SYSKEY, TNG freeze, 2.0.x->TNG merge and other thoughts

[Luke Kenneth Casson Leighton]

On Wed, 9 Feb 2000, Jean Francois Micouleau wrote:

> > well, they're sent over-the-wire, right (you know this)?
> 
> yes I damn know ! I'm currently spending between 8 and 10 hours a day
> rewritting that code ! 

[i was stating things for other readers :)]

> > well, the msrpc request is split into multiple PDUs.  those PDUs are sent
> > _to_ you with SMBtrans+SMBtranssecondary OR SMBwriteXes, depending on
> > client/server DCE/RPC negotiations.
> > 
> > therefore, in order to support spoolss properly, you need that code.
> 
> what's the size of the PDUs ?

it's negotiated.  AS/U negotiates 2k.  NT and samba negotiate 0x1630
(5680) bytes.

> But in 2.0.x you can have RpcEnumShares with 10000 shares, and that's
> split in multiples PDUs ?
>

no, because the _request_ fits into one PDU (the [in] arguments are only a
serv-ername and an info level).

the _Response_ goes back in multiple PDUs (the [out] arguments).

unfortunately for the spoolss (and the svcctl code), you have to send a
STUPID buffer as an [in] argument with NOTHING in it which is the same
size as the [out] argument.

potentially this could be up to a megabyte of data.  actually, it can't be
a megabyte, because the NT MSRPC marhsalling code can't _cope_ with a
megabyte of data, SPOOLSS.EXE dr-watsons if you do that.