On Sun, 7 Aug 2005 16:10:36 +0200 Dominik Vogt <[EMAIL PROTECTED]> wrote:
> On Sun, Aug 07, 2005 at 03:35:32PM +0200, T. Baumann wrote:
> > On Sun, 7 Aug 2005 14:09:04 +0200 Dominik Vogt <[EMAIL PROTECTED]>
> > wrote:
> >
> > > On Sun, Aug 07, 2005 at 12:38:00PM +0200, T. Baumann wrote:
> > > > attached.
> > >
> > > Thanks. (Please put replies below the quted text on this mailing
> > > list).
> > >
> > > It does not get me much further though:
> > >
> > > > 4126 if (last_style_in_list &&
> > > > 3: ps->id = {name = 0x811f3e8 "*", window_id = 0, flags = {has_name = 1,
> > > > has_window_id = 0}}
> > > > 1: ps = (window_style *) 0x81213a0
> > > > (gdb)
> > > >
> > > > Program received signal SIGSEGV, Segmentation fault.
> > > > 0x402f8348 in strcmp () from /lib/tls/libc.so.6
> > > > (gdb) info local
> > > > No symbol table info available.
> > > > (gdb) up
> > > > #1 0x0805a695 in style_ids_are_equals (a={name = 0x1 <Address 0x1 out
> > > > of bounds>, window_id = 0, flags = {has_name = 1, has_window_id = 0}},
> > > > b=
> > > > {name = 0x810ce00 "*", window_id = 0, flags = {has_name = 1,
> > > > has_window_id = 0}}) at style.c:141
> > > > 141 !strcmp(SID_GET_NAME(a), SID_GET_NAME(b)))
> > >
> > > in style.c:4126, ps->id is still okay, but in the calles function
> > > a.name is 0x01 (= ps->id.name). Is that because fvwm was compiled
> > > optimized?
> > >
> > > Please recompile fvwm with
> > >
> > > $ make clean
> > > $ make CFLAGS="-g -O0"
> > >
> > > In gdb:
> > >
> > > * Put a breakpoint at style.c:4126
> > > * run fvwm in gdb and skip the bp once with "cont"
> > > * (gdb) display ps
> > > * (gdb) display ps->id
> > > * (gdb) display last_style_in_list
> > > * (gdb) display last_style_in_list->id
> > > * single-step with "step" until gdb enters the
> > > style_ids_are_equals function
> > > * display a
> > > * display a->id
> > > * display b
> > > * display b->id
> > > * single-step until fvwm crashes
> >
> > your last changes to the cvs code seem to be effective. I'm running:
> >
> > fvwm 2.5.14 (from cvs) compiled on Aug 7 2005 at 14:53:53 with support
> > for: ReadLine, Stroke, XPM, PNG, Shape, XShm, SM, Xinerama, XRender,
> > XFT, NLS
> >
> > Should I still do the debugging?
>
> Yes, please. Although I fixed a couple of bugs, neither could
> have caused the crash. Ignore the CVS code for now and debug with
> the code that does crash.
sorry, I do not have the old source to compile with -O0 (seems I was
too fast with the cvs-update...)
the debug-trace is attached
Thomas
fvwm.debug.bz2
Description: application/bzip
