Running gnome-shell under valgrind, I saw the attached invalid write.
Basically we can destroy a window during event processing, and the old
window_remove_filters simply called g_free() on the filter, ignoring
the refcount. Then later in event processing we call filter->refcount--,
which is writing to free()d memory.
Fix this by centralizing list mutation and refcount handling inside
a new shared _gdk_window_filter_unref() function, and using that
everywhere.
==13876== Invalid write of size 4
==13876== at 0x446B181: gdk_event_apply_filters (gdkeventsource.c:86)
==13876== by 0x446B411: _gdk_events_queue (gdkeventsource.c:188)
==13876== by 0x44437EF: gdk_display_get_event (gdkdisplay.c:410)
==13876== by 0x446B009: gdk_event_source_dispatch (gdkeventsource.c:317)
==13876== by 0x4AB7159: g_main_context_dispatch (gmain.c:2436)
==13876== by 0x4AB7957: g_main_context_iterate.clone.5 (gmain.c:3087)
==13876== by 0x4AB806A: g_main_loop_run (gmain.c:3295)
==13876== by 0x8084D6B: main (main.c:722)
==13876== Address 0x1658bcac is 12 bytes inside a block of size 16 free'd
==13876== at 0x4005EAD: free (vg_replace_malloc.c:366)
==13876== by 0x4ABE515: g_free (gmem.c:263)
==13876== by 0x444BCC9: window_remove_filters (gdkwindow.c:1873)
==13876== by 0x4454BA3: _gdk_window_destroy_hierarchy (gdkwindow.c:2043)
==13876== by 0x447BF6E: gdk_window_destroy_notify (gdkwindow-x11.c:1115)
==13876== by 0x43588E2: _gtk_socket_windowing_filter_func (gtksocket-x11.c:518)
==13876== by 0x446B170: gdk_event_apply_filters (gdkeventsource.c:79)
==13876== by 0x446B411: _gdk_events_queue (gdkeventsource.c:188)
==13876== by 0x44437EF: gdk_display_get_event (gdkdisplay.c:410)
==13876== by 0x446B009: gdk_event_source_dispatch (gdkeventsource.c:317)
==13876== by 0x4AB7159: g_main_context_dispatch (gmain.c:2436)
==13876== by 0x4AB7957: g_main_context_iterate.clone.5 (gmain.c:3087)
https://bugzilla.gnome.org/show_bug.cgi?id=637464
The old functions to get core pointer and devices list are gone as
well. This slice is entirely replaced internally by multidevice
handling and may just go.
The xi2 device manager now handles slaves being detached and/or
attached to a master.
gdk_device_list_slaves() has been added so it is possible to
know how slaves relate with masters. The other backends (X11 and not)
don't neeed to to anything special here since their hierarchy is
fully flat.
When the slave device changes, the master takes the shape of the
new one, modifying its axes, this signal is more useful to catch
this situation than the n-axes property
This function may be used to know the hardware device that triggered
an event, it could resort to the master device in the few cases there's
not a direct hardware device to relate to the event (i.e.: crossing events
due to grabs)
gdk_enable_multidevice() has been replaced with gdk_disable_multidevice(),
so applications may call that function if they want to go back at the
previous behavior.
There would be usually little reasons to call that function, unless the
application is doing X calls itself that count on old fashioned core
devices.
Scanner doesn't detect this case correctly because it's a typedef;
work around it here by explicitly specifying that it's caller
allocates. See bug 636393 for the scanner issue.
One less magic function. Also refactored it to make it easier to
implement. It now returns TRUE if it beeped and FALSE if it failed to do
so. A default implementation exists that just returns FALSE for all the
backends that can't beep windows (read: everything but X11 with XKB -
and why on earth do keyboard libs implement beeping?)
Trying to get rid of all the _gdk_windowing_something() functions that
we expect backends to magically know about and instead put them in a
proper interface (mostly GdkWindowImplClass).
An event filter may add or remove filters itself. This patch does
two things to address this case. The first is to take a temporary
reference to the filter while it is being used. The second is
to wait until after the filter function is run before determining
the next node in the list to process. This guards against
changes to the next node. It also does not run functions
that have been marked as removed. Though I'm not sure if this
case can arise.
https://bugzilla.gnome.org/show_bug.cgi?id=635380
Since what we are doing is turning an icon with alpha into a
no-alpha icon + mask for legacy window managers, it makes more sense
to use the system visual than the window's visual, which might
be ARGB.
https://bugzilla.gnome.org/show_bug.cgi?id=634821
gdk_settings_map needs to be updated when gdk_settings_names is changed,
and it's easier to add the setting at the end, so we don't need to
recalculate everything.
Really fixing:
https://bugzilla.gnome.org/show_bug.cgi?id=634697
In this case, gnome-terminal sets an RGBA visual on its window,
and we need to be careful when creating the icon pixmap, to create
the pixmap with the same depth as the visual, or we risk a BadMatch
from XRenderCreatePicture deep inside cairo.
Allow -1 for the hotspot coordinates in gdk_cursor_new_from_pixbuf,
if the pixbuf contains the x_hot/y_hot options with appropriate values.
Bug #632140.
In particular, the following functions are gone:
- gdk_screen_get_default_colormap()
- gdk_screen_set_default_colormap()
- gdk_screen_get_system_colormap()
- gdk_screen_get_rgba_colormap()
This way, we can specify a colormap when constructing windows. And ew
must do that to avoid BadMatch from XCreateWindow when we use a
different visual.
Also add a priv pointer to GdkVisual and use it for the GdkVisualPrivate
structure. Then Make GdkVisualPrivate actually private to
gdkvisual-x11.c and make other callers use proper function calls to
access it.
Direct and TrueColor visuals don't alloc colors, so they don't need to
fiddle with colormaps. Just copy the code that computes the pixel value
from gdkcolor-x11.c and use it. For other visual types, don't set the
background color and fallback to background = None.
No more GdkPixmap to store the icon and its mask, but instead use cairo
surfaces. Also render the icon into the surfaces using Cairo instead of
gdk_pixbuf_render_threshold_alpha().
Now the window background is a cairo_pattern_t. The backends will try to
set this as good as they can on the windowing system, but no guarantees
are made on wether the windowing system supports the pattern.
Also gets rid of GDK_NO_BG as undefined behavior is not a good idea to
support, and GDK_NO_BG effectively made the window's contents undefined.
It wasn't effectively used in GTK anyway.
Basically copies the code for setting the WM icon hint from GtkWindow to
GdkWindow. This achieves the following:
- Putting this X11 specific code into the X11 backend
- Enables removal of gdk_window_set_icon()
- Gets rid of Pixmap/Bitmap usage outside of GDK.
trap->end_sequence is the first serial for which we don't
ignore errors anymore, so we know the trap is dead if
end_sequence <= processed_serial.
Bug 629608
Currently fprintf(stderr, ...) is used for X error and X IO errors
(unless compiled with debugging, in which case g_error() is used for
X errors.)
But if an application is redirecting log messages, we really want
X errors to appear in that same log, so it's better to use a g_logv()
variant.
Though g_warning() isn't really appropriate for "lost connection to the
server", G_LOG_LEVEL_INFO isn't part of the "prefixed log levels"
so will produce output without the application name and PID.
https://bugzilla.gnome.org/show_bug.cgi?id=630216
When an error occurs with nested traps in place, only the innermost
trap should have the error code stored in it; outer traps are
shielded by the inner trap.
https://bugzilla.gnome.org/show_bug.cgi?id=629608
* don't lose track of previous X error handler
if nested traps are pushed
* free any remaining traps when display
is finalized
Test will fail unless bug 630032 is closed so
gdk_display_close() works.
https://bugzilla.gnome.org/show_bug.cgi?id=630033
* _gdk_device_set_associated_device() did not allow NULL device
* GdkDisplay should dispose device manager to avoid devices
trying to touch the display in finalize
* GdkDeviceManagerXI did not ref devices in id hash
* GdkDisplayX11 did not ref devices in ->input_devices
* add per-display gdk_x11_display_error_trap_push()
(X11-specific because gdk_error_trap_push() probably
should have been)
* make gdk_error_trap_push() handle only GDK displays
not displays opened without a GDK wrapper
* make gdk_error_trap_pop() and gdk_x11_display_error_trap_pop()
automatically sync only if needed, so manual gdk_flush() is not
required
* add gdk_error_trap_pop_ignored() which just asynchronously
ignores errors, so never needs to sync
* add G_GNUC_WARN_UNUSED_RESULT to plain pop(), because
if you use plain pop() and don't need the return value,
the async gdk_error_trap_pop_ignored() should be used
instead. This results in lots of warnings to clean
up in a later patch.
The main objective here was to avoid the need to sync just
to ignore an error. Now, syncing is automatic, and only
happens when we need to know the error code.
https://bugzilla.gnome.org/show_bug.cgi?id=629608
Add the composite overlay window to the cache, as this can be a reasonable Xdnd proxy as well.
This is only done when the screen is composited in order to avoid mapping
the COW. We assume that the CM is using the COW (which is true for pretty
much any CM currently in use).
https://bugzilla.gnome.org/show_bug.cgi?id=601731
This was the last exported variable; it wasn't multihead safe,
and there's easy replacement with gdk_display_get_default().
Also drop the GDK_DISPLAY() macro which was just a wrapper around
the variable.
The keysyms create a lot of potential namespace conflicts for
C, and are especially problematic for introspection, where we take
constants into the namespace, so GDK_Display conflicts with GdkDisplay.
For C application compatiblity, add gdkkeysyms-compat.h which uses
the old names.
Just one user in GTK+ continues to use gdkkeysyms-compat.h, which is
the gtkimcontextsimple.c, since porting that requires porting more
custom Perl code.
Bug 628656 - _gdk_windowing_get_startup_notify_id memory leak
get_display_name() returns a newly allocated string, which was being fed
directory info a g_strdup_printf() call.
While X11 surfaces can be resized, this is not the case for Quartz
surfaces. Instead of resizing we will invalidate the surface instead.
