another trivial rpc_api_pipe_req patch

[Ronan Waide]

On April 27, waider at waider.ie said:
> When a data packet overflows a single write request, the current code
> increments the RPC callid for each request. This results in a "pipe
> busy" error since the NT side is attempting to assemble the data for
> the first call on the FID in question before it does anything
> else. The trivial fix is to move the callid declaration out of the

Just to add that the trivial fix may not be sufficient. I've not
tested against Win2K, but NT4SP6 definitely seems to require that raw
pipe mode is enabled and that the unexplained "- 8" in the max_data
calculation is removed.

Cheers,
Waider.
-- 
waider at waider.ie / Yes, it /is/ very personal of me.
"What is this talk of "release?"  We do not make software "releases."  Our
 software "escapes" leaving a bloody trail of designers and quality assurance
 people in its wake." - MoncriefJM at gvl.esys.com