Re: Debugging tips needed - application dies on WinXP SP2

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance

I've not seen this problem, but there are several possible ideas. Note that VS.NET has a
lot more checking of stack integrity, including detecting buffer overruns in the stack.
This is done by adding a few hundred bytes to each stack frame in debug mode, and filling
them with a pattern. A buffer overrun will clobber these bytes (instead of, say, your
return address), and you'll get a runtime error. I don't recall the 1282ness of it, but I
was startled the first time I saw it (we'd been running the code for years under VS6).
Turns out the client had decided to use 'bool', had called a function that wanted
(LPBOOL), and had cast the bool * to an LPBOOL. Of course, the called function wrote 4
bytes, which overwrote three filler bytes on the stack, and the runtime caught it! (I'd
warned them this would get them in trouble...)

I can't pin it down specifically to the list control operations, but are you using
callbacks, owner-draw, or something like that for your list view? The DrawItem looks
mildly suspicious in this regard. (but this is just sort-of-guesswork).
joe

On 22 Sep 2005 07:41:05 -0700, erik.finnman@xxxxxxxxx wrote:

>Well, it turns out that I was actually able to get some info from
>Windbg, since the callstack was avilable (which it wasn't in Visual
>Studio). Right now it looks like there may be some problems in the
>drawing code of a subclassed ListView. After exiting and returning the
>control to:
>
>comctl32.dll!_ListView_RDrawItem
>
>everything dies. So far I have no idea why, but at least there is
>something to go on...
>
>Does anyone know about new parameter checks introduced in this method
>in SP2?
>
>Erik
Joseph M. Newcomer [MVP]
email: newcomer@xxxxxxxxxxxx
Web: http://www.flounder.com
MVP Tips: http://www.flounder.com/mvp_tips.htm
.

Relevant Pages