Dylan Noblesmith [Sat, 30 Jun 2012 19:50:09 +0000 (19:50 +0000)]
wayland-server: fix socket ownership race condition
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.
Dylan Noblesmith [Sat, 16 Jun 2012 22:47:42 +0000 (22:47 +0000)]
wayland-server: fix fd leak on error path
And restructure get_socket_lock() so it's clearer that it's
allocating a new file descriptor.
Uncovered by an upcoming test.
Dylan Noblesmith [Fri, 15 Jun 2012 22:01:06 +0000 (22:01 +0000)]
drop use of strerror()
It isn't thread-safe. Use the %m conversion specifier instead,
like the rest of the code already does.
Dylan Noblesmith [Fri, 15 Jun 2012 21:09:00 +0000 (21:09 +0000)]
wayland-server: reduce use of magic numbers
Make it clear what the significance of '5' and '113' actually is.
Also drop an unneeded function argument.
Kristian Høgsberg [Fri, 29 Jun 2012 02:01:58 +0000 (22:01 -0400)]
Allocate client proxy automatically for new objects
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.
Ander Conselvan de Oliveira [Thu, 28 Jun 2012 15:08:03 +0000 (18:08 +0300)]
protocol: Change wording of pointer.set_cursor()
Change the description of pointer.set_cursor() so that a client without
pointer focus but that currently owns the pointer surface can update the
hotspot using it.
Kristian Høgsberg [Wed, 27 Jun 2012 14:30:08 +0000 (10:30 -0400)]
wl_shell: Don't require that parent surfaces are wl_shell_surfaces
The set_transient and set_popup requests take a wl_shell_surface as
the parent surface argument. We don't need the parent surface to be
a shell surface and this restricts the types of surfaces we can use
the transient and popup surface types on.
Kristian Høgsberg [Mon, 25 Jun 2012 22:09:00 +0000 (18:09 -0400)]
configure.ac: Bump version to 0.94.90
Kristian Høgsberg [Fri, 22 Jun 2012 18:38:12 +0000 (14:38 -0400)]
Send modifier event to pointer focus when it changes too
Kristian Høgsberg [Fri, 22 Jun 2012 16:20:00 +0000 (12:20 -0400)]
Send keyboard modifier event using the keyboard resource
Daniel Stone [Fri, 22 Jun 2012 12:21:26 +0000 (13:21 +0100)]
Maintain keyboard modifier state in wayland-server
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>
Ander Conselvan de Oliveira [Tue, 19 Jun 2012 10:45:39 +0000 (13:45 +0300)]
wayland-cursor: Plug leak of filename on load_all_cursors_from_dir()
Daniel Stone [Mon, 18 Jun 2012 19:01:05 +0000 (20:01 +0100)]
Xcursor: Don't ignore unknown file types
XFS doesn't return the file type with opendir(), and instead only
returns it when you stat() the file itself.
c66f2602 introduced a check
to ensure that only files and symlinks were loaded, while not checking
for DT_UNKNOWN. Properly handle DT_UNKNOWN by just proceeding anyway.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Kristian Høgsberg [Mon, 18 Jun 2012 16:09:47 +0000 (12:09 -0400)]
data-device: Fix list corruption when the source goes away
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.
Kristian Høgsberg [Fri, 15 Jun 2012 20:09:39 +0000 (16:09 -0400)]
tests: Update test cases to new closure allocate convention
Kristian Høgsberg [Fri, 15 Jun 2012 15:09:36 +0000 (11:09 -0400)]
TODO: Remove cursor-as-regular-surface item
Ander Conselvan de Oliveira [Fri, 15 Jun 2012 14:27:33 +0000 (17:27 +0300)]
cursor: Add function for calculating current frame in cursor animation
Given a pointer to struct wl_cursor and an elapsed time in ms.,
wl_cursor_frame() will return the index of the cursor image the should
be displayed.
Ander Conselvan de Oliveira [Fri, 15 Jun 2012 14:27:32 +0000 (17:27 +0300)]
protocol: replace pointer.attach with pointer.set_cursor
Kristian Høgsberg [Wed, 13 Jun 2012 17:33:01 +0000 (13:33 -0400)]
TODO: Update to account for changes from Daniel
Kristian Høgsberg [Tue, 12 Jun 2012 21:48:19 +0000 (17:48 -0400)]
connection: Set closure->start in demarshal too
Kristian Høgsberg [Tue, 12 Jun 2012 21:45:25 +0000 (17:45 -0400)]
connection: Always malloc closure
This lets us allocate the closure just big enough and is a first step towards
a message queue.
