Re: Test 10 Crashing and/or Client Freeze on Cilent Exit

["Robert Collins" <robert dot collins at itdomain dot com dot au>]

(from the peanut gallery)
It might be signal handling with cygwin....

Rob

----- Original Message -----
From: "Harold Hunt" <huntharo@msu.edu>
To: "Cygx (E-mail)" <cygwin-xfree@sources.redhat.com>
Sent: Wednesday, March 28, 2001 8:41 AM
Subject: Test 10 Crashing and/or Client Freeze on Cilent Exit


> I have been debugging the client exit crashes for the past, oh, 24
hours,
> and I have convinced myself that the crash is not being caused by my
code.
> I am convinced because of the following reasons:
>
> 1) Client exits that cause XWin.exe to crash do not result in an
> STATUS_ACCESS_VIOLATION, indicating that the crash is
> not a problem caused by an invalid memory pointer.
> 2) Client exits that cause XWin.exe to crash do not result in a call
> to AbortDDX (), ddxGiveUp (), or FatalError ().  This indicates
> that some code, somewhere, is calling exit ().
> 3) The last Windows message I see is either a key release or a
> mouse button release; I never see a WM_QUIT, WM_DESTROY,
> WM_CLOSE, or any other strange message.  I even modified
> the window processing loop to output the code for *every*
> message it received; the last messages I get are
> completely normal.
> 4) winWakeupHandler returns correctly from the last call that
> it processes.
> 5) When I fixed a typo in xc/programs/Xserver/os/WaitFor.c,
> client exits that crashed into an infinite loop seemed
> to get less likely.
>
> Now, the error may or may not actually be in my code; however, I need
> someone to convince me otherwise :)
>
> I need a talented debugger to tell me what function XWin.exe is in
when it
> exits.
>
> Any takers?
>
> Also, I need someone to run a leak checker against XWin.exe.
>
> Well, let me know if you are interested :)
>
> Harold
>
>