The GLib slice allocator has been obsoleted. Having a mixture of two
allocators can lead to hidden issues on systems that use different
implementations. Take the leap and replace everything with g_new/g_free.
Change-Id: I025c8383ef47b2c2472573360407fd6a6ca339b3
Special codec handler to support not forwarding (nor any processing
whatsoever) of particular payload types at all. Support this in the
kernel module as well.
Change-Id: If10227affa54307e1e9b448eadd0bf2bfc5774ba
Only use the codec handler's "kernelize" flag to determine whether a
particular payload type can be handled by the kernel or not. It's
supposed to be a reliable indicator, making checking the other flags
unnecessary.
Change-Id: I5fe58eded55ea973ed555e4376ee8200f112d162
Split function into one part handling the "target" (i.e. ingress)
portion of the forwarding chain, and one function handling just the
outputs.
Change-Id: I3766da3c4bc5caee4eb6bae8978f177a83cc231a
First step to refactoring the kernelize* functions. Keep all state in a
central struct that is passed to all functions. Requires an auxiliary
bool to facilitate a case distinction in kernelize_one, which can be
removed once refactoring is complete.
Change-Id: I5de8404d55d5b320ef5cb8cd27d5b684802fa7e7
This seems to be an acceptable and reliable way to detect RTCP
multiplexed with RTP, even if `a=rtcp-mux` wasn't advertised in the SDP.
Take the opportunity to clean up __streams_set_sinks() a bit by giving
the variables better names.
Change-Id: I0cdc5e4a544641591fc2aabca12fb11bab3453f7
Separate out the two distinct use cases of requesting a specific port,
and requesting (one or more) random ports.
Change-Id: Ifbe7725d8638d49f620c9f4b9445342d180c8770
Add helper functions wrapping around reserve_port which directly return
a socket_port_link object. This allows us to rework
__get_consecutive_ports and release_reserved_port to eliminate mixing
different port numbers in the same list, which makes for cleaner code.
Change-Id: Ia9808f71c854f6efd0e93ce8df4527c4aea311bd
Turns out that g_ptr_array_new_full or g_ptr_array_new_sized only
preallocates array members, but doesn't set the length.
Change-Id: I722e29a13c415f2a5ff62830dce60205f273b445
Add a simple function that acts as an encryption callback to just update
the ext_seq (index) of the egress SSRC context. The kernel module
already does this, but the daemon only did it when SRTP was involved.
This now tracks egress packet indexes in all cases.
Change-Id: I9460744de55ead4b05aceb322fd8482442ff2b41
Avoid repeated calls to the memory allocated for port pool list
management by picking out the list elements without freeing them,
storing them in the stream_fd object, and then returning them to the
list when the port is released.
Change-Id: I67cd5039e62e4d2965e85d7ba7f0454f08f40494
Make the macro return the appropriate pointer type, and make sure the
free function takes an argument of the same type. This also eliminates
some boilerplate type-casting code.
Change-Id: I3094271fa2c53ec93b9ff9f837d461cf422e0f12
under normal circumstances, the rtp sink enters this function first
and updates the reti->local.family var to AF_INET so that when the
function is called for an rtcp sink, it exits the function early.
However, if media is being blocked the rtcp sink is the first to
enter the function and does so with a NULL payload type. this NULL
check is therefore required to prevent a sefgault on the call to `assert`.
closes#1876
Change-Id: Id95cbbc9ae388f50dd51417c351f4a9ba09eb5f0
Use provided macros for initrialise str objects everywhere instead of
manually setting .s and .len. This allows for flexibility to extend str
objects.
Change-Id: I5ebca0bc82b348301c2c8e94b974c7acbf249b82
If a stream has been pushed to the kernel from anything other than RTP,
even though RTP is expected, we get a forwarding entries without any
SSRCs. This is valid, but once actual RTP is received, it needs to be
passed on to user space, so that SSRC contexts can be set up.
Possible fix for #1855
Change-Id: I51b82d3cf79cf66780fdde154bebe56e0f43174b
Use the __preferred_lists_for_family queues as storage for all
logical_intf objects instead of the __logical_intf_name_family_hash hash
table. Move cleanup of logical_intf objects to cleanup of the respective
queues. Use the hash table key free function to manage the key objects.
Change-Id: I3f9628fbde0c9c71dc6afcfc95bfd54a5b12d6fa
For typed hash tables, enforce the correct type in the arguments to the
hashing and equality functions.
Adapt existing affected callback functions and change their arguments
from void* to the respective types.
Add reverse casts to GHashFunc and GEqualFunc in instances where these
functions are used in non-typed hash tables (that should be converted at
a later point).
Add convenience macro to create typed wrapper functions for hash tables
that use "direct" hashing (i.e. the pointer value).