Jonas Ådahl [Tue, 12 Jun 2012 21:57:33 +0000 (23:57 +0200)]
wayland-server: Add destroy signal to wl_seat
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Jonas Ådahl [Tue, 12 Jun 2012 21:57:32 +0000 (23:57 +0200)]
wayland-util: wl_list_insert_list() should accept empty lists
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Ander Conselvan de Oliveira [Fri, 8 Jun 2012 08:58:02 +0000 (11:58 +0300)]
wayland-cursor: add an option that changes the cursor theme directory
Some distros (e.g. gentoo) install cursor themes in non-standard
directories. Add option --with-icondir to configure.sh that sets the
directory in which to look for cursors.
Pekka Paalanen [Wed, 6 Jun 2012 11:30:19 +0000 (14:30 +0300)]
server: require XDG_RUNTIME_DIR
An arbitrary fallback to the current directory is only confusing.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Wed, 6 Jun 2012 11:30:18 +0000 (14:30 +0300)]
client: require XDG_RUNTIME_DIR
An arbitrary fallback to the current directory is only confusing.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Kristian Høgsberg [Tue, 5 Jun 2012 13:56:42 +0000 (09:56 -0400)]
Add a focus change signal for wl_pointer and wl_keyboard
Kristian Høgsberg [Mon, 4 Jun 2012 03:11:34 +0000 (23:11 -0400)]
TODO: Drop clipboard manager item
We have a reasonable prototype in weston now and we're going to
conclude that the persistent selection functionality is either
going to be part of the compositor (as for weston) or use a custom
protocol.
Kristian Høgsberg [Sun, 3 Jun 2012 21:30:12 +0000 (17:30 -0400)]
data-device: Don't implement data source through data offer object
The wl_data_source object used to specify the implementation for data
offers created for it. This means you need a data offer to retrieve the
data from the source, which makes it awkward to use in-process in a
compositor. Now we instead have three virtual functions that can be
connected to a protocol object or in-process data-sources such as an
X server proxy or clipboard.
Daniel Stone [Fri, 1 Jun 2012 11:13:37 +0000 (12:13 +0100)]
Add wl_keyboard::keymap event
With this event, the compositor generates one canonical keymap for all
clients, and then sends an mmap()able fd over the wire, rather than all
the clients generating possibly disparate keymaps.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Kristian Høgsberg [Fri, 1 Jun 2012 03:29:08 +0000 (23:29 -0400)]
data-device: Handle setting a NULL selection source
Kristian Høgsberg [Fri, 1 Jun 2012 03:28:31 +0000 (23:28 -0400)]
data-device: Emit selection_signal also when we lose the selection
Kristian Høgsberg [Fri, 1 Jun 2012 03:26:50 +0000 (23:26 -0400)]
data-device: Only send out NULL selection if we have a resource
If the current keyboard focus client doesn't have a resource for the
data_device, don't send out the selection event.
Daniel Stone [Wed, 30 May 2012 15:31:49 +0000 (16:31 +0100)]
Convert wl_pointer::axis to fixed
Which allows for smooth scrolling, among other things.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Daniel Stone [Wed, 30 May 2012 15:31:48 +0000 (16:31 +0100)]
Add key_state and button_state enums
Rather than hardcoding 0 and 1 everywhere.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Daniel Stone [Wed, 30 May 2012 15:31:47 +0000 (16:31 +0100)]
wl_keyboard: Add modifier event
This event sends the current keyboard modifier/group state from the
compositor to the client, allowing all clients to have a consistent view
of the keyboard state (e.g. current layout, Caps Lock, et al). It
should be sent after a keyboard enter event, and also immediately after
any key event which changes the modifier state.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Daniel Stone [Wed, 30 May 2012 15:31:46 +0000 (16:31 +0100)]
Always reset keyboard and pointer focus
If wl_pointer_set_focus or wl_keyboard_set_focus have been called before
a listener has been established for that seat and client combination,
the focus window will be set but the focus resource will be NULL. This
changes these functions to always attempt to search for the relevant
focus resource, allowing the resource to be set by calling
wl_keyboard_set_focus and wl_pointer_set_focus again when a listener has
been established.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Daniel Stone [Wed, 30 May 2012 15:31:45 +0000 (16:31 +0100)]
cursor: Fix warning with objdir != srcdir build
Make sure we find the right include paths.
krh: Edited to use $(top_builddir)
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Martin Minarik [Tue, 29 May 2012 15:38:51 +0000 (17:38 +0200)]
Wayland: logging: replace printf by wl_log
Martin Minarik [Tue, 29 May 2012 15:37:02 +0000 (17:37 +0200)]
Wayland: logging
The core libwayland libraries should not handle logging, only passing
the error messages to subscribed functions.
