In most cases the pointer equality test is sufficient. However, in
some cases, depending on how things are split across shared objects,
we can end up with multiple instances of the interface metadata
constants. So if the pointers match, the interfaces are equal, if
they don't match we have to compare the interface names.
On the client side where we queue up multiple events before dispatching, we
need to look up the receiving proxy and argument proxies immediately before
calling the handler. Between queueing up multiple events and eventually
invoking the handler, previous handlers may have destroyed some of the
proxies.
The only way to make the global object listener interface thread safe is to
make it its own interface and make different listeners different wl_proxies.
The core of the problem is the callback we do when a global show up or
disappears, which we can't do with a lock held. On the other hand we can't
iterate the global list or the listener list without a lock held as new
globals or listeners may come and go during the iteration.
Making a copy of the list under the lock and then iterating after dropping
the lock wont work either. In case of the listener list, once we drop the
lock another thread may unregister a listener and destroy the callbackk
data, which means that when we eventually call that listener we'll pass it
free memory and break everything.
We did already solve the thread-safe callback problem, however. It's what
we do for all protocol events. So we can just make the global registry
functionality its own new interface and give each thread its own proxy.
That way, the thread will do its own callbacks (with no locks held) and
destroy the proxy when it's no longer interested in wl_registry events.
We used to special case this because of the untyped new-id argument in
the bind request. Now that the scanner can handle that, we can
remove the special case.
Switching to the generated stubs does bring an API change since we now
also take the interface version that the client expects as an argument.
Previously we would take this from the interface struct, but the
application may implement a lower version than what the interface struct
provides. To make sure we don't try to dispatch event the client
doesn't implement handlers for, we have to use a client supplied version
number.
This makes the scanner generate the code and meta data to send the
interface name and version when we pass a typeless new_id. This way, the
generic factory mechanism provided by wl_display.bind can be provided by
any interface.
This introduces wl_event_queue, which is what will make multi-threaded
wayland clients possible and useful. The driving use case is that of a
GL rendering thread that renders and calls eglSwapBuffer independently of
a "main thread" that owns the wl_display and handles input events and
everything else. In general, the EGL and GL APIs have a threading model
that requires the wayland client library to be usable from several threads.
Finally, the current callback model gets into trouble even in a single
threaded scenario: if we have to block in eglSwapBuffers, we may end up
doing unrelated callbacks from within EGL.
The wl_event_queue mechanism lets the application (or middleware such as
EGL or toolkits) assign a proxy to an event queue. Only events from objects
associated with the queue will be put in the queue, and conversely,
events from objects associated with the queue will not be queue up anywhere
else. The wl_display struct has a built-in event queue, which is considered
the main and default event queue. New proxies are associated with the
same queue as the object that created them (either the object that a
request with a new-id argument was sent to or the object that sent an
event with a new-id argument). A proxy can be moved to a different event
queue by calling wl_proxy_set_queue().
A subsystem, such as EGL, will then create its own event queue and associate
the objects it expects to receive events from with that queue. If EGL
needs to block and wait for a certain event, it can keep dispatching event
from its queue until that events comes in. This wont call out to unrelated
code with an EGL lock held. Similarly, we don't risk the main thread
handling an event from an EGL object and then calling into EGL from a
different thread without the lock held.
The update callback for the file descriptors was always a bit awkward and
un-intuitive. The idea was that whenever the protocol code needed to
write data to the fd it would call the 'update' function. This function
would adjust the mainloop so that it polls for POLLOUT on the fd so we
can eventually flush the data to the socket.
The problem is that in multi-threaded applications, any thread can issue
a request, which writes data to the output buffer and thus triggers the
update callback. Thus, we'll be calling out with the display mutex
held and may call from any thread.
The solution is to eliminate the udpate callback and just require that
the application or server flushes all connection buffers before blocking.
This turns out to be a simpler API, although we now require clients to
deal with EAGAIN and non-blocking writes. It also saves a few syscalls,
since the socket will be writable most of the time and most writes will
complete, so we avoid changing epoll to poll for POLLOUT, then write and
then change it back for each write.
When integrating the wayland event-loop into another event-loop, we
currently have no chance of checking whether there are pending idle
sources that have to be called. This patch exports the
"dispatch_idle_sources()" call so other event loops can call this before
going to sleep. This is what wl_event_loop_dispatch() currently does so we
simply allow external event-loops to do the same now.
To avoid breaking existing applications, we keep the call to
dispatch_idle_sources() in wl_event_loop_dispatch() for now. However, if
we want we can remove this later and require every application to call
this manually. This needs to be discussed, but the overhead is negligible
so we will probably leave it as it is.
