, just rename before the release,
and add prelude headers to define the new decorator to GST_EXPORT.
This allows us to later map signals from rtpbin/rtpsource back to the
corresponding stream transport, and allows to do keep-alive based on
RTCP packets in case of TCP media transport.
https://bugzilla.gnome.org/show_bug.cgi?id=789646
As READY_TO_PAUSED can no longer return async, the RECORD
command will be queued before the OPEN command fails
(for example in case the server could not be connected),
and record then waits for ever.
https://bugzilla.gnome.org/show_bug.cgi?id=793896
The test_record case was working because async=false had
been added in https://bugzilla.gnome.org/show_bug.cgi?id=757488
but that was incorrect, as it should not be needed.
Removing async=false made the test fail as expected, this is
fixed by not trying to preroll when preparing the media for
RECORD, as start_prepare is called upon receiving ANNOUNCE,
and our peer will not start sending media until it has received
a response to that request, and sent and received a response
to RECORD as well, thus obviously preventing preroll.
https://bugzilla.gnome.org/show_bug.cgi?id=793738
res is a boolean variable which is defined in the function scope and
redefined, with no reason, in the loop scope. This patch removes the
redefinition.
https://bugzilla.gnome.org/show_bug.cgi?id=793592
...and replace all checks for RECORD in GstRTSPMedia which are really
for "sender-only". This way the code becomes more generic and introducing
support for onvif-backchannel later on will require no changes in
GstRTSPMedia.
They are wrong in the ONVIF streaming spec. The backchannel should be
recvonly and the normal media should be sendonly: direction is always
from the point of view of the SDP offerer (the server) according to
RFC 3264.
This adds a new RTSP server, client, media-factory and media subclass
for handling the specifics of the backchannel. Ideally this later can be
extended with other ONVIF specific features.
The return value type is defined with G_DEFINE_POINTER_TYPE,
and gi emits the following warning:
Invalid non-constant return of bare structure or union; register as
boxed type or (skip)
This maps _new_empty() to _new(), which also makes RTSPToken()
work properly now. Since this API wasn't usable from bindings
before, this should hopefully be fine.
https://bugzilla.gnome.org/show_bug.cgi?id=787073
In the multicast case (as in test-multicast, not test-multicast2), the
address could be allocated/reserved (and thus set) already without
allocating the actual socket. We need to allocate the socket here still
instead of just claiming that it was already allocated.
See https://bugzilla.gnome.org/show_bug.cgi?id=791743#c2
SDP are now provided *before* the pipeline is fully complete. In order
to know whether a media is seekable or not therefore requires asking
the invididual streams.
API: gst_rtsp_stream_seekable
https://bugzilla.gnome.org/show_bug.cgi?id=790674
There is not need of adding fakesink elements to the media
pipeline in the dynamic-payloader case.
The media pipeline itself is dynamically updated with
the receiver and sender parts that are based on the client
transport information known after SETUP has been received.
Change-Id: I4e88c9b500c04030669822f0d03b1842913f6cb9
https://bugzilla.gnome.org/show_bug.cgi?id=790674
Media is complete when all the transport based parts are
added to the media pipeline. At this point ASYNC_DONE is
posted by the media pipeline and media is ready to enter
the PREPARED state.
Change-Id: I50fb8dfed88ebaf057d9a35fca2d7f0a70e9d1fa
https://bugzilla.gnome.org/show_bug.cgi?id=790674
If we still have some dynamic paylaoders which haven't posted
no-more-pads yet, don't go to PREPARED if one of the streams
blocked.
The risk was that we would end up not exposing/using all specified
streams.
The downside is that if you have _multiple_ _live_ _dynamic_ payloaders
then it will take a bit more time to start. But only if those 3
conditions are present.
https://bugzilla.gnome.org/show_bug.cgi?id=769521
If we have more than one dynamic payloader in the pipeline, we need
to wait until the *last* one emits 'no-more-pads' before switching
to PREPARED.
Failure to do so would result in a race where some of the streams
wouldn't properly be prepared
https://bugzilla.gnome.org/show_bug.cgi?id=769521
The initial pipeline does not contain specific transport
elements. The receiver and the sender parts are added
after PLAY.
If the media is shared, the streams are dynamically
reconfigured after each PLAY.
https://bugzilla.gnome.org/show_bug.cgi?id=788340
