Re: ddeml: More calling convention fixes

[Dmitry Timoshkov]

Tristan Schmelcher <tristan_schmelcher(a)alumni.uwaterloo.ca> wrote:
> +    if (cbData != (DWORD)-1)
> +    {
> +        /* pData is not a pointer if cbData is -1, so we linearize the address
> +         * here rather than in the calling code. */
> +        pData = MapSL((SEGPTR)pData);
> +    }

-1 here should be replaced by CBR_BLOCK, and this piece of the patch
should be sent separately from .spec file fixes.

-- 
Dmitry.

ddeml: More calling convention fixes

[Tristan Schmelcher]

I think CBR_BLOCK is only for values of type HDDEDATA. The cbData
value here is a DWORD (specifying a length). (DWORD)-1 seems to be the
preferred way of checking this condition in the existing code, e.g.
see dlls/user32/dde_client.c lines 637, 648, etc.

I will re-send the patch in two parts.

On Thu, Dec 2, 2010 at 12:19 AM, Dmitry Timoshkov
<dmitry(a)codeweavers.com> wrote:
> Tristan Schmelcher <tristan_schmelcher(a)alumni.uwaterloo.ca> wrote:
> > +    if (cbData != (DWORD)-1)
> > +    {
> > +        /* pData is not a pointer if cbData is -1, so we linearize the address
> > +         * here rather than in the calling code. */
> > +        pData = MapSL((SEGPTR)pData);
> > +    }
>
> -1 here should be replaced by CBR_BLOCK, and this piece of the patch
> should be sent separately from .spec file fixes.
>
> --
> Dmitry.

ddeml: More calling convention fixes

[Dmitry Timoshkov]

Tristan Schmelcher <tristan_schmelcher(a)alumni.uwaterloo.ca> wrote:
> I think CBR_BLOCK is only for values of type HDDEDATA. The cbData
> value here is a DWORD (specifying a length). (DWORD)-1 seems to be the
> preferred way of checking this condition in the existing code, e.g.
> see dlls/user32/dde_client.c lines 637, 648, etc.

You are right.

-- 
Dmitry.