This finally allows to fully integrate the wayland-server API into
existing event-loops without any nasty workarounds.
Signed-off-by: David Herrmann <dh.herrmann@googlemail.com>
wl_client_add_resource() used to return no error even though the new
resource wasn't added to the client. This currently makes it very easy to
DOS weston by simply posting thousands of "create_surface" requests with
an invalid ID. Weston simply assumes the wl_client_add_resource() request
succeeds but will never destroy the surface again as the "destroy" signal
is never called (because the surface isn't linked into the wl_map).
This change makes wl_client_add_resource() return the new ID of the added
object and 0 on failure. Servers (like weston) can now correctly
immediately destroy the surface when this call fails instead of leaving
the surface around and producing memory-leaks.
Instead of returning -1 on failure and 0 on success, I made it return the
new ID as this seems more appropriate. We can directly use it when calling
it with new_id==0.
Signed-off-by: David Herrmann <dh.herrmann@googlemail.com>
There is really no need to increment "n" if we never read the value. The
do-while() loop overwrites the value before it is read the first time.
Signed-off-by: David Herrmann <dh.herrmann@googlemail.com>
Expose these to other files using wayland-private.h, so wayland-client.c
can walk NULLables properly.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
If a new object id arrives ensure that there is an empty array entry
created, otherwise we might get out of sync for new ids if object isn't
created by interface implementation.
Creation of new client resources was silently ignored when
wl_client_add_resource() was used on server side and new object id was out
of range.
An error is now send out to the client in such case.
Also changed error message in wl_client_add_object, since
wl_map_insert_at() returns -1 only at invalid new id.
I've found a bug during wayland exploration - if you make two
drag'n'drops in weston client example, dnd - weston crashes with
segfault. I've tried to investigate it and found a problem.
In function drag_grab_button we first call data_device_end_drag_grab,
which sets seat->drag_data_source to NULL. Then we remove
listener from list only if drag_data_source is not NULL.
So if client will not free wl_data_source and start another drag'n'drop,
after the first one. Then two wl_data_source structures will be
free'd on client exit (let's name them s1 and s2).
next and prev pointer of
wl_data_source.resource.destroy_signal.listener_list in both
wl_data_source structures will be seat->drag_data_source_listener,
but next and prev in seat->drag_data_source_listener.link point
to listener_list in s2.
So if you try to iterate over listener_list in s1
then you get drag_data_source_listener as first item and
(struct wl_listener *)(&s2.resource.destroy_signal.listener_list)
Iteration over that list occurs in
wl_resource_destroy->destroy_resource->wl_signal_emit->wl_signal_emit
and try to call function at address of wl_resource->client, so
weston segfaults there.
This makes wl_seat_set_keyboard similar to wl_seat_set_pointer in that
it's a no-op, if you try to set keyboard to NULL when it already is
NULL, instead of refusing to set it to NULL ever.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Most of the time it does not make sense to pass a NULL object, string, or array
to a protocol request. This commit adds an explicit “allow-null” attribute
to mark the request arguments where NULL makes sense.
Passing a NULL object, string, or array to a protocol request which is not
marked as allow-null is now an error. An implementation will never receive
a NULL value for these arguments from a client.
Signed-off-by: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Attempting to write anything longer into the embedded char
array would create a non-null-terminated string, and all
later reads would run off the end into invalid memory.
This is a hard limitation of AF_LOCAL/AF_UNIX sockets.
Attempting to write anything longer into the embedded char
array would create a non-null-terminated string, and all
later reads would run off the end into invalid memory.
This is a hard limitation of AF_LOCAL/AF_UNIX sockets.
Always unlink() the lock file before closing the file
descriptor for it. Otherwise, there is a race like this:
Process A closes fd, releasing the lock
Process B opens the same file, taking the lock
Process A unlinks the lock file
Process C opens the same file, which now no longer exists,
and takes the lock on the newly created lock file
Process B and C both 'own' the same display socket.
unlink()ing while holding the lock is effectively a better
way to release the lock atomically.
When the server send a new object ID, the client used to have to allocate
the proxy manually and without type-safety. We now allocate the proxy
in a client-side post-processing step on the incoming closure.
Provide a slot for keyboard modifier state inside wl_keyboard for
implementations to update, and use this to send wl_keyboard:;modifier
events whenever the keyboard or pointer focus changes.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
If the data source is destroyed, the corresponding offers may stay around for
a little longer (until the owning client destroys it). When the offer is
finally destroyed, we have to be careful to only remove the source
destroy listener if the source hasn't yet been destroyed.
Thanks to Martin Minarik for tracking down where the corruption happened.
