Re: [Flightgear-devel] Crash on linux with latest CVS code

2002-01-24 Thread David Megginson 

Curtis L. Olson writes:

   If you're using Emacs, try
   
 M-x cvs-examine
   
   and then enter the root directory for the base package when propted.
   Use 'x' to clear any warning messages, etc., then see what's left.
  
  Interesting tool.  In Debian anyway, make sure you have the cvs-pcl
  package installed.

You need that only for Gnu Emacs; XEmacs (my editor of choice) has
everything bundled, so you need only the regular Debian cvs package.

I've been happy with how much easier CVS management is with this tool,
and with how well integrated it is into the rest of XEmacs.  There's
little interactive help, though, so it's worth taking ten minutes to
read the info pages carefully.  It's especially nice for constructing
.cvsignore files.


All the best,


David

-- 
David Megginson
[EMAIL PROTECTED]


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] Crash on linux with latest CVS code

2002-01-24 Thread James Gallagher 

On Wed, 2002-01-23 at 19:40, Jim Wilson wrote:
 James,
 
 I rebuilt again from CVS tonight (2nd time this week) and no problems. When I
 do it I do a full rebuild including all the config and makefiles (even if I
 think it might not be necessary).
 
 This is a gcc2.95.2/glibc2.1 system for what thats worth.  Ran default just as
 you did and it loads up fine.
 
 This really looks like corrupted XML.  Is it possible that a merge conflict
 occured that you missed?  You can grep for that.  Or you could check any xml
 files flagged M or rm -rf the xml files and reupdate.

I checked both the base and source for any conflicts or modified files
and found none. I got new modules from cvs anyway just to be sure and
rebuilt. Same problem.

I'm building with gcc 2.95.3 on RH 7.1 (kernel 2.4.9-12). I have plib
1.4.2.

I think I'll try linking with efence next.

James

 
 Best,
 
 Jim
 
 James Gallagher [EMAIL PROTECTED] said:
 
  On Wed, 2002-01-23 at 07:52, Curtis L. Olson wrote:
   James Gallagher writes:
I get the following seg fault with the latest CVS version of FG (using
the latest CVS SimGear):

[jimg@dcz FlightGear]$ src/Main/fgfs 
FlightGear:  Version 0.7.9

Scanning for root: command line
Scanning for root: /home/jimg/.fgfsrc
fg_root = /usr/local/src/X11/fgfsbase
Segmentation fault (core dumped)
   
   Hmmm, this looks like you are dying in the property manager code some
   how.  Do you have the latest cvs version of the base package?  Perhaps
   a config file there got corrupted?
  
  I double checked this. I have the latest fg base (from CVS).
   
   Another things worth trying is to rebuild simgear from scratch, make clean;
   make, then install it, then make clean; make in the FlightGear tree.
  
  Yup, I tried this too, with metakit 2.4.2 (both with and without Erik's
  suggestion of removing the #define bool int (which on gcc won't be
  defined, but just to be sure...). 
  
  It still crashes. 
  
  Stepping through the code in gdb/ddd I saw the following:
  
  In
  inline bool fgSetDouble (const string name, double val)
  {
return globals-get_props()-setDoubleValue(name, val);
  }
  
  The call to globals-get_props()-setDoubleValue(...) has stuff that
  makes sense. But inside setDoubleValue the formal param relative_path is
  null and that's where the seg fault comes from.
  
  bool
  SGPropertyNode::setDoubleValue (const string relative_path, double
  value)
  {
return getNode(relative_path, true)-setDoubleValue(value);
  }
  
  I'm sort of swamped right now, so I haven't taken this any further.
  Sorry. Maybe this will be enough of a clue for someone else on the list.
  
  Oh, I was building from CVS just fine until at least late last week, so
  whatever is causing this happened relatively recently.
  
  James
  
   
   Regards,
   
   Curt.
   
   


I poked around with gdb/ddd. Here's the stack from the core file:

#0  0x82e5204 in SGPropertyNode::getNode (this=0x0,
relative_path=@0xb2e0, create=true) at props.cxx:1290
#1  0x82e5c24 in SGPropertyNode::setDoubleValue (this=0x0,
relative_path=@0xb2e0, value=-110.6642444)
at props.cxx:1472
#2  0x8089d64 in fgSetDefaults () at fg_props.hxx:328
#3  0x80689d5 in fgInitConfig (argc=1, argv=0xb724) at
fg_init.cxx:219
#4  0x8059089 in mainLoop (argc=1, argv=0xb724) at main.cxx:1487
#5  0x805dd72 in main (argc=1, argv=0xb724) at main.cxx:1816
#6  0x4054b306 in __libc_start_main (main=0x805dd54 main, argc=1,
ubp_av=0xb724, init=0x804d4a4 _init, 
fini=0x83a2a78 _fini, rtld_fini=0x4000d2dc _dl_fini,
stack_end=0xb71c)
at ../sysdeps/generic/libc-start.c:129

I tried various things, like not using my ~/.fgfsrc file, et c. but I
still get the seg fault. 

Any one else see this?

