[cairo] Multithreaded cairo-test

[Behdad Esfahbod]

On Fri, 2007-03-23 at 18:58 -0400, Chris Wilson wrote:
> I wrote: 
> > The first causality exposed by multi-threaded cairo-test is the
> > _cairo_xlib_screen_list which lacked reference counting of the
> > _cairo_xlib_screen_info_t.
> 
> Not only that but later, a recursive deadlock was exposed:
> 
> ==25095== Deadlock[recursive] for mutex 0x409130C at:
> ==25095==    at 0x401F478: pthread_mutex_lock (ld_preload.c:51)
> ==25095==    by 0x4063ABE: _cairo_xlib_screen_info_destroy (cairo-xlib-screen.c:284)
> ==25095==    by 0x4062E7B: _cairo_xlib_surface_finish (cairo-xlib-surface.c:345)
> ==25095==    by 0x403E402: cairo_surface_finish (cairo-surface.c:471)
> ==25095==    by 0x403E4AF: cairo_surface_destroy (cairo-surface.c:396)
> ==25095==    by 0x4041C49: _cairo_pattern_release_surface (cairo-pattern.c:1630)
> ==25095==    by 0x4062144: _cairo_xlib_surface_composite (cairo-xlib-surface.c:1419)
> ==25095==    by 0x403D5FD: _cairo_surface_composite (cairo-surface.c:1092)
> ==25095==    by 0x403A8D4: _cairo_scaled_font_show_glyphs (cairo-scaled-font.c:1136)
> ==25095==    by 0x403F828: _cairo_surface_old_show_glyphs_draw_func (cairo-surface-fallback.c:901)
> ==25095==    by 0x403F2B2: _clip_and_composite (cairo-surface-fallback.c:391)
> ==25095==    by 0x403F6C5: _cairo_surface_fallback_show_glyphs (cairo-surface-fallback.c:952)
> ==25095==    by 0x403DE6D: _cairo_surface_show_glyphs (cairo-surface.c:1806)
> ==25095==    by 0x402FDAE: _cairo_gstate_show_glyphs (cairo-gstate.c:1595)
> ==25095==    by 0x4029AF0: cairo_show_text (cairo.c:2825)
> ==25095==    by 0x804A203: draw (glyph-cache-pressure.c:80)
> ==25095==    by 0x804B714: cairo_test_targets (cairo-test.c:293)
> ==25095==    by 0x426631A: start_thread (in /lib/tls/i686/cmov/libpthread-2.5.so)
> ==25095==    by 0x436F3ED: clone (in /lib/tls/i686/cmov/libc-2.5.so)
> ==25095== Mutex 0x409130C was taken by thread 9 at:
> ==25095==    at 0x401F4F2: pthread_mutex_lock (ld_preload.c:54)
> ==25095==    by 0x40639F1: _cairo_xlib_close_display (cairo-xlib-screen.c:300)
> ==25095==    by 0x418E995: XCloseDisplay (ClDisplay.c:71)
> ==25095==    by 0x804BE2B: cleanup_xlib (cairo-boilerplate.c:1038)
> ==25095==    by 0x804B23F: cairo_test_targets (cairo-test.c:355)
> ==25095==    by 0x426631A: start_thread (in /lib/tls/i686/cmov/libpthread-2.5.so)
> ==25095==    by 0x436F3ED: clone (in /lib/tls/i686/cmov/libc-2.5.so)
> ==25095== which waits on mutex 0x80F9CEC, taken by us, thread 10, at:
> ==25095==    at 0x401F4F2: pthread_mutex_lock (ld_preload.c:54)
> ==25095==    by 0x403DDD0: _cairo_surface_show_glyphs (cairo-surface.c:1796)
> ==25095==    by 0x402FDAE: _cairo_gstate_show_glyphs (cairo-gstate.c:1595)
> ==25095==    by 0x4029AF0: cairo_show_text (cairo.c:2825)
> ==25095==    by 0x804A203: draw (glyph-cache-pressure.c:80)
> ==25095==    by 0x804B714: cairo_test_targets (cairo-test.c:293)
> ==25095==    by 0x426631A: start_thread (in /lib/tls/i686/cmov/libpthread-2.5.so)
> ==25095==    by 0x436F3ED: clone (in /lib/tls/i686/cmov/libc-2.5.so)

Woot.  Is this yet another valgrind skin you've wrote?


> --
> Chris Wilson

-- 
behdad
http://behdad.org/

"Those who would give up Essential Liberty to purchase a little
 Temporary Safety, deserve neither Liberty nor Safety."
        -- Benjamin Franklin, 1759