Overlapped calls and crash dumps?

Blair,

What happens if an overlapped call goes bad?  I'm making overlapped calls to
something that might(???) itself be using other OS threads, and have an
intermittent crash.  The good news is that the crash occurs only if I start
and stop the offending code.  The bad news is that the crash dump says
little more than the library and address (hopefully the RVA will tell me
something) and that it did not occur on the main Dolphin thread.

Is that what would happen with a crash on an overlapped thread, or does it
suggest that the library has created a thread of its own that has faulted?

Have a good one,

Bill

--
Wilhelm K. Schwab, Ph.D.
[hidden email]

"Bill Schwab" <[hidden email]> wrote in message
news:boh498$1e79fs$[hidden email]...
> Blair,
>
> What happens if an overlapped call goes bad?  I'm making overlapped calls
to
> something that might(???) itself be using other OS threads, and have an
> intermittent crash.  The good news is that the crash occurs only if I
start
> and stop the offending code.  The bad news is that the crash dump says
> little more than the library and address (hopefully the RVA will tell me
> something) and that it did not occur on the main Dolphin thread.
>
> Is that what would happen with a crash on an overlapped thread, or does it
> suggest that the library has created a thread of its own that has faulted?

Yes, you would see that report if an overlapped call crashed. [The crash
dump code is not aware of the overlapped call threads, which seems a bit of
a silly oversight I have to say.] However you would also see that if an
external thread started by the library were to crash.

Regards

Blair

Blair,

> Yes, you would see that report if an overlapped call crashed. [The crash
> dump code is not aware of the overlapped call threads, which seems a bit
of
> a silly oversight I have to say.]

OTOH, it took until now for me to make a mistake that lead to it.  Still, it
would be nice to see dumps apply to overlapped call threads.


> However you would also see that if an
> external thread started by the library were to crash.

That seems quite reasonable.

BTW, the crash appears to have been due to my telling the library to free
something twice.

Have a good one,

Bill

--
Wilhelm K. Schwab, Ph.D.
[hidden email]