An application linked to libwayland-server or libwayland-client
will be able to set own functions (one per library) to handle error
messages.
Change in this series: make the wl_log return int, because
of compatibility with printf. It will return the number of bytes logged.
Pekka Paalanen [Tue, 29 May 2012 13:53:21 +0000 (16:53 +0300)]
tests: update gitignore
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Tue, 29 May 2012 13:53:20 +0000 (16:53 +0300)]
build: add uninstalled pkg-config files
Required for building apps against wayland libraries that have not been
installed.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Rob Bradford [Tue, 29 May 2012 13:10:30 +0000 (14:10 +0100)]
cursor: Link libwayland-cursor to libwayland-client
LDADD is only used when linking programs not libraries.
Spotted-by: Pekka Paalanen <ppaalanen@gmail.com>
Ander Conselvan de Oliveira [Tue, 29 May 2012 07:58:26 +0000 (10:58 +0300)]
data-device: implement drags with no data source for self-dnd
Properly handle a drag with no data source, i.e., don't crash and send
events only to the client that initiated the drag. This way a client can
do self drag and drop without offering anything to other clients.
Ander Conselvan de Oliveira [Fri, 25 May 2012 07:40:28 +0000 (10:40 +0300)]
data-device: fix crash when a client dies during a drag with an icon
The commit that split wl_input_device into wl_seat and friends changed
erroneously the drag icon destroy listener, causing it to operate into
an invalid pointer to a wl_seat.
Ander Conselvan de Oliveira [Thu, 24 May 2012 13:17:48 +0000 (16:17 +0300)]
wayland-cursor: load all cursors from a theme on wl_cursor_theme_load
Ander Conselvan de Oliveira [Thu, 24 May 2012 13:17:47 +0000 (16:17 +0300)]
wayland-cursor: remove enum wl_cursor_type
Defining a list of all cursors a theme, client or toolkit should have
is not the purpose of libwayland-cursor. The cursor type enum existed
for making lookups faster, but this kind of optmization belongs in the
toolkits.
Ander Conselvan de Oliveira [Wed, 23 May 2012 13:09:55 +0000 (16:09 +0300)]
shm: use mremap on pool resize
Yan Wang [Wed, 23 May 2012 06:14:30 +0000 (14:14 +0800)]
Wayland: Add missing extern "C"
Add extern "C" macro definition for C++.
Diego Viola [Wed, 23 May 2012 06:08:08 +0000 (02:08 -0400)]
fix typo
Kristian Høgsberg [Wed, 23 May 2012 02:41:21 +0000 (22:41 -0400)]
cursor: Fix unsigned/signed comparisons
Rob Bradford [Tue, 22 May 2012 17:48:14 +0000 (18:48 +0100)]
shm: Zero busy count for allocated buffers
This avoids a valgrind error like:
==31496== Conditional jump or move depends on uninitialised value(s)
==31496== at 0x407620: weston_buffer_post_release (compositor.c:928)
==31496== by 0x406AEB: weston_surface_attach (compositor.c:725)
==31496== by 0x409EB8: pointer_attach (compositor.c:2009)
==31496== by 0x34ECE05D63: ffi_call_unix64 (unix64.S:75)
==31496== by 0x34ECE05784: ffi_call (ffi64.c:486)
==31496== by 0x5674C4D: wl_closure_invoke (connection.c:770)
==31496== by 0x566ECCB: wl_client_connection_data (wayland-server.c:255)
==31496== by 0x56722F9: wl_event_source_fd_dispatch (event-loop.c:79)
==31496== by 0x5672C99: wl_event_loop_dispatch (event-loop.c:410)
==31496== by 0x56705FF: wl_display_run (wayland-server.c:1004)
==31496== by 0x40C775: main (compositor.c:2937)
==31496== Uninitialised value was created by a heap allocation
==31496== at 0x4A074CD: malloc (vg_replace_malloc.c:236)
==31496== by 0x5670EA7: shm_pool_create_buffer (wayland-shm.c:113)
==31496== by 0x34ECE05D63: ffi_call_unix64 (unix64.S:75)
