Please describe the problem:
I was playing around with valgrind on nautilus and noticed:

==4407== 3932160 bytes in 1 blocks are still reachable in loss record 13246 of 13246
==4407==    at 0x3414A584: malloc (in /usr/lib/valgrind/vgpreload_addrcheck.so)
==4407==    by 0x34D9581B: IA__g_try_malloc (gmem.c:194)
==4407==    by 0x3488A8E5: IA__gdk_pixbuf_new (gdk-pixbuf.c:300)
==4407==    by 0x34890EA1: IA__gdk_pixbuf_scale_simple (gdk-pixbuf-scale.c:252)
==4407==    by 0x34250674: eel_background_pre_draw (eel-background.c:586)
==4407==    by 0x34251B8E: eel_background_get_pixmap_and_color
(eel-background.c:771)
==4407==    by 0x34251D40: eel_background_set_up_widget (eel-background.c:1518)
==4407==    by 0x34251EB6: eel_widget_background_changed (eel-background.c:1550)
==4407==    by 0x34D14EBC: IA__g_cclosure_marshal_VOID__VOID (gmarshal.c:77)
==4407==    by 0x34D09895: IA__g_closure_invoke (gclosure.c:437)

What is interesting is that we also hold the pixmap that is used on the X server
to display the image. So it looks like this 4 mb of memory is wasted.

The pixbuf is allocated in `eel_background_ensure_image_scaled'. However, it is
never freed (though a reference is kept).

In `eel_background_ensure_realized' the background is drawn to a pixmap with:

pixmap = gdk_pixmap_new (window, pixmap_width, pixmap_height, -1);
		gc = gdk_gc_new (pixmap);
		eel_background_pre_draw (background,  entire_width, entire_height);
		eel_background_draw (background, pixmap, gc,
				     0, 0, 0, 0,
				     pixmap_width, pixmap_height);
		g_object_unref (gc);
		background->details->background_pixmap = pixmap;

However, we don't free the pixbuf that was used to hold the image.

Steps to reproduce:


Actual results:


Expected results:


Does this happen every time?


Other information: