Bug 1228724 - Suppress Valgrind leak when the window is maximized. r=karlt
authorDão Gottwald <dao@mozilla.com>
Tue, 22 Dec 2015 09:51:57 +0100
changeset 277122 63f78eaa94e88c96ba58068ab69dd7fd4d8b9cac
parent 277121 b96e844b0289c5363b5988214d6343b05a7996e8
child 277123 054fa50c23c5c0bee40ff3b389cd4711dc675ab9
push id16738
push userdgottwald@mozilla.com
push dateTue, 22 Dec 2015 08:52:16 +0000
treeherderfx-team@63f78eaa94e8 [default view] [failures only]
perfherder[talos] [build metrics] [platform microbench] (compared to previous push)
reviewerskarlt
bugs1228724
milestone46.0a1
Bug 1228724 - Suppress Valgrind leak when the window is maximized. r=karlt
build/valgrind/x86_64-redhat-linux-gnu.sup
--- a/build/valgrind/x86_64-redhat-linux-gnu.sup
+++ b/build/valgrind/x86_64-redhat-linux-gnu.sup
@@ -35,29 +35,27 @@
    Bug 1187649
    Memcheck:Leak
    match-leak-kinds: definite
    fun:realloc
    fun:FcPatternObjectInsertElt
    ...
 }
 # Leaks due to either Gtk+3 or cairo, but Gecko is not directly involved with
-# those cairo interactions.
+# those cairo interactions.  One suspected cause is Gecko not closing the
+# display to work around a bug in old Gtk+3 versions.  See also bug 1228724.
 {
    Bug 1187649
    Memcheck:Leak
    match-leak-kinds: definite
    fun:malloc
    fun:_cairo_freelist_alloc
    fun:_cairo_xlib_display_queue_resource
    fun:_cairo_xlib_surface_finish
    ...
-   fun:gtk_widget_realize
-   fun:_ZN8nsWindow6CreateEP9nsIWidgetPvRKN7mozilla3gfx12IntRectTypedINS4_12UnknownUnitsEEEP16nsWidgetInitData
-   ...
 }
 # The three following leaks are deep in Gtk+3, and it doesn't seem we're doing
 # anything wrong on our end with the container objects. Those suppressions
 # are purposefully verbose so as to avoid them catching actual leaks due to
 # Gecko code.
 # Note: valgrind doesn't support more than 24 elements in a suppression stack,
 # which explains why the second has an ellipsis above g_slice_alloc.
 {