DM> If str is a global pointer, then yes, you need to protect it.

 PC> str is defined in a class.  The mutex semaphore seems to have fixed my
 PC> traps.

Ah, so you're using C++.  Couldn't tell.  :-)

BTW, your code had something like:  request.MutexSem() and
release.MutexSem().  Was this a typo?  I'm not sure how that could work.

I have a few classes for this... a generic semaphore (it would only work
for mutex'ing...) and a 16-bit and 32-bit version of it, all in C++.  You
could then have:

MutexSem ms;
ms.Open("WHATEVER.SEM"); // opens \SEM32\WHATEVER.SEM
ms.Claim(timeout_val_in_ms); // defaults to an indefinate wait

// code...

ms.Release();


For example:

ULONG someFn()
{
  static MutexSem ms;
  if (!ms) // if this is the first call...
  {
    if (!ms.Open("someFn.SEM"))
    {
      return NO_SEM;
    }
  }

  ms.Claim();

  // do stuff

  ms.Release();

  return rc;
}

I'll be adding one more class so you can have:

ULONG someFn()
{
  static MutexSem ms;
  if (!ms) // if this is the first call...
  {
    if (!ms.Open("someFn.SEM"))
    {
      return NO_SEM;
    }
  }

  DoSem ds(ms);

  ...

  return rc; // ~DoSem will release the sem AFTER the return value is set!
}


Interested?  :-)


--- Maximus/2 3.01