Bug 1887 - Samba panic handler breaks OpenVMS traceback facility.
Samba panic handler breaks OpenVMS traceback facility.
Status: RESOLVED WONTFIX
Product: Samba 3.0
Classification: Unclassified
Component: Config Files
3.0.7
Alpha OpenVMS
: P3 enhancement
: none
Assigned To: Samba Bugzilla Account
Samba QA Contact
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2004-10-07 19:00 UTC by John Malmberg
Modified: 2010-04-26 03:44 UTC (History)
0 users

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description John Malmberg 2004-10-07 19:00:09 UTC
In the module fault.c, a panic handler is established to cause a core dump on
fatal exceptions.

This is the wrong thing to do on OpenVMS.

On OpenVMS, unless overridden at link time, a traceback facility is enabled that
gives the debugging information of where the fault occurred, including the
source module name and line numbers of all routines on the stack.  It also
displays diagnostic information about the fault.

If the SAMBA fault handler intercepts the signals, instead of getting this
diagnostic information, all that that gets displayed is that the panic handler
called exit from some unknown module.

My sugestion is allow the configure process to add a macro to indicate that the
host operating system traceback/panic handler should be used if it is defined as
part of the configure process.

Example:

void fault_setup(void (*fn)(void *))
{
#ifndef USE_HOST_PANIC_HANDLER

        cont_fn = fn;

#ifdef SIGSEGV
        CatchSignal(SIGSEGV,SIGNAL_CAST sig_fault);
#endif
#ifdef SIGBUS
        CatchSignal(SIGBUS,SIGNAL_CAST sig_fault);
#endif
#ifdef SIGABRT
        CatchSignal(SIGABRT,SIGNAL_CAST sig_fault);
#endif

}
#endif
Comment 1 Stefan Metzmacher 2010-04-26 03:44:48 UTC
If it's still broken in 3.5, please reopen.
3.0 isn't supported anymore.