https://bugs.documentfoundation.org/show_bug.cgi?id=95241
--- Comment #7 from Luke Kendall <l...@zeta.org.au> ---
I am right now experiencing a similar problem. I had just clicked into the body
of a comment, and went to choose the Format al comments option, and LO locked
up almost completely. I noticed that the main Format menu had dropped down -
I'm not sure why.
(Earlier in the day, before I split the many-comments file in two, I had yet
another bizarre behaviour: when clicking on the small "down arrow" in the
little scrollbar within a comment, instead of the comment scrolling, the text
cursor insertion point in the main document moved down one line each time I
clicked on the down-scroll button in the comment scrollbar!)
Anyway, LO is once again using over 99% CPU:
top - 22:56:54 up 92 days, 2:32, 15 users, load average: 1.74, 1.69, 1.51
Tasks: 267 total, 3 running, 263 sleeping, 0 stopped, 1 zombie
%Cpu(s): 37.6 us, 0.4 sy, 0.0 ni, 62.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem: 16101596 total, 14675220 used, 1426376 free, 2485200 buffers
KiB Swap: 9820156 total, 1087620 used, 8732536 free. 4834772 cached Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
612 luke 20 0 2361204 829372 79872 R 100.0 5.2 40:06.64 soffice.bin
29840 luke 20 0 3746448 2.033g 47356 S 43.5 13.2 1167:08 firefox
...
In fact, at first I could do nothing whatsoever in my X1 session - even Alt-Tab
wouldn't let me change focus. I did Ctrl-Alt-F1 to get a text console and ran
gdb to get a stack traceback, and when I went back to the X session via
Ctrl-Alt-F7 I could at least change focus again, and mouse events could get
through to everything in a normal way.
One unusual thing that may be related: it's pretty likely that a "Rest break"
window popped up while I was in the middle of using LO. Could it be that LO is
vulnerable to having the focus stolen from it unexpectedly? I have the
Workrave software installed to remind me to take breaks. It's quite possible
that a brief "30 second break" window had popped up to take focus during each
of the earlier incidents. I know that the longer "Rest break" panel had only
been on screen for 15 seconds when the latest problem occurred. I'm so
accustomed to them that I may not have even noticed it, if that was correlated
to the LO problems.
The stack backtrace from gdb is different and much shorter this time:
0x00007f79c08cf5eb in XCheckIfEvent ()
from /usr/lib/x86_64-linux-gnu/libX11.so.6
(gdb) where
#0 0x00007f79c08cf5eb in XCheckIfEvent ()
from /usr/lib/x86_64-linux-gnu/libX11.so.6
#1 0x00007f79b0834e1f in X11SalInstance::AnyInput(VclInputFlags) ()
from /opt/libreoffice5.0/program/libvclplug_genlo.so
#2 0x00007f798a631192 in SwLayIdle::_DoIdleJob(SwContentFrm const*,
SwLayIdle::IdleJobType) () from
/opt/libreoffice5.0/program/../program/libswlo.so
#3 0x00007f798a6313dd in SwLayIdle::DoIdleJob(SwLayIdle::IdleJobType, bool) ()
from /opt/libreoffice5.0/program/../program/libswlo.so
#4 0x00007f798a6343c0 in SwLayIdle::SwLayIdle(SwRootFrm*, SwViewShellImp*) ()
from /opt/libreoffice5.0/program/../program/libswlo.so
#5 0x00007f798a964a57 in SwViewShell::LayoutIdle() ()
from /opt/libreoffice5.0/program/../program/libswlo.so
#6 0x00007f798a48f176 in sw::DocumentTimerManager::DoIdleJobs(Idle*) ()
from /opt/libreoffice5.0/program/../program/libswlo.so
#7 0x00007f79c4df50bf in ImplSchedulerData::Invoke() ()
from /opt/libreoffice5.0/program/libmergedlo.so
#8 0x00007f79c4df522f in Scheduler::ProcessTaskScheduling(bool) ()
from /opt/libreoffice5.0/program/libmergedlo.so
#9 0x00007f79c4e02400 in Application::Yield() ()
from /opt/libreoffice5.0/program/libmergedlo.so
#10 0x00007f79c4e024b5 in Application::Execute() ()
from /opt/libreoffice5.0/program/libmergedlo.so
#11 0x00007f79c3f14a33 in desktop::Desktop::Main() ()
from /opt/libreoffice5.0/program/libmergedlo.so
#12 0x00007f79c4e07729 in ImplSVMain() ()
from /opt/libreoffice5.0/program/libmergedlo.so
#13 0x00007f79c4e07772 in SVMain() ()
from /opt/libreoffice5.0/program/libmergedlo.so
#14 0x00007f79c3f338a2 in soffice_main ()
from /opt/libreoffice5.0/program/libmergedlo.so
#15 0x000000000040075b in main ()
Now, I said LO was almost completely locked up. Although I could not get the
screen to refresh, or any UI element to respond, at one point when I clicked
into the window, one small paragraph of text (about 3 lines long) did get
redrawn in the LO canvas, along with the cursor (which was in that paragraph).
No other UI element would operate, Ctrl-S would not save the document, and nor
could I close it by clicking on the Close icon.
I "rolled up" both windows ("MS" and "MS-ed" - part 1) to add this comment.
Just now I "unrolled" the MS window and tried another Ctrl-S, and this time I
saw the "Saving" message and "save progress bar draw. Also, LO redrew just the
Comment identifiers. I'll take a new screenshot (LO-5.0-lockup-PM.png).
Unfortunately, from looking in the directory, I can see that LO has not
completed saving the file - the newest version is from over an hour ago.
Because I *really* don't want to lose the creative work I've done, which I may
not be able to reproduce, and because LO seems to respond somewhat, I'll leave
it running at 100% CPU in the hope that sometime overnight it completes the
Save.
--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Libreoffice-bugs mailing list
Libreoffice-bugs@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs
