Re: d3d9 : do not proceed reset or release from another thread than the one which created the device.

[Henri Verbeet]

2008/11/30 Jérôme Gardou <jerome.gardou(a)gmail.com>:
> -    if (ref == 0) {
> +    if (ref <= 0) {
> This doesn't work, of course. ref is an ULONG. The check for Reset
> should probably go into wined3d.

d3d9 : do not proceed reset or release from another thread than the one which created the device.

[Jérôme Gardou]

Henri Verbeet a écrit :
> 2008/11/30 Jérôme Gardou <jerome.gardou(a)gmail.com>:
> > -    if (ref == 0) {
> > +    if (ref <= 0) {
>
> This doesn't work, of course. ref is an ULONG.
> Why not just a cast in this case ?
>  The check for Reset
> should probably go into wined3d.

OK... but a test should check is this is the same behaviour in d3d8. Not 
sure...

d3d9 : do not proceed reset or release from another thread than the one which created the device.

[Jérôme Gardou]

Jérôme Gardou a écrit :
> Henri Verbeet a écrit :
> > 2008/11/30 Jérôme Gardou <jerome.gardou(a)gmail.com>:
> > > -    if (ref == 0) {
> > > +    if (ref <= 0) {
> >
> > This doesn't work, of course. ref is an ULONG.
>
> Why not just a cast in this case ?

According to d3d9_private.h , ref is a LONG.

d3d9 : do not proceed reset or release from another thread than the one which created the device.

[Henri Verbeet]

2008/12/6 Jérôme Gardou <jerome.gardou(a)gmail.com>:
> According to d3d9_private.h , ref is a LONG.

ref is not defined in d3d9_private.h.

d3d9 : do not proceed reset or release from another thread than the one which created the device.

[Jérôme Gardou]

Henri Verbeet a écrit :
> 2008/12/6 Jérôme Gardou <jerome.gardou(a)gmail.com>:
> > According to d3d9_private.h , ref is a LONG.
>
> ref is not defined in d3d9_private.h.

Taken from d3d9_private.h :

typedef struct IDirect3DDevice9Impl
{
    /* IUnknown fields */
    const IDirect3DDevice9ExVtbl   *lpVtbl;
    LONG                          ref;
...

That's why the compiler did not complained... But release actually 
returns a ULONG, and the IUnknown ref is an ULONG anyway...
I guess this should be changed for consistency.

d3d9 : do not proceed reset or release from another thread than the one which created the device.

[Jérôme Gardou]

Jérôme Gardou a écrit :
> Henri Verbeet a écrit :
> > 2008/12/6 Jérôme Gardou <jerome.gardou(a)gmail.com>:
> > > According to d3d9_private.h , ref is a LONG.
> >
> > ref is not defined in d3d9_private.h.
>
> Taken from d3d9_private.h :
>
> typedef struct IDirect3DDevice9Impl
> {
>     /* IUnknown fields */
>     const IDirect3DDevice9ExVtbl   *lpVtbl;
>     LONG                          ref;
> ...
>
> That's why the compiler did not complained... But release actually 
> returns a ULONG, and the IUnknown ref is an ULONG anyway...
> I guess this should be changed for consistency.

Oh dear... I upgraded my winXP VM to SP3 yesterday, and the tests I 
wrote now crash... seems like d3d9 does not care about threads anymore...