==31496== by 0x34ECE05784: ffi_call (ffi64.c:486)
==31496== by 0x5674C4D: wl_closure_invoke (connection.c:770)
==31496== by 0x566ECCB: wl_client_connection_data (wayland-server.c:255)
==31496== by 0x56722F9: wl_event_source_fd_dispatch (event-loop.c:79)
==31496== by 0x5672C99: wl_event_loop_dispatch (event-loop.c:410)
==31496== by 0x56705FF: wl_display_run (wayland-server.c:1004)
==31496== by 0x40C775: main (compositor.c:2937)
Rob Bradford [Tue, 22 May 2012 17:48:13 +0000 (18:48 +0100)]
server: Initialise serial number counter for wl_display
This ends up propagating through and creating a valgrind error like:
==22573== Conditional jump or move depends on uninitialised value(s)
==22573== at 0x409E57: pointer_attach (compositor.c:1999)
==22573== by 0x34ECE05D63: ffi_call_unix64 (unix64.S:75)
==22573== by 0x34ECE05784: ffi_call (ffi64.c:486)
==22573== by 0x5674C45: wl_closure_invoke (connection.c:770)
==22573== by 0x566ECCB: wl_client_connection_data (wayland-server.c:255)
==22573== by 0x56722F1: wl_event_source_fd_dispatch (event-loop.c:79)
==22573== by 0x5672C91: wl_event_loop_dispatch (event-loop.c:410)
==22573== by 0x56705F4: wl_display_run (wayland-server.c:1003)
==22573== by 0x40C775: main (compositor.c:2937)
Ander Conselvan de Oliveira [Tue, 22 May 2012 12:39:41 +0000 (15:39 +0300)]
Introduce libwayland-cursor, a cursor helper library
The purpose of this library is to be the equivalent of libXcursor in
the X world. This library is compatible with X cursor themes and loads
them directly into an shm pool making it easy for the clients to get
buffer for each cursor image.
The code for handling the X cursor theme was taken from libXcursor. The
files cursor/xcursor.[ch] are a stripped down version of that library
containing only the interfaces necessary for implementing the wayland
counterpart.
Ander Conselvan de Oliveira [Tue, 22 May 2012 12:39:40 +0000 (15:39 +0300)]
shm: Add request for resizing pools
Tiago Vignatti [Mon, 21 May 2012 13:47:44 +0000 (16:47 +0300)]
protocol: use transient flags for activate or not surfaces
Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com>
Kristian Høgsberg [Wed, 16 May 2012 18:25:45 +0000 (14:25 -0400)]
doc: Update docs to refer to wl_seat instead of wl_input_device
Daniel Stone [Wed, 16 May 2012 17:44:40 +0000 (18:44 +0100)]
Convert wl_input_device to wl_seat (and friends)
wl_input_device has been both renamed and split. wl_seat is now a
virtual object representing a group of logically related input devices
with related focus.
It now only generates one event: to let clients know that it has new
capabilities. It takes requests which hand back objects for the
wl_pointer, wl_keyboard and wl_touch interfaces it exposes which all
provide the old input interface, just under different names.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Darxus [Mon, 14 May 2012 20:41:10 +0000 (16:41 -0400)]
Add missing alt attributes in publican docs.
Haven't figured out how to regenerate the html, so it's not tested.
Looks like remaining html errors will be fixed in the next publican
release: https://bugzilla.redhat.com/show_bug.cgi?id=788576
Pekka Paalanen [Tue, 15 May 2012 07:29:12 +0000 (10:29 +0300)]
tests: add noop to fixed-benchmark
I was just curious of how much the looping takes time without
conversion, so I added this.
My results on Intel(R) Core(TM) i5-2410M CPU @ 2.30GHz:
benchmarked noop: 1.876349827s
benchmarked magic: 2.245844470s
benchmarked div: 12.709085309s
benchmarked mul: 7.504838141s
Mul seems to take 15x the time magic does, cool!
