Yes - it fails (Samba returns error invalid parameter) without this change.
On Mon, Feb 25, 2019 at 8:21 AM Jeremy Allison <jra@xxxxxxxxx> wrote:
>
> On Mon, Feb 25, 2019 at 08:43:33AM +0100, Stefan Metzmacher wrote:
> > Yes, that correct, I missed that sorry.
> >
> > Does your client still work without that chance, just sending 8 zero
> > bytes? Or does that fail?
> >
> > I guess it will fail at
> >
> > + if ((inbuflen % 16) != 0) {
> > + return smbd_smb2_request_error(
> > + req, NT_STATUS_INVALID_PARAMETER);
> > + }
>
> Yes, it's supposed to fail there :-).
>
> > I think we need to have something like this before:
> >
> > static const uint64_t reserved64;
> > DATA_BLOB reserved = data_blob_const(&reserved64,
> > sizeof(reserved6464));
> > int cmp;
> >
> > cmp = data_blob_cmp(&in_posix->data, &reserved_value);
> > if (cmp == 0) {
> > inbuf = (const uint8_t *)SMB2_CREATE_TAG_POSIX;
> > inbuflen = 16;
> > outbuf = reserved.blob;
> > outbuflen = reserved.length;
> > }
>
> I don't want to do that. None of this code is in any
> mainline/master/released code bases so I don't want
> to support what is "test" code in the Linux client.
>
> We are close to the point of making something
> that we can commit to support going forward,
> but we're not there yet.
>
> So I really don't want to support any Negprot
> variants other than the latest "list of create
> contexts" version.
--
Thanks,
Steve