Add wrappers for existing GLib functions that have generic arguments so
that they can be used in typed hash tables.
Change-Id: I43bb32969208f4aae49584d95c0df8353df6e2a0
This only exists because of RTCP indexing issues, but with the index
being in shared memory now, we no longer need this.
Change-Id: Ib0a69214f24a7c1edec8aa53139212ee861a6c4d
The requirement to continuously update stats from the kernel has been
eliminates. Remove all relevant code.
Change-Id: Idc8093b64290b89ff3d015f3e61e4bbe25df6e23
We now have accurate "last packet" timestamps for both user and kernel
space. Since we iterate all calls anyway, we can use these timestamps to
update the gauges.
Change-Id: If0522c387ed1fba51058129ef37a41f45d0c0c8b
... and change to 32 bit.
Remove functions to retrieve kernel stats as they don't do anything any
more.
Add RTCP index information to /list output.
Change-Id: Ia087c8c3913daf83a0749d6a048682b7248eb38b
This field is duplicated with `ext_seq`, which is already kept in shared
memory and therefore preferable.
Also update all sequence number fields to 32-bit (down from 64-bit) as
there's no need for all these extra bits.
Add output seq information to /list.
Change-Id: I0b8d8dfdf5c007c6ca75e7b50a8c900c16be21ed
Use the already existing ext_seq (in shm) in place of srtp_index as
these carry duplicate information.
Kernel still tracks its own index in the crypto context at this point.
Change data type from 64-bit to (unsigned) int as we don't need that
many bits.
Change-Id: I2468c6a814964eb64a19d84c918d3a37c2494b66
Allocate per-payload-type stats buffers in shared bufferpool. Push these
to the kernel module for direct tracking of bytes/packets stats. This
eliminates the need to return these values in the /blist data and
continuously update the userspace stats.
The accomodate updating the payload-type tracker, we use the "last
received" timestamp from the kernel module together with the "last seen
payload type" to give similar results as before.
Change-Id: I524791a1940b5d1ff2d82716c3b7a262f64e106c
This obsoletes the need to retain the last seen kernel stats and track
updates to the counters in order to set call states.
Change-Id: Ie5338bd630b679af205a16933f847ccdcd6a477e
Keep two separate timestamps, one updated by userspace code only and the
other updated by kernel only. This way we can tell where the packet
processing happens. For code that wants to report only the last
timestamp regardless of which one of the two it is, we add a convenience
function that just returns the newer one.
Change-Id: Ib3af7aa55006d8b32e2bc3db4f8bfa5514c57e40
This leaves the TOS value as the only leftover to be not in shm.
The kernel_stats structures remain for the time being as they're needed
to determine whether kernel forwarding is done or not, as well as for
global stats.
Change-Id: I158f18098e018d2870b797f1a196baa03a0e0fb7
Allocate memory from bufferpool for per-stream stats. No functional
change, but it allows sharing these between kernel and user space.
Change-Id: I370a49e1d94bb91c7fd0a2bc7d00ba65f99c4f6a
Use the kernel/user shared memory to keep track of interface stats. This
eliminates the need to update the userspace stats counters with the
values from the kernel, and vice versa when updating the kernel streams.
Add a function to reverse map userspace mapped memory address to kernel
space addresses.
Change-Id: Iaa5f9488061a12103e57faf27b3979521778cea8
There isn't any immediate benefit to this, but it prepares the code for
use of shared memory for statistics.
Use the opportunity to switch accesses to these to relaxed memory order.
Change-Id: I585fef7579202179fbbcbc1b843d3bbe440a723b
Switch all memory buffers used for RTP I/O from generic stack or heap
allocated memory to the bufferpool implementation. Use a per-thread
bufferpool to minimise lock contention.
This commit is just a one-for-one swap and doesn't use the bufferpool's
reference counting semantics yet.
Change-Id: I9cba4ec97bd0afcd374bf6c0be2b608a46e73e57
To support asynchronous pollers which may hold references on underlying
sockets, let the poller close the socket after it has released its
references. This prevents cases of file descriptor re-use while an
underlying socket is still open.
Add reset_socket() to be used in place of close_socket() which does the
same thing except the actual closing of the socket.
Add poller_del_item_callback() for cases where more action than just
closing the file descriptor is needed.
Change-Id: Iefda1487ecb89263729120ecb964436dd79b2a0e
This has been broken for a while and nobody seems to be using it since
nobody complained. Remove it.
Change-Id: I114e7b1859ecd1982338c625f4523f372af3bbe8
If multiple pollers are in use, use a single poller per call instead of
assigning pollers round-robin to each socket used in the call.
Change-Id: Iec49bd9d2fbd75d947d6232bcccfdfe87c4c6d7c
The poller-per-thread feature was broken with a division by zero. Take
the opportunity to rework it and eliminate the poller_map object. Use a
simple array of pollers for media sockets, plus one global poller for
control sockets. In the regular case only one poller is created and
everything points to that poller. In the poller-per-thread case, one
poller per thread is created, plus one poller (also with its own single
thread) for control connections. All control sockets use the single
control poller, while all media sockets get assigned one poller from the
pool in a round-robin fashion.
closes#1801
Change-Id: Iae91a3e10b7206455c6df33b1a472254c700ce21
In the file implementations follow the rules:
1. Firstly goes the correlated header file, then one empty row.
2. Secondly go system headers, so in angle-brackets, then one empty row.
3. Thirdly, go custom header files, so in double quotes,
then one empty row.
4. If there is "xt_RTPENGINE.h", it's mentioned next, but separately,
then one empty row.
5. If there are pre-processor definitions, they are added.
6. And eventually at least one empty row before the code.
In some situations it's allowed to step aside from the rules,
when inclusions are dependent on each other, so on the sequence,
and also possibly on some inline objects definitions, but if possible
to follow the rules, it's being done.
Change-Id: Ie512a970e230fe202398656d1942e8874bb14cd9
Instead of using transcoding flag `_TRANSCODING`
on the monologue level, we have to use that on the media
level in order to properly reflect the level on which
transocding is being used, and also to be able to selectively
set this for specific media sessions.
Change-Id: I9a25dc7be24f80b2b6ada816448a67933c762d86
There's two flavours of this struct being in use, even though the
structs' signatures are the same. One contains socket_t objects, the
other contains stream_fd objects. Separate them out and be explicit
about which is which.
Change-Id: I5ef1d154cc442528149f69be2e6a02625a6c650d
Start using the media subscriptions model
(based on newly introduced `media_subscription` objects)
in scope of `stream_packet()`.
Change-Id: Ia8d9442810c4afc12b00cad8b1a354d66bb5051c
We always need to (re)set the ssrc_map_out SSRC, even for the input
context, as this SSRC is used to look up the receiving SSRC context,
where the appropriate RX stats are stored, including the payload type in
use.
Change-Id: I434c6df4761bf56e36c7267dd3a44b0bb9c9b852
Don't explicitly set the "non forwarding" flag just because an RTCP
stream is being kernelised. The distinction between RTP and RTCP,
together with zero-length outputs setting "non forwarding" automatically
takes care of it.
Change-Id: Ie363fc369b7ace21ed172a0ec141c83afd1ba21b
Make it possible for a looper thread function to break out of the loop
by returning an appropriate status code.
Change-Id: I22e7789270eed4bf3340e7dae941929de58700ea
Perform endpoint address check before doing ingress SSRC handling. We
don't want to act on an apparent SSRC change if the endpoint address
doesn't match and the packet should have been ignored.
Closes#1655
Change-Id: I251e23b1583b1916c7c4904cf40d3b37b3533117
As this is running in a separate thread now, rtpe_now needs to be set
excplicitly in each iteration.
Fix-up for 6fe9cc9
Change-Id: Icb1ec8a27389bb48ee893a9a29e4e1c11447d61c
To do the work more efficiently,
and not be dependent on the call_timer runs by poller,
we should move the ports iterations (stats update from the kernel)
functionality to a separate thread, to make it faster and
not be dependent on what happens in the `call_timer` at all.
Since it has nothing to do with the call timers.
As an additional benefit: we unload the `call_timer` runner.
Change-Id: I511529ce504ef3d29f4e9d6d731ffd470d78d27a
Due to multiple threads polling all sockets for read events, it's
possible for one socket to receive a read event in one thread, then
immediately receive another read event in another thread, resulting in
two threads reading packets from the same socket at the same time.
While this is perfectly valid and correctly handled by mutex etc, it can
result in packets being processed out of order. In media passthrough
scenarios which don't do sequencing this can result in packets being
reordered.
Using a simple atomic counter we can ensure that only one thread is
reading from any one socket at a time.
Relevant to #1638
Change-Id: I406491d6ae5e13e618e153ba5463fd9169636016
To do the work more efficiently and not be dependent on
the `call_timer` runs by poller, we should move
the releasing of sockets to a separate thread, to make it
faster and not be dependent on what happens in the `call_timer`
at all. Since it has nothing to do with the call timers.
Since now we have two queues:
- thread scope (local): ports_to_release
- global one: ports_to_release_glob
`sockets_releaser()` uses the ports_to_release_glob,
meanwhile appending in the `call_timer()` happens using the
ports_to_release.
Change-Id: Iadd966ac895b2dd64f81269d4fdf5d83747fe0b7
Richard Fuchs
Peter Pulham
Donat Zenichev
Dennis Yurasov
Orgad Shaneh