By giving _gdk_windowing_set_cairo_surface_size() a boolean return
value, we can signal back whether or not resizing was possible. If not
possible, we invalidate the surface.
The window move code needs special attention for multiple reasons:
- invalid areas for expose events need to be modified
- self-copy is not supported by Cairo
- in X11, copying from an overlapped Window might cause unexposed areas
to be copied in, spo expose events for those need to be generated.
This was all special cased in various parts of the code. By making it an
explicit vfunc, we can work around it.
Includes fixing all callers to use the cairo region API instead. This is
usually just replacing the function names, the only difference is
gdk_region_get_rectangles() being replaced by
cairo_region_num_rectangles() and cairo_region_get_rectangle() which
required a bit more work.
https://bugzilla.gnome.org/show_bug.cgi?id=613284
Only changes in headers: GdkRegion and GdkRectangle are typedeffed to
cairo_region_t and cairo_rectangle_int_t respectively. The region type
was opaque anyway so it doesn't matter and the rectangle types are
identical.
https://bugzilla.gnome.org/show_bug.cgi?id=613284
This function makes a better replacement for
gdk_display_get_core_pointer(), wherever it might yet be needed, for
XI2 resorts to XIGetClientPointer(), for the others return the only
core pointer.
https://bugzilla.gnome.org/show_bug.cgi?id=621685
Conceivably this GdkEventTranslator may still be asked for events such as
NoExpose or GraphicsExpose for pixmaps. It is not meant to translate such events,
but should at least behave with them.
We now exit early from gdk_window_register_dnd() to avoid crashing if the
window type is GDK_WINDOW_OFFSCREEN and does not support dnd operations.
This makes it possible to use any dnd-enabled widgets, such as GtkEntry,
within a GtkOffscreenWindow.
In particular, rename
- libraries to lib*-3.0.so
- pc files to *-3.0.pc
- include paths to /usr/include/gtk-3.0/*
- module paths to /usr/lib/gtk-3.0/*
- rc files names to gtk-3.0/gtkrc
- commandline utilities to *-3.0
- adjust documentation
Also change the install location for unix-print headers to
/usr/include/gtk-3.0/unix-print/gtk.
Bug 615853 - BadMatch when pressing keyboard volume keys while pointer in
secondary X screen
_gdk_windowing_get_shape_for_mask() is using the default screen, not taking
into account that its GdkBitmap could have been created for a different one,
causing BadMatch errors.
It doesn't make sense to determine an index into an array, if
you are going to sort the array afterwards...pointed out in
bug 615128 by Florian Scandella.
Currently gdk_screen_get_primary_monitor just returns the first monitor,
in this case but both the panel and now gnome-shell use an additional
heuristic to prefer LVDS as primary if present.
Move this heuristic to gdk_screen_get_primary_monitor to avoid duplicating
it all over the place.
The fallback heuristic is also used when no primary output is set.
https://bugzilla.gnome.org/show_bug.cgi?id=614894
It turns out that my attempt at handling Super, Hyper and Meta better
is causing problems, mostly because Alt and Meta are commonly colocated
in the modmap, and apps do a check for the Alt modifier regularly.
See e.g bug 607697.
This patch changes GDK to add all matching virtual modifiers in
the state field of the key event. The corresponding GTK+ change makes
use of a new GdkKeymap function to map virtual modifiers back to
real modifiers and detect conflicts while doing so.
This should fix bug 603190 and bug 427409.
The Gdk-custom.c file in gir-repository contained a number of
introspection annotations. Merge those into the GDK source files.
Some documentation was moved from the tmpl/ files to accomodate
the addition of annotations.
https://bugzilla.gnome.org/show_bug.cgi?id=592279
The fix in 786b589d95 for the
"Cannot click buttons more than once without moving the mouse cursor"
did not correctly handle setups with multiple screens in one display.
We need to handle the case where the first XQueryPointer returns
a different root window than the default one.
This fixes bug 597386 (agaion)
Either g_type_register_static_simple (used by G_DEFINE_TYPE_EXTENDED)
and G_IMPLEMENT_INTERFACE use automatic variables for GTypeInfo and
GInterfaceInfo structs, while tutorials and source code often use
static variables. This commit consistently adopts the former method.
https://bugzilla.gnome.org/show_bug.cgi?id=600158
Backends that support native window background setting (and that clears
new window areas to this color/pixmap) should set this to true.
Currently only X11 supports this.
Fixed the prototypes of MyEnhancedXkbTranslateKeyCode() and
translate_keysym() to take a gint* because those functions treat
these arguments as an int, so we get around the need to cast.
We need to do a final XQueryWindow to get the coordinates inside the
windows rather than in the parent window.
This fixes bug #597386, "Cannot click buttons more than once...", which
failed due to the grab tracking stuff getting the wrong coordinates as
per the above.
Thanks gcc :
gdkwindow-x11.c:1731: warning: passing argument 3 of
'XReconfigureWMWindow' makes integer from pointer without a cast
/usr/include/X11/Xlib.h:1871: note: expected 'int' but argument is of
type 'struct GdkScreen *'
Signed-off-by: Pascal Terjan <pterjan@mandriva.com>
