Add a limit to the amount of queued bytes or messages we allow on the watch.
API: GstRTSPConnection::gst_rtsp_watch_set_send_backlog()
API: GstRTSPConnection::gst_rtsp_watch_get_send_backlog()
return false from dispatch () if the read and write sockets have been
unset in tunnel_complete ()
Setting up HTTP tunnels causes segfaults since the watch for the second
connection is not destroyed anymore in tunnel_complete () and the connection
will still be used even though it is not valid anymore.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=686276
When closing the connection, unref the currently used sockets. This should close
them when not in use. We need to do this because else we cannot reconnect
anymore after a close, the connect function requires that the sockets are NULL.
Clear the GError after g_socket_connect tells us that the connection is pending.
If we don't do this, glib complains when we try to reuse the non-NULL GError
variable a little below.
Even if watch->messages->length is 0 there may still be some
data from a message that was only written partially at the
previous attempt stored in watch->write_data, so check for
that as well. We don't want to write data into the middle
of another message, which could happen when there wasn't
enough bandwidth.
https://bugzilla.gnome.org/show_bug.cgi?id=669039
Add private replacements for deprecated functions such as
g_mutex_new(), g_mutex_free(), g_cond_new() etc., mostly
to avoid the deprecation warnings. We'll change these
over to the new API once we depend on glib >= 2.32.
Replace g_thread_create() with g_thread_try_new().
Unlike linux, OSX wakes up select with POLLOUT (instead of POLLERR) when
connect() is done async and the connection is refused. Therefore always check
for the socket error state using getsockopt (..., SO_ERROR, ...) after a
connection attempt.
When closing rtspsrc the state change blocks until the polling in the
connection timeouts. This is because the second time we loop to read a
full message controllable is set to FALSE in the poll group, even though no
message is half read.
This can be avoided by not setting controllable to FALSE the poll group
unless we had begin to read a message.
Fixes#610916
We want to send the keealive message a little earlier than the timeout value
specifies. Scale this based on the value of the timeout instead of just assuming
5 seconds.
Because we should act before the rtsp server does a timeout, we
reduce the timeout-time with 5 seconds, this should be safe to always
keep te rtsp connection alive.
https://bugzilla.gnome.org/show_bug.cgi?id=633455
Catch more socket errors.
Rework how sockets are managed in the GSource, wake up the maincontext instead
of adding/removing the sockets from the source.
Add callback for when the tunnel connection is lost. Some clients (Quicktime
Player) close the POST connection in tunneled mode and reopen the socket when
needed.
See #612915
Be careful when allocating the amount of bytes specified in the Content-Length
because it can be an insanely huge value. Try to allocate the memory but fail
gracefully with a nice error when the allocation failed.
Use send() instead of write() so that we can pass the MSG_NOSIGNAL flags to
avoid crashing with SIGPIPE when the remote end is not listening to us anymore.
Fixes#601772
gstrtspconnection.c:gst_rtsp_connection_receive() can hang when an error occured
on a socekt. Fix this problem by checking for error on 'other' socket after poll
return.
Fixes#596159
The new API to send messages using GstRTSPWatch will first try to send the
message immediately. Then, if that failed (or the message was not sent
fully), it will queue the remaining message for later delivery. This avoids
unnecessary context switches, and makes it possible to keep track of
whether the connection is blocked (the unblocking of the connection is
indicated by the reception of the message_sent signal).
This also deprecates the old API (gst_rtsp_watch_queue_data() and
gst_rtsp_watch_queue_message().)
API: gst_rtsp_watch_write_data()
API: gst_rtsp_watch_send_message()
With gst_rtsp_connection_set_http_mode() it is possible to tell the
connection whether to allow HTTP messages to be supported. By enabling HTTP
support the automatic HTTP tunnel support will also be disabled.
API: gst_rtsp_connection_set_http_mode()
The error_full callback is similar to the error callback, but allows for
better error handling. For read errors a partial message is provided to
help an RTSP server generate a more correct error response, and for write
errors the write queue id of the failed message is returned.
Rewrote read_line() to support LWS (Line White Space), the method used by
RTSP (and HTTP) to break long lines. Also added support for \r and \n as
line endings (in addition to the official \r\n).
From RFC 2068 section 4.2: "Multiple message-header fields with the same
field-name may be present in a message if and only if the entire
field-value for that header field is defined as a comma-separated list
[i.e., #(values)]." This means that we should not split other headers which
may contain a comma, e.g., Range and Date.
Due to the odd syntax for WWW-Authenticate (and Proxy-Authenticate) which
allows commas both to separate between multiple challenges, and within the
challenges themself, we need to take some extra care to split these headers
correctly.
Do not abort message parsing as soon as there is an error. Instead parse
as much as possible to allow a server to return as meaningful an error as
possible.
Remove any existing Session and Date headers before adding new ones
when sending a request. This may happen if the user of this code reuses
a request (rtspsrc does this when resending after authorization fails).
Do not use sizeof() on an array passed as an argument to a function and
expect to get anything but the size of a pointer. As a result only the
first 4 (or 8) bytes of the response buffer were initialized to 0 in
auth_digest_compute_response() which caused it to return a string which
was not NUL-terminated...
gst_rtsp_watch_queue_data() is similar to gst_rtsp_watch_queue_message()
but allows for queuing any data block for writing (much like
gst_rtsp_connection_write() vs. gst_rtsp_connection_send().)
API: gst_rtsp_watch_queue_data()
The base64 decoding in fill_bytes() expected the size of the read data to
be evenly divisible by four (which is true for the base64 encoded data
itself). This did not, however, take whitespace (especially line breaks)
into account and would fail the decoding if any whitespace was present.
Previously the messages_sent() callback was only called for messages
which had a CSeq, which excluded all data messages. Instead of using the
CSeq as ID, use a simple index counter.
People might queue messages from a thread other than the thread in which
the main context which this watch is attached is iterated from, so use
a GAsyncQueue instead of a GList, so g_list_append() doesn't trample
over list nodes just freed in the other thread. This just fixes issues
I've had with gst-rtsp-server. We might need more locking in various
places here.
We were returning a pointer to a stack variable with the resolved hostname,
which doesn't work.
return a copy of the resolved ip address instead.
Fixes#575256.
