:) I've read some where that memcpy not hmemcpy works in Nt/XP. Never tried it yet.

The problems with pret or stepover.
 

The problems have more to do with the logistics of
the XP threading model. I doubt this will be fixed easily
or least of all quickly.

If you break within (for example) MessageBoxA, have a look at your stack ( d esp ) and set a break point at your return address.