Btw. the simple default cast of int32_t to double is slower than magic
for me, hence the use of union.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Kristian Høgsberg [Sat, 12 May 2012 03:36:20 +0000 (23:36 -0400)]
Optimize wl_fixed_t to/from double conversion functions
Kristian Høgsberg [Fri, 11 May 2012 15:27:22 +0000 (11:27 -0400)]
TODO: Strike the last xkbcommon issues
Kristian Høgsberg [Thu, 10 May 2012 18:13:05 +0000 (14:13 -0400)]
protocol: Specify that the fullscreen framerate parameter is also mHz
Kristian Høgsberg [Thu, 10 May 2012 16:09:56 +0000 (12:09 -0400)]
protocol: Back out accidentally committed wl_seat changes
Kristian Høgsberg [Thu, 10 May 2012 15:56:39 +0000 (11:56 -0400)]
protocol: Add documentation for wl_output
Kristian Høgsberg [Wed, 9 May 2012 15:26:30 +0000 (11:26 -0400)]
TODO: More updates
Kristian Høgsberg [Tue, 8 May 2012 18:49:44 +0000 (14:49 -0400)]
Update TODO
Daniel did fixed point coordinates.
Daniel Stone [Tue, 8 May 2012 16:17:26 +0000 (17:17 +0100)]
Input: Convert pointer co-ordinates to signed_24_8
This offers more precision when using devices like tablets, and also for
transformed surfaces.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Daniel Stone [Tue, 8 May 2012 16:17:25 +0000 (17:17 +0100)]
Add support for signed 24.8 decimal numbers
'fixed' is a signed decimal type which offers a sign bit, 23 bits of
integer precision, and 8 bits of decimal precision. This is exposed as
an opaque struct with conversion helpers to and from double and int on
the C API side.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Kristian Høgsberg [Tue, 8 May 2012 14:42:42 +0000 (10:42 -0400)]
event-loop: Delete fd from epoll when removing event source
Closing an fd will remove it from the epoll set only if it hasn't been
dup'ed. In other words, the fd is only removed from epoll when all file
descriptors referring to the open file has been close. We now dup
fd for fd sources, so we need to use EPOLL_CTL_DEL directly now.
Kristian Høgsberg [Tue, 8 May 2012 13:58:01 +0000 (09:58 -0400)]
tests: prefix event loop tests with event_loop_
Kristian Høgsberg [Tue, 8 May 2012 13:57:17 +0000 (09:57 -0400)]
tests: assert that mask is zero is post_dispatch callback
Kristian Høgsberg [Tue, 8 May 2012 13:55:55 +0000 (09:55 -0400)]
tests: Add timerfd test
Kristian Høgsberg [Tue, 8 May 2012 13:37:34 +0000 (09:37 -0400)]
tests: Add signal test case
Doesn't necessarily catch the signalfd bug just fixed, since that only
triggers when an uninitialized int is negative.
Kristian Høgsberg [Tue, 8 May 2012 13:30:49 +0000 (09:30 -0400)]
event-loop: Consolidate code for setting up new sources
Tiago Vignatti [Tue, 8 May 2012 11:58:02 +0000 (14:58 +0300)]
event-loop: fix conditional checking
Introduced in:
commit
80f4f0d5127ebc8d5e35969a29691cf61a79997d
Author: Jonas Ådahl <jadahl@gmail.com>
Date: Wed Mar 21 10:31:24 2012 +0100
Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com>
Kristian Høgsberg [Tue, 8 May 2012 03:45:20 +0000 (23:45 -0400)]
TODO: drop scanner namespace item
Not worh the trouble.
Kristian Høgsberg [Tue, 8 May 2012 03:41:48 +0000 (23:41 -0400)]
scanner: Generalize desc_dump() to handle hanging indents
Kristian Høgsberg [Tue, 8 May 2012 02:43:56 +0000 (22:43 -0400)]
scanner: Fix desc_dump() to not extend beyond column 72
Kristian Høgsberg [Tue, 8 May 2012 02:23:05 +0000 (22:23 -0400)]
scanner: Use indent() function in desc_dump()
Daniel Stone [Fri, 4 May 2012 10:21:20 +0000 (11:21 +0100)]
Use unsigned for key/button up/down state in grabs
Since we just use it as a simple boolean flag, might as well convert it
to unsigned.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Kristian Høgsberg [Thu, 3 May 2012 19:41:50 +0000 (15:41 -0400)]
Use class_ instead of class in protocol
This way we can include the generated code from C++.