Thanks,
James

-- 
__
James Gallagher  The Distributed Oceanographic Data System
[EMAIL PROTECTED]  
http://unidata.ucar.edu/packages/dods
Voice: 775.337.8612  Fax: 775.337.2105

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
   
   -- 
   Curtis Olson   Intelligent Vehicles Lab FlightGear Project
   Twin Cities[EMAIL PROTECTED]  [EMAIL PROTECTED]
   Minnesota  http://www.menet.umn.edu/~curt   http://www.flightgear.org
   
   ___
   Flightgear-devel mailing list
   [EMAIL PROTECTED]
   http://mail.flightgear.org/mailman/listinfo/flightgear-devel
  --

Re: [Flightgear-devel] Crash on linux with latest CVS code

2002-01-23 Thread Jim Wilson 

James,

I rebuilt again from CVS tonight (2nd time this week) and no problems. When I
do it I do a full rebuild including all the config and makefiles (even if I
think it might not be necessary).

This is a gcc2.95.2/glibc2.1 system for what thats worth.  Ran default just as
you did and it loads up fine.

This really looks like corrupted XML.  Is it possible that a merge conflict
occured that you missed?  You can grep for that.  Or you could check any xml
files flagged M or rm -rf the xml files and reupdate.

Best,

Jim

James Gallagher [EMAIL PROTECTED] said:

 On Wed, 2002-01-23 at 07:52, Curtis L. Olson wrote:
  James Gallagher writes:
   I get the following seg fault with the latest CVS version of FG (using
   the latest CVS SimGear):
   
   [jimg@dcz FlightGear]$ src/Main/fgfs 
   FlightGear:  Version 0.7.9
   
   Scanning for root: command line
   Scanning for root: /home/jimg/.fgfsrc
   fg_root = /usr/local/src/X11/fgfsbase
   Segmentation fault (core dumped)
  
  Hmmm, this looks like you are dying in the property manager code some
  how.  Do you have the latest cvs version of the base package?  Perhaps
  a config file there got corrupted?
 
 I double checked this. I have the latest fg base (from CVS).
  
  Another things worth trying is to rebuild simgear from scratch, make clean;
  make, then install it, then make clean; make in the FlightGear tree.
 
 Yup, I tried this too, with metakit 2.4.2 (both with and without Erik's
 suggestion of removing the #define bool int (which on gcc won't be
 defined, but just to be sure...). 
 
 It still crashes. 
 
 Stepping through the code in gdb/ddd I saw the following:
 
 In
 inline bool fgSetDouble (const string name, double val)
 {
   return globals-get_props()-setDoubleValue(name, val);
 }
 
 The call to globals-get_props()-setDoubleValue(...) has stuff that
 makes sense. But inside setDoubleValue the formal param relative_path is
 null and that's where the seg fault comes from.
 
 bool
 SGPropertyNode::setDoubleValue (const string relative_path, double
 value)
 {
   return getNode(relative_path, true)-setDoubleValue(value);
 }
 
 I'm sort of swamped right now, so I haven't taken this any further.
 Sorry. Maybe this will be enough of a clue for someone else on the list.
 
 Oh, I was building from CVS just fine until at least late last week, so
 whatever is causing this happened relatively recently.
 
 James
 
  
  Regards,
  
  Curt.
  
  
   
   
   I poked around with gdb/ddd. Here's the stack from the core file:
   
   #0  0x82e5204 in SGPropertyNode::getNode (this=0x0,
   relative_path=@0xb2e0, create=true) at props.cxx:1290
   #1  0x82e5c24 in SGPropertyNode::setDoubleValue (this=0x0,
   relative_path=@0xb2e0, value=-110.6642444)
   at props.cxx:1472
   #2  0x8089d64 in fgSetDefaults () at fg_props.hxx:328
   #3  0x80689d5 in fgInitConfig (argc=1, argv=0xb724) at
   fg_init.cxx:219
   #4  0x8059089 in mainLoop (argc=1, argv=0xb724) at main.cxx:1487
   #5  0x805dd72 in main (argc=1, argv=0xb724) at main.cxx:1816
   #6  0x4054b306 in __libc_start_main (main=0x805dd54 main, argc=1,
   ubp_av=0xb724, init=0x804d4a4 _init, 
   fini=0x83a2a78 _fini, rtld_fini=0x4000d2dc _dl_fini,
   stack_end=0xb71c)
   at ../sysdeps/generic/libc-start.c:129
   
   I tried various things, like not using my ~/.fgfsrc file, et c. but I
   still get the seg fault. 
   
   Any one else see this?
   
   Thanks,
   James
   
   -- 
   __
   James GallagherThe Distributed Oceanographic Data System
   [EMAIL PROTECTED]  
   http://unidata.ucar.edu/packages/dods
   Voice: 775.337.8612Fax: 775.337.2105
   
   ___
   Flightgear-devel mailing list
   [EMAIL PROTECTED]
   http://mail.flightgear.org/mailman/listinfo/flightgear-devel
  
  -- 
  Curtis Olson   Intelligent Vehicles Lab FlightGear Project
  Twin Cities[EMAIL PROTECTED]  [EMAIL PROTECTED]
  Minnesota  http://www.menet.umn.edu/~curt   http://www.flightgear.org
  
  ___
  Flightgear-devel mailing list
  [EMAIL PROTECTED]
  http://mail.flightgear.org/mailman/listinfo/flightgear-devel
 -- 
 __
 James GallagherThe Distributed Oceanographic Data System
 [EMAIL PROTECTED]  
 http://unidata.ucar.edu/packages/dods
 Voice: 775.337.8612Fax: 775.337.2105
 
 ___
 Flightgear-devel mailing list
 [EMAIL PROTECTED]
 http://mail.flightgear.org/mailman/listinfo/flightgear-devel