Re: [MACTCL] Getting segfault running with 8.5.2 and Mac OS X 10.5.3

[Trevor W. <pha...@gm...>]

The Tcl/Tk is stock.  I have not made any modifications to its code;
however, I would agree that the stack trace doesn't make sense to me either
(from what little I have looked at the source code).  I attempted to run
valgrind on it and saw several errors from within Tk libraries (I don't have
access to my machine at the moment, but I can provide those error messages
this evening if folks would like to see them).  Additionally, I am not
calling the 'tk inactive reset' command from within any of my code, so most
likely the stack trace is corrupted as you have suggested.

My project is open source, so I would be very comfortable providing the
source (about 2MB tar/gzipped) and  build instructions (the project doesn't
require any third-party libraries -- just stock GNU tools) if someone would
be willing to replicate this problem exactly.

Let me know what would be helpful.

Thanks,
Trevor


On Mon, Jun 9, 2008 at 11:59 AM, Daniel A. Steffen <
da...@us...> wrote:

> Hi Trevor,
>
> I cannot reproduce this crash, it looks like you may not be running your
> application on stock Tcl/Tk?
>
> your backtrace seems to indicate that the crashing
> Tk_ResetUserInactiveTime() is called from non-standard places, in stock tk
> this is only every called from the implementation of [tk inactive reset]
> (i.e. from Tk_TkObjCmd()). Are you using this command?
>
> the backtrace as a whole does not make a lot of sense to me, I don't
> understand e.g. why SendEventToEventTarget() should be called from
> Tk_ResetUserInactiveTime() or why DispatchEventToHandlers() would be calling
> Tk_ResetUserInactiveTime() in turn...
>
> possibly the stacktrace is bogus due to stack corruption, you can try
> putting a breakpoint in Tk_ResetUserInactiveTime() to get a backtrace from
> immediately before the crash.
>
> otherwise you will need to provide more detailed instructions to reproduce
> the problem and/or debug yourself any custom code and modifications you made
> to tcl/tk.
>
>
> On 07/06/2008, at 18:58, Trevor Williams wrote:
>
>  #0  0x0b0d2978 in Tk_ResetUserInactiveTime ()
>> #1  0x0b0d30ff in Tk_UpdatePointer ()
>> #2  0x0b0c59f2 in TkGenerateButtonEvent ()
>> #3  0x0b0b9d07 in TkMacOSXGetHostToplevel ()
>> #4  0x0b0d1b5f in Tk_ResetUserInactiveTime ()
>> #5  0x94661763 in DispatchEventToHandlers ()
>> #6  0x94660b9d in SendEventToEventTargetInternal ()
>> #7  0x9467d4ee in SendEventToEventTarget ()
>> #8  0x9468fb90 in ToolboxEventDispatcherHandler ()
>> #9  0x94661b1c in DispatchEventToHandlers ()
>> #10 0x94660b9d in SendEventToEventTargetInternal ()
>> #11 0x9467d4ee in SendEventToEventTarget ()
>> #12 0x0b0d1f4b in Tk_ResetUserInactiveTime ()
>> #13 0x0b0c5dd7 in Tk_MacOSXSetupTkNotifier ()
>> #14 0x0a07c871 in Tcl_DoOneEvent ()
>> #15 0x0b013cc8 in Tk_MainLoop ()
>> #16 0x00076757 in command_report (argc=3, last_arg=1, argv=0xbffff740)
>> at report.c:983
>> #17 0x00067eb5 in main (argc=3, argv=0xbffff740) at main.c:196
>>
>
>
> Cheers,
>
> Daniel
>
> --
> ** Daniel A. Steffen                   **
> ** <mailto:da...@us...>  **
>
>
>