[Tcl-bugs] [ tcl-Bugs-3002655 ] Semi-repeatable SIGSEGV doing info level

[SourceForge.net <no...@so...>]

Bugs item #3002655, was opened at 2010-05-17 11:59
Message generated for change (Comment added) made by jaspertheperson
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110894&aid=3002655&group_id=10894

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: 17. Commands I-L
Group: development: 8.6b1.1
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Jasper  (jaspertheperson)
Assigned to: miguel sofer (msofer)
Summary: Semi-repeatable SIGSEGV doing info level

Initial Comment:
Started getting occasional segfaults in my app. Narrowed it down to a call to "info level" -- arg around 14. Checked out HEAD and built with debugging symbols enabled (32bit, multithreaded) and still get bug. gdb not very informative:
% jaspert@spacehopper:~/Build/tk/unix$ ./wish
% info patchlevel
8.6b1.2
% array get tcl_platform
osVersion 2.6.32-22-generic pointerSize 4 byteOrder littleEndian threaded 1 machine i686 platform unix pathSeparator : os Linux user jaspert wordSize 4
...
[source app and attach gdb, then cause crash]
...
(gdb) c
Continuing.
[New Thread 0x178bb70 (LWP 24951)]

Program received signal SIGSEGV, Segmentation fault.
0x0043ffb5 in NRCommand (data=0x95a1fc8, interp=0x0, result=157867776)
    at /home/jaspert/Build/tcl/generic/tclBasic.c:4332
4332	    ((Interp *)interp)->numLevels--;
(gdb) bt
#0  0x0043ffb5 in NRCommand (data=0x95a1fc8, interp=0x0, result=157867776)
    at /home/jaspert/Build/tcl/generic/tclBasic.c:4332
#1  0x0a4738ec in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb) c
Continuing.
[Thread 0x178bb70 (LWP 24951) exited]
[Thread 0xb77cdb70 (LWP 24908) exited]

Program terminated with signal SIGSEGV, Segmentation fault.
The program no longer exists.

Please feel free to request any further tests


----------------------------------------------------------------------

>Comment By: Jasper  (jaspertheperson)
Date: 2010-05-17 13:44

Message:
The app loads some non-Tcl c code, e.g., Tktable, but does not use it; I
will try not loading it to see if that makes any difference.


----------------------------------------------------------------------

Comment By: miguel sofer (msofer)
Date: 2010-05-17 13:37

Message:
I read that gdb output a bit diffferently: "corrupt stack?" means "no valid
info here, look elsewhere".

In order to help please provide more info:

  1. is this a pure tcl script, or is there any non-tcl C code involved?
Extensions, your own code, ... 

  2. if you use gdb, you probably can run under valgrind too. Please do

  (a) make distclean; GCC=-DPURIFY ./configure --disabled-shared
--enable-symbols
  (b) valgrind ./tclsh, source app, crash
  (c) let's see the full valgrind output

----------------------------------------------------------------------

Comment By: Donal K. Fellows (dkf)
Date: 2010-05-17 13:19

Message:
I've no idea why interp is NULL there. It shouldn't be.

----------------------------------------------------------------------

Comment By: Jasper  (jaspertheperson)
Date: 2010-05-17 12:08

Message:
This may be of some help: I reconfigured my app to do everything in one
thread, and the crash appears no longer to happen.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110894&aid=3002655&group_id=10894