Daniel Stone [Tue, 1 May 2012 19:30:12 +0000 (20:30 +0100)]
tests: Fix out-of-tree builds
Make sure we include any generated headers from src/ as well, like
wayland-version.h.
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Kristian Høgsberg [Wed, 2 May 2012 13:39:17 +0000 (09:39 -0400)]
Add shell_surface set_title and set_class requests
Nicolas Pouillon [Wed, 2 May 2012 06:46:47 +0000 (08:46 +0200)]
server: Post new globals to current clients
Globals are bursted to client on conntect. Now when new global
objects are added to the server state, existing clients also get
notified.
Kristian Høgsberg [Tue, 1 May 2012 17:50:23 +0000 (13:50 -0400)]
protocol: Move ping event down to other events
Kristian Høgsberg [Fri, 27 Apr 2012 15:31:07 +0000 (11:31 -0400)]
Add wl_proxy_get_id()
Kristian Høgsberg [Fri, 27 Apr 2012 15:28:06 +0000 (11:28 -0400)]
Add wl_client_get_object()
Kristian Høgsberg [Fri, 27 Apr 2012 15:26:57 +0000 (11:26 -0400)]
Drop stale prototype
Kristian Høgsberg [Thu, 26 Apr 2012 12:06:43 +0000 (08:06 -0400)]
tests: Just list tests under the TESTS variable
Kristian Høgsberg [Wed, 25 Apr 2012 14:45:48 +0000 (10:45 -0400)]
Merge remote-tracking branch 'pq/for-krh'
Kristian Høgsberg [Wed, 25 Apr 2012 14:12:21 +0000 (10:12 -0400)]
Add COPYING
Pekka Paalanen [Thu, 22 Mar 2012 12:24:18 +0000 (14:24 +0200)]
server: do not create wl_client if accept() failed
Spotted during wl_os_accept_cloexec() work.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Thu, 22 Mar 2012 12:16:10 +0000 (14:16 +0200)]
os: wrap accept4(SOCK_CLOEXEC)
Some system C libraries do not have SOCK_CLOEXEC, and completely miss
accept4(), too. Provide a fallback for this case.
This changes the behaviour: no error messages are printed now for
failing to set CLOEXEC but the file descriptor is closed.
The unit test for this wrapper is NOT included.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Thu, 22 Mar 2012 13:02:05 +0000 (15:02 +0200)]
os: wrap epoll_create
Some system C libraries do not have epoll_create1() nor EPOLL_CLOEXEC,
provide a fallback.
Add tests for the wrapper.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Wed, 21 Mar 2012 14:00:23 +0000 (16:00 +0200)]
os: wrap recvmsg(MSG_CMSG_CLOEXEC)
Some system C libraries do not have MSG_CMSG_CLOEXEC. This flag would
automatically set O_CLOEXEC flag on any received file descriptors.
Provide a fallback that does it manually. If setting CLOEXEC fails, the
file descriptor is closed immediately, which will lead to failures but
avoid leaks. However, setting CLOEXEC is not really expected to fail
occasionally.
Add tests for the wrapper. The setup is copied from connection-test.c.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Mon, 23 Apr 2012 11:36:32 +0000 (14:36 +0300)]
tests: silence warnings from pipe()
warning: ignoring return value of ‘pipe’, declared with attribute
warn_unused_result
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Mon, 23 Apr 2012 10:55:55 +0000 (13:55 +0300)]
os: wrap F_DUPFD_CLOEXEC
Some system C libraries do not have F_DUPFD_CLOEXEC. Provide a fallback.
Add tests for the new wl_os_dupfd_cloexec() wrapper.
Add per-wrapper call counters in os_wrappers-test.c. Makes it easier to
determine the minimum required number of wrapped calls.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Mon, 23 Apr 2012 09:10:45 +0000 (12:10 +0300)]
os: define SOCK_CLOEXEC on Linux
If it's not already defined, and we are on Linux, #define it. This gets
rid of a load of #ifdefs. This should also allow to use it when the
kernel supports it, but the libc does not define it.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>
Pekka Paalanen [Mon, 23 Apr 2012 08:37:40 +0000 (11:37 +0300)]
os: remove unneeded errno assignment
If socket() fails, it will return failure code. No need to reset errno
beforehand.
Signed-off-by: Pekka Paalanen <ppaalanen@gmail.com>