triple allocation of large buffers.

[Jeremy Allison]

On Thu, Jan 17, 2008 at 10:00:24AM +0000, Amin Azez wrote:
> Jeremy/Stevef please can I have a trans2 sub-opcode?
> 
> * Andrew Bartlett wrote, On 16/01/08 22:16:
> > ...
> > Not having looked at the code, I wonder if vfs_cifs gets away with a lot
> > because the structures and packets being passed along are identical, so
> > much less munging is involved.
> >   
> There's actually a lot of munging as the calls are all mapped to a
> generic form, but they all preserve the pointer to the pre-allocated
> response (which gets munged back to) so when the proxied read is
> received the read-data is unpacked directly to the pre-allocated
> response as referenced by a pointer that passed unscathed through a few
> layers of munging.
> 
> As said to Stefan, I would have to provide storage for an [in,out]
> uint8* x; and merely supply the pointer in the rpc struct, so I am sure
> something can be done based on that to have the blob unpacked directly
> to the pre-allocated response. I'll investigate.
> > I'm sorry you seem to be getting the run-around.  This is a new area,
> > and trying the different options is I hope worthwhile, in the long term.
> > (If frustrating in the short term). 
> >   
> :-) It needs doing anyway... I'm grateful for the guidance of you guys.
> > Perhaps I could make another suggestion:  Put the data in normal CIFS
> > packets, but marshal your header with NDR?
> >   
> I was thinking this last night and Stefan made the same suggestion. The
> code in dcerpc.c/dcerpc_ndr_request_send() seems to lend itself to this
> idea, so I think that is what I will do.
> 
> I've tried to get a sub-opcode from Mr French but not been successful
> yet, maybe now I've exhausted alternatives... please guys?

Not ignoring you, just really busy. I'll talk to Stevef in the
next couple of days and get back to you.

Thanks,

Jeremy.