Just because Wine and Windows do it that way doesn't make it any more correct or portable.
RtlQueueWorkItem expects a function of this type: typedef VOID (NTAPI *WORKERCALLBACKFUNC)(IN PVOID Context);
The supplied function is of this type: typedef ULONG (NTAPI *PTHREAD_START_ROUTINE)(PVOID Parameter);
This doesn't crash on x86 and "works" because the return value is simply ignored, but it's definitely not portable and MAY crash on another architecture where it does make a difference. I don't know if it would matter on ARM/PPC/... but the old code at least implemented it correctly/portable. It's not really an issue but IMO a bad hack. Typecasting a function pointer to a function with a different signature is hardly ever good practice.
- Thomas
Dmitry Chapyshev wrote:
On Fri, 17 Jul 2009 23:03:18 +0400, Thomas Bluemel thomas@reactsoft.com wrote:
This explains why we are using a trampoline function and not just typecast there... Is there a reason you changed this?
- Thomas
dchapyshev@svn.reactos.org wrote:
- /* NOTE: Don't use Function directly since the callback signature
differs. This might cause problems on certainplatforms... */
- Status = RtlQueueWorkItem(InternalWorkItemTrampoline,
WorkItemContext,
Ros-dev mailing list Ros-dev@reactos.org http://www.reactos.org/mailman/listinfo/ros-dev
Wine and Windows do so. Why we should do in another way? Other reasons are necessary?