RE: cygwin runtime woes

["Pierre Muller" <muller@ics.u-strasbg.fr>]

Pierre@d620-muller ~/gdbcvs/build-normal/gdb
$ export GDBTK_DEBUG=1

Pierre@d620-muller ~/gdbcvs/build-normal/gdb

Pierre@d620-muller ~/gdbcvs/build-normal/gdb
$ gdbcvs ./insight.exe
GNU gdb (GDB) 6.8.50.20090306-cvs
Copyright (C) 2009 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later
<http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "i686-pc-cygwin".
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>...
Setting up the environment for debugging gdb.
During symbol reading, struct/union type gets multiply defined: struct type.
Breakpoint 1 at 0x40c8d0: file ../../purecvs/gdb/utils.c, line 993.
Breakpoint 2 at 0x41cced: file ../../purecvs/gdb/cli/cli-cmds.c, line 199.
(top-gdb) set new-c
(top-gdb) r ./gdb.exe
Starting program: /usr/local/src/gdbcvs/build-normal/gdb/insight.exe
./gdb.exe
[New Thread 168.0x35c]
[New Thread 168.0x15d4]
[New Thread 168.0x111c]
[New Thread 168.0x1224]
[New Thread 168.0x5a0]
[New Thread 168.0xaa8]
[New Thread 168.0x1300]
[New Thread 168.0x1544]
[New Thread 168.0x1114]

Program received signal SIGSEGV, Segmentation fault.
0x7473d529 in TF_GetLangIcon () from /cygdrive/d/WINDOWS/system32/msctf.dll
(top-gdb) bt
#0  0x7473d529 in TF_GetLangIcon ()
   from /cygdrive/d/WINDOWS/system32/msctf.dll
#1  0x7472513a in TF_CreateCategoryMgr ()
   from /cygdrive/d/WINDOWS/system32/msctf.dll
#2  0x74725382 in TF_CreateCategoryMgr ()
   from /cygdrive/d/WINDOWS/system32/msctf.dll
#3  0x74721dd3 in ?? () from /cygdrive/d/WINDOWS/system32/msctf.dll
#4  0x74720000 in ?? ()
#5  0x00000000 in ?? ()
(top-gdb) inf thr
* 1 Thread 168.0x35c  0x7473d529 in TF_GetLangIcon ()
   from /cygdrive/d/WINDOWS/system32/msctf.dll
(top-gdb) q
The program is running.  Quit anyway (and kill it)? (y or n) y


$ cat insight.log
I global pref_set_defaults {}
I global set_bg_colors {}
I global r_setcolors {. -background #cccccc}
I global r_setcolors {. -highlightbackground #cccccc}
I global r_setcolors {. -textbackground #ffffff}
I global r_setcolors {. -troughcolor #b8b8b8}
I global r_setcolors {. -activebackground #b8b8b8}
I global gdbtk_locate_main {}
I global gdbtk_clear_file {}
I global gdbtk_locate_main {}
I ManagedWin find SrcWin
I global gdbtk_locate_main {}
I SrcWin point_to_main {could not find main}
I global gdbtk_tcl_warning {struct/union type gets multiply defined: struct
type
}
I ManagedWin startup {Got active list {ManagedWin::open ProcessWin}}
I ManagedWin _open {ProcessWin }
I ManagedWin _create {win=processwin args=}
I GDBWin constructor {}
I EmbeddedWin constructor {}
I GDBEventHandler GDBEventHandler::dispatch {posting event "busy" to
"::.process
win0.processwin"}
I GDBEventHandler GDBEventHandler::dispatch {posting event "idle" to
"::.process
win0.processwin"}
I EmbeddedWin destructor {}
I global pref_save {pref_save E:/cygwin/home/Pierre/gdbtk.ini}

Does that help you?

Pierre

> -----Message d'origine-----
> De : insight-owner@sourceware.org [mailto:insight-owner@sourceware.org]
> De la part de Keith Seitz
> Envoyé : Wednesday, May 06, 2009 6:11 PM
> À : Pierre Muller
> Cc : insight@sourceware.org
> Objet : Re: cygwin runtime woes
> 
> Pierre Muller wrote:
> > I debugged the problem a bit:
> >
> > gdb_loc is call a few times
> > for 'main' 'MAIN___' 'MAIN__' 'cyg_user_start' 'cyg_start'
> > and then finally with '*' which causes the abort.
> 
> That sounds normal. When insight starts up and gdb loads an executable,
> it will attempt to locate the entry point for the executable so that
> the
> source window can display it. These function names are actually from
> the
> preference gdb/main_names, defined in prefs.tcl.
> 
> The '*' argument is odd. I don't know where that is coming from.
> Passing
> '*' to gdb_loc will cause gdb_loc to error, though.
> 
> Out of curiosity, have you tried inhibiting preferences at startup?
> (insight -nx) Does it make any difference?
> 
> > the first calls all fail with
> > "No symbol table loaded. Use "file" command"
> > even though I used ./gdb.exe as arg for run command.
> 
> That sounds odd, too. Can you start vanilla gdb this way? I wonder what
> insight might be complaining about. Have you tried running insight in
> debug mode? To enable this, set GDBTK_DEBUG=1 in your environment
> before
> starting insight. It will write out the file "insight.log" (which you
> can change by setting GDBTK_DEBUGFILE).
> 
> Keith