DBGRID32.OCX fix - RFC

[Alex Villací­s Lasso]

I would like to know if the attached patch (which prevents DBGRID32.OCX 
from crashing, as well as adding additional debugging information when a 
reference is not found) would step on anybody's toes (especially Robert 
Shearman, since he is sending several oleaut32 patches himself). As of 
2005-12-01 CVS, any VisualBasic application that tries to use 
DBGRID32.OCX will systematically crash unless the patch is applied. 
However, since Robert Shearman is sending oleaut32 fixes, I think he 
might implement some functionality that will, as a side effect, fix the 
DBGRID32.OCX bug.

Alex Villacís Lasso

[Robert Shearman]

Alex Villací­s Lasso wrote:
> I would like to know if the attached patch (which prevents 
> DBGRID32.OCX from crashing, as well as adding additional debugging 
> information when a reference is not found) would step on anybody's 
> toes (especially Robert Shearman, since he is sending several oleaut32 
> patches himself). As of 2005-12-01 CVS, any VisualBasic application 
> that tries to use DBGRID32.OCX will systematically crash unless the 
> patch is applied. However, since Robert Shearman is sending oleaut32 
> fixes, I think he might implement some functionality that will, as a 
> side effect, fix the DBGRID32.OCX bug.

No, I have no plans to change anything that should affect this.
> ITypeInfoImpl *This = (ITypeInfoImpl *)iface;
> HRESULT result = E_FAIL;
>
> - if (hRefType == -1 &&
> - (((ITypeInfoImpl*) This)->TypeAttr.typekind   == TKIND_DISPATCH) &&
> - (((ITypeInfoImpl*) This)->TypeAttr.wTypeFlags &  TYPEFLAG_FDUAL))
> - if (
> -
>    (
> -
>        hRefType == -1 &&
> -
>        (This->TypeAttr.typekind   == TKIND_DISPATCH) &&
> -
>        (This->TypeAttr.wTypeFlags &  TYPEFLAG_FDUAL)
> -
>    )
> -
>    ||
> -
>    (
> -
>        (This->TypeAttr.typekind   == TKIND_DISPATCH) &&
> -
>        (This->TypeAttr.wTypeFlags &  TYPEFLAG_FDISPATCHABLE)
> -
>    )
> -
>   )
> {

What hRefType is passed to the function? I don't think the above is 
quite correct. Can you try changing the if statement to something like this:

if (hRefType == -1 &&
    (This->TypeAttr.typekind   == TKIND_DISPATCH) &&
    (This->TypeAttr.wTypeFlags &  (TYPEFLAG_FDUAL|TYPEFLAG_FDISPATCHABLE)))

[Alex Villací­s Lasso]

Robert Shearman wrote:
> What hRefType is passed to the function? I don't think the above is 
> quite correct. Can you try changing the if statement to something like 
> this:
>
> if (hRefType == -1 &&
>     (This->TypeAttr.typekind   == TKIND_DISPATCH) &&
>     (This->TypeAttr.wTypeFlags &  
> (TYPEFLAG_FDUAL|TYPEFLAG_FDISPATCHABLE)))

That would not keep the app from crashing. For the case of DBGRID32.OCX, 
hRefType is systematically == 19, and it says so in the ERR message just 
before the crash. However, on a second look, maybe the problem is not 
that the function performs the wrong action, but that it returns the 
wrong value (E_FAIL, should be TYPE_E_ELEMENTNOTFOUND). I have to check 
this alternate theory on my home machine.

Alex Villacís Lasso

[Alex Villací­s Lasso]

Alex Villací­s Lasso wrote:
> Robert Shearman wrote:
> > What hRefType is passed to the function? I don't think the above is 
> > quite correct. Can you try changing the if statement to something 
> > like this:
> >
> > if (hRefType == -1 &&
> >     (This->TypeAttr.typekind   == TKIND_DISPATCH) &&
> >     (This->TypeAttr.wTypeFlags &  
> > (TYPEFLAG_FDUAL|TYPEFLAG_FDISPATCHABLE)))
>
> That would not keep the app from crashing. For the case of 
> DBGRID32.OCX, hRefType is systematically == 19, and it says so in the 
> ERR message just before the crash. However, on a second look, maybe 
> the problem is not that the function performs the wrong action, but 
> that it returns the wrong value (E_FAIL, should be 
> TYPE_E_ELEMENTNOTFOUND). I have to check this alternate theory on my 
> home machine.
>
> Alex Villacís Lasso

More information: if the flag dumping portion of the patch is applied, 
but the condition portion is not, and a dump of the available references 
is added, the following is the result when trying to load DBGRID32.OCX:

fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x0640
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x0190
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x03e8
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x0031
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x0708
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x0258
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x00c8
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x003d
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x0578
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0x07d0
fixme:ole:ITypeInfo_fnGetRefTypeInfo    pRefType->reference == 0xffffffff
fixme:ole:ITypeInfo_fnGetRefTypeInfo Can't find pRefType for ref 19
fixme:ole:ITypeInfo_fnGetRefTypeInfo    typekind is TKIND_DISPATCH
fixme:ole:ITypeInfo_fnGetRefTypeInfo    wTypeFlags are: TYPEFLAG_FHIDDEN 
TYPEFLAG_FNONEXTENSIBLE TYPEFLAG_FDISPATCHABLE
wine: Unhandled page fault on read access to 0x00000010 at address 
0x230f4f31 (thread 0009), starting debugger...

BTW, no luck on the bad-return-value theory, the app keeps crashing.

Alex Villacís Lasso