[SOC] proposal -> valgrind

[Eric Pouech]

Another potential proposal for SOC would be upgrading valgrind to work 
with Wine.

I've tweaked it a bit (some patches still floating around), but:
- that's not complete (some items still don't work)
- what's also completly missing is the support of native modules 
(especially when generating backtraces)
- 16bit support (but we can safely leave it out for now)
- and testing (and fixing) most of the implementations we have

I've updated the Wiki page accordingly.
A+

[Robert Shearman]

Eric Pouech wrote:
> Another potential proposal for SOC would be upgrading valgrind to work 
> with Wine.
>
> I've tweaked it a bit (some patches still floating around), but:
> - that's not complete (some items still don't work)
> - what's also completly missing is the support of native modules
>
> (especially when generating backtraces)
> - 16bit support (but we can safely leave it out for now)
> - and testing (and fixing) most of the implementations we have
>
> I've updated the Wiki page accordingly.
> A+

It would be really good for valgrind know about our heap implementation 
so that it could report memory leaks. A solution I've done in the past 
is to change all of the references to HeapAlloc/HeapReAlloc/HeapFree to 
malloc/realloc/free so that it can report leaks, but that isn't very 
convenient.

I think that is something that could benefit Wine a lot and wouldn't 
require very much Wine knowledge at all (although, it could require some 
knowledge about valgrind).