Bug 837809

Created attachment 596379 [details]
gdb transcript of the issue

Description of problem:
Suddenly, gnome-shell hogs the CPU and renders the system unsuable.


Version-Release number of selected component (if applicable):
f16

How reproducible:
I don't know. But it happened to me at least twice this week.


Steps to Reproduce:
1.
2.
3.
  
Actual results:
system hangs, i.e. gnome-shell hogs the CPU, strace showing sched_yield every nanosecond or so.

Expected results:
...

Additional info:

I see at least a similar, if not the same issue on Fedora 16: strace shows sched_yield like crazy and the GNOME Shell hogs the CPU and cannot be used at all.

(gdb) t a a bESC[Kt full

Thread 7 (Thread 0x7f76e652f700 (LWP 1875)):
#0  0x00000039f20e85c3 in __GI___poll (fds=<optimized out>, nfds=<optimized out>, timeout=<optimized out>) at ../sysdeps/unix/sysv/linux/poll.c:87
        resultvar = <optimized out>
        oldtype = 0
        result = <optimized out>
#1  0x00000039f4045448 in g_main_context_poll (n_fds=1, fds=0x7f76dc001170, priority=<optimized out>, timeout=-1, context=0x23b4c30) at gmain.c:3402
        poll_func = 0x39f4053860 <g_poll>
#2  g_main_context_iterate (context=0x23b4c30, block=<optimized out>, dispatch=1, self=<optimized out>) at gmain.c:3084
        max_priority = 2147483647
        timeout = -1
        some_ready = <optimized out>
        nfds = 1
        allocated_nfds = <optimized out>
        fds = 0x7f76dc001170
        __PRETTY_FUNCTION__ = "g_main_context_iterate"
#3  0x00000039f4045c85 in g_main_loop_run (loop=0x7f76dc001150) at gmain.c:3297
        self = 0x23b1790
        __PRETTY_FUNCTION__ = "g_main_loop_run"
#4  0x00007f76e65349fb in dconf_context_thread (data=0x23b4c30) at dconfcontext.c:11
        context = 0x23b4c30
        loop = <optimized out>
        __PRETTY_FUNCTION__ = "dconf_context_thread"



Thread 1 (Thread 0x7f76ef55e9c0 (LWP 1859)):
#0  0x00000039f20d9bd7 in sched_yield () at ../sysdeps/unix/syscall-template.S:8
2
No locals.
#1  0x00007f76e7ebee26 in r600_fence_finish (pscreen=<optimized out>, fence=0x26
fe9c0, timeout=18446744073709551615) at r600_pipe.c:625
        rfence = 0x26fe9c0
        ctx = 0x22b1d60
        start_time = 0
        spins = 2655484928
#2  0x00007f76e7f9426d in st_finish (st=0x2348930) at state_tracker/st_cb_flush.c:106
        fence = 0x26fe9c0
#3  0x00007f76e7f942a0 in st_glFinish (ctx=<optimized out>) at state_tracker/st_cb_flush.c:141
        st = 0x2348930
#4  0x0000003a1365494c in _cogl_winsys_onscreen_swap_region (onscreen=<optimized out>, user_rectangles=<optimized out>, n_rectangles=1) at winsys/cogl-winsys-glx.c:1205
        framebuffer = <optimized out>
        context = 0x23513a0
        xlib_renderer = 0x21111c0
        glx_renderer = 0x21111c0
        xlib_onscreen = 0x2003700
        glx_onscreen = 0x2003700
        drawable = 16777229
        end_frame_vsync_counter = 0
        have_counter = 1
        can_wait = 1
        blit_sub_buffer_is_synchronized = 1
        framebuffer_height = <optimized out>
        rectangles = 0x7fff44cd1fb0
        i = <optimized out>
#5  0x0000003a140316a0 in clutter_stage_cogl_redraw (stage_window=0x23539b0) at cogl/clutter-stage-cogl.c:534
        clip = 0x2353a20
        copy_area = {33, 0, 1217, 1017}
        stage_cogl = 0x23539b0 [ClutterStageCogl]
        wrapper = 0x23c2050 [ClutterStage]
        backend = <optimized out>
        backend_cogl = <optimized out>
        may_use_clipped_redraw = <optimized out>
        use_clipped_redraw = <optimized out>
        stage_x11 = 0x23539b0 [ClutterStageCogl]
#6  0x0000003a140934ba in clutter_stage_do_redraw (stage=0x23c2050 [ClutterStage]) at ./clutter-stage.c:1023
        backend = 0x1fdf060 [ClutterBackendCogl]
        actor = 0x23c2050 [ClutterStage]
        priv = 0x23c2400
#7  _clutter_stage_do_update (stage=0x23c2050 [ClutterStage]) at ./clutter-stage.c:1079
        priv = 0x23c2400
#8  0x0000003a1407ac90 in clutter_clock_dispatch (source=<optimized out>, callback=<optimized out>, user_data=<optimized out>) at ./clutter-master-clock.c:384
        clock_source = <optimized out>
        master_clock = 0x210c5c0 [ClutterMasterClock]
        stage_manager = <optimized out>
        stages_updated = <optimized out>



I have more of that stacktrace in a separate file.



(gdb) call gjs_dumpstack ()
(gdb) quit



$ lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc RV730 PRO [Radeon HD 4650]
$


FWIW: Just before the hang, the screens (I have two) would flicker a couple of times, i.e. turn off for a second, then turn on for a couple of seconds, then turn off for a second until they turned on and finally froze.



I also commented on bug 706971, but as I can't set the version flag, I opened a new one.