Re: dmband: Indirection levels fix

[Robert Shearman]

Andrew Talbot wrote:
> diff --git a/dlls/dmband/band.c b/dlls/dmband/band.c
> index 891fb5a..8b89573 100644
> --- a/dlls/dmband/band.c
> +++ b/dlls/dmband/band.c
> @@ -173,7 +173,7 @@ static HRESULT WINAPI IDirectMusicBandImpl_IDirectMusicObject_SetDescriptor (LPD
>    	This->pDesc->ftDate = pDesc->ftDate;
>    if (pDesc->dwValidData & DMUS_OBJ_MEMORY) {
>    	This->pDesc->llMemLength = pDesc->llMemLength;
> -
> memcpy (This->pDesc->pbMemData, pDesc->pbMemData, sizeof (pDesc->pbMemData));
> -
> memcpy (&This->pDesc->pbMemData, &pDesc->pbMemData, sizeof (pDesc->pbMemData));
> }
> if (pDesc->dwValidData & DMUS_OBJ_STREAM) {
> /* according to MSDN, we copy the stream */

This isn't correct. Judging by the surrounding code, this should be 
allocating a block of memory of This->pDesc->pbMemData and then passing 
pDesc->llMemLength into memcpy, possibly validating that 
pDesc->llMemLength isn't greater than UINT_MAX to avoid an overflow.

dmband: Indirection levels fix

[Andrew Talbot]

Robert Shearman wrote:
> This isn't correct. Judging by the surrounding code, this should be
> allocating a block of memory of This->pDesc->pbMemData and then passing
> pDesc->llMemLength into memcpy, possibly validating that
> pDesc->llMemLength isn't greater than UINT_MAX to avoid an overflow.

Yes. I realised later that there were two possibilites: pointer copying or
memory copying, so I sent an email to Alexandre, since I didn't have my
normal usenet access at the time.

I was unable to figure out whether This->pDesc->pbMemData needs its own
slice of memory or whether it can use that which comes with
pDesc->pbMemData. I shall leave this one (and a similar one in
dmcompos/chordmap.c) and sweep on with the easy fixes.

Thanks,