Re: Another crash in scrolling.

[David Kastrup]

address@hidden (Kim F. Storm) writes:

> David Kastrup <address@hidden> writes:
>
>> Here is at least the dump, "it" does not seem set to any useful
>> value.  Probably I should set the breakpoint on abort in order to get
>> a useful backtrace.
>
> I cannot reproduce the crash here.

Well, basically I run preview-latex on some larger buffer, then page
through the buffer forwards and backwards.

>> Breakpoint 1, abort () at /home/tmp/emacs/src/emacs.c:456
>> 456       kill (getpid (), SIGABRT);
>> (gdb) bt
>> #0  abort () at /home/tmp/emacs/src/emacs.c:456
>> #1  0x080997b0 in move_it_vertically_backward (it=0xe6c, dy=0)
>>     at /home/tmp/emacs/src/xdisp.c:6370
>
> Could you try to remove the assert in that line, and see if things
> work well without it.

There you are, here is the next one:

456       kill (getpid (), SIGABRT);
(gdb) bt
#0  abort () at /home/tmp/emacs/src/emacs.c:456
#1  0x080997b0 in move_it_vertically_backward (it=0xe6c, dy=0)
    at /home/tmp/emacs/src/xdisp.c:6321
#2  0x08099364 in move_it_by_lines (it=0xbfecf6e0, dvpos=-2, need_y_p=0)
    at /home/tmp/emacs/src/xdisp.c:6496
#3  0x080be80f in window_scroll (window=139571676, n=-1074989344, whole=1, 
    noerror=0) at /home/tmp/emacs/src/window.c:4809
#4  0x080be931 in scroll_command (n=137490449, direction=-1)
    at /home/tmp/emacs/src/window.c:4999
#5  0x080be96f in Fscroll_down (arg=3353) at /home/tmp/emacs/src/window.c:5029
#6  0x08171bf9 in Ffuncall (nargs=2, args=0xbfecfa64)
    at /home/tmp/emacs/src/eval.c:2783
#7  0x0816e90c in Fcall_interactively (function=137687993, 
    record_flag=137490449, keys=137547324) at /home/tmp/emacs/src/callint.c:877
#8  0x081194a3 in Fcommand_execute (cmd=137687993, record_flag=137490449, 
    keys=137490449, special=137490449) at /home/tmp/emacs/src/keyboard.c:9697
#9  0x08120129 in command_loop_1 () at /home/tmp/emacs/src/keyboard.c:1792
#10 0x0816ff26 in internal_condition_case (bfun=0x811fdd0 <command_loop_1>, 
    handlers=137551441, hfun=0x8119db0 <cmd_error>)
    at /home/tmp/emacs/src/eval.c:1385
#11 0x0811485a in command_loop_2 () at /home/tmp/emacs/src/keyboard.c:1319
#12 0x0816fe35 in internal_catch (tag=3353, func=0x811483c <command_loop_2>, 
    arg=137490449) at /home/tmp/emacs/src/eval.c:1144
#13 0x08114669 in command_loop () at /home/tmp/emacs/src/keyboard.c:1298
#14 0x08114703 in recursive_edit_1 () at /home/tmp/emacs/src/keyboard.c:991
#15 0x081147fe in Frecursive_edit () at /home/tmp/emacs/src/keyboard.c:1052
#16 0x08113b85 in main (argc=3, argv=0xbfed0274)
    at /home/tmp/emacs/src/emacs.c:1766
(gdb) xbacktrace
"scroll-down"
"call-interactively"
(gdb) 

> I suspect the two asserts in move_it_vertically_backward are bogus
> (although they do appear to be perfectly sane).

In connection with preview-latex, it is dangerous to talk about
sanity.  It changes the _displayed_ text on the screen in filter
routines, but exclusively by changing _overlays_, not actually
touching the text, but replacing its visual representation with images
of differing sizes instead.  One crash had a "sit-for" in its
backtrace IIRC, and that is one of the opportunities where filter
routines run.

However, I think that most crashes occured at a time where this filter
routine business was not active.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum