Add a function to check if a specfic custom log level has been defined
for a topic.
We can use this to dynamically check if we need to do the connection debug
messages.
We can also get rid of the conn.* pattern hack to disable connection
messages by default.
Emit error if message construction is not holding appropriate locks or
runs from wrong thread, so that the flush may be running concurrently.
This generally causes corrputed messages to be sent silently, and can be
fairly hard to debug otherwise in client applications.
If no socket configuration specified for a server, create two sockets,
"CORENAME" and "CORENAME-manager" where CORENAME is the value computed
by get_server_name.
Add client key PW_KEY_SEC_SOCKET that indicates which socket the client used
to connect to the server.
This can be used by other modules as an access control mechanism.
For all the modules that include the private header we require that the
library and compiler versions match.
Otherwise we might end up poking into new or old fields that got moved or
changed in the private struct and crash.
See #3243
When we consumed all the buffer data, don't clear all the fds but only
those that were already consumed in the message. It is possible that we
already have fds for the next message and we don't want to discard
those.
Fixes some intermittend memory map errors.
Always first use the env var and then check the properties. So that
PIPEWIRE_CORE=pipewire-1 PIPEWIRE_REMOTE=pipewire-1 make run runs
everything on pipewire-1 sockets regardless of the config files.
Also PIPEWIRE_NODE always needs to be taken into account first.
Support fork-free readiness notifications.
Without this, a service supervisor that does not implement socket
activation has no way of knowing whether or not pipewire is ready to
accept connections on the socket.
s6 is the most popular service manager that supports this mechanism.
See here: https://skarnet.org/software/s6/notifywhenup.html
Init the compat types map a little earlier so that it is initialized
when we try to clear it on error.
Add the client listener earlier so that we can use the events to clean
up the map, source and connection.
When the client destroys the protocol-native module, the server
and the client are destroyed but the client is still reffed (not freed).
It will be unreffed after its messages are processed, after which point
it will be freed and removed from the server client_list that is already
destroyed.
Fix this by removing the client from the server list when it is
destroyed.
See #565
Add a refcount to resource and client so that we can keep them alive
while the native protocol is using them.
One problem might be that the protocol destroys the client or resource
while handling it and that would cause errors.
Fixes#565
Message footer should be handled before attempting to find the object
the main message is sent to / checking permissions, because it is not
aimed at a specific object. E.g. the registry generation updates should
be handled regardless of whether the main message is valid or not,
because the updates will not be re-sent.
Fixes registry generation updates sometimes going missing.
There's an assumption that marshaled messages consist of a single POD,
since we now tag on a footer after it. This is true for the
protocol-native implementations, which all wrap the message in a single
POD Struct.
To catch protocol-native implementation bugs here later, add assert that
marshaling produces a single POD.
Extend version 3 protocol with message footers, which are for passing
around global state data that is not addressed to a specific object.
The extension is backward compatible with previous v3 clients, and won't
e.g. result to error spam in logs.
The footer is a single SPA POD, appended after the main message POD.
Because both the protocol message and the message POD record their
length, it's possible to append trailing data. Earlier clients will
ignore any data trailing the message POD.
The footer POD contains a sequence [Id opcode, Struct {...}]*,
so there is room to extend with new opcodes later as necessary.
There are separate marshal/demarshal routines for messages aimed at
resources and proxies.
Most feature checks already use #ifdef, and do not care about
the value of the macro. Convert all feature checks to do that,
and simplify the meson build scripts by replacing
if cond
cdata.set('X', 1)
endif
with
cdata.set('X', cond)
This is more complicated than a normal module because we have two
logging topics: mod.protocol-native and conn.protocol-native for wire
messages. Because the latter use spa_debug (through spa_debug_pod) we
need to #define our way around so those too use the right topics.
Note that this removes the previous "connection" category, it is now
"conn.protocol-native" instead.
Leaving sockets in the home directories is bad form, so let's not do
this.
This effectively requires XDG_RUNTIME_DIR to be set for pipewire to
work - it is set correctly on most setups anyway and on custom setups
this needs to be addressed with a custom environment.
Fixes#1443
This also brings the advantage that all tools, examples, modules, components
can also be compiled standalone out-of-tree using libpipewire from the system
Do not return an error immediately if connect() fails with EAGAIN. Check
if it completed successfully with getsockopt() when the socket becomes
writable instead.
This is the way to handle non-blocking connect() by the book but after
testing it seems that the case when connect() fails with EAGAIN is when
the listen backlog is full on the server side and in that case the
server socket is closed. So even though connect() completes successfully
according to getsockopt() the client socket is no longer usable
(on_remote_data() will get both SPA_IO_OUT and SPA_IO_HUP in mask on the
first call after connect() returned EAGAIN).
If buffer type char[] is 4-byte aligned, higher 3-byte on char could be non-zero if data is not initialized, which make 'buffer[i] >> 4' larger than 0x0f.
Use type uint8_t[] on SEC_LABEL buffer to fix it.
The message structures returned by pw_protocol_native_connection_get_next
point to data that is contained in the buffer of the connection.
The data was invalidated when pw_protocol_native_connection_get_next was
called the next time, which made the connection loop non-reentrant, in
cases where it was re-entered from demarshal callbacks.
Fix this by allocating new buffers when reentering and stashing the old
buffers onto a stack. The returned message structure is also stored on
the stack to make lifetimes to match.
If a client becomes unbusy again, signal a resume event so that the
messages are processed in the next mainloop iteration. This gives the
current iteration time to perform cleanups if needed.
Remove the look hook and always do flushing with an IO_OUT event. Rework
some things so that we can flush right after processing input without
having to go through a loop iteration.
See #298
