Re: [V9fs-developer] [PATCH v5 11/11] net/9p: allocate appropriate reduced message buffers

[Christian Schoenebeck]

On Mittwoch, 13. Juli 2022 11:29:57 CEST Dominique Martinet wrote:
> Christian Schoenebeck wrote on Wed, Jul 13, 2022 at 11:19:48AM +0200:
> > > - for this particular patch, we can still allocate smaller short buffers
> > > for requests, so we should probably keep tsize to 0.
> > > rsize there really isn't much we can do without a protocol change
> > > though...
> > 
> > Good to know! I don't have any RDMA setup here to test, so I rely on what
> > you say and adjust this in v6 accordingly, along with the strcmp -> flag
> > change of course.
> 
> Yeah... I've got a connect-x 3 (mlx4, got a cheap old one) card laying
> around, I need to find somewhere to plug it in and actually run some
> validation again at some point.
> Haven't used 9p/RDMA since I left my previous work in 2020...
> 
> I'll try to find time for that before the merge
> 
> > As this flag is going to be very RDMA-transport specific, I'm still
> > scratching my head for a good name though.
> 
> The actual limitation is that receive buffers are pooled, so something
> to like pooled_rcv_buffers or shared_rcv_buffers or anything along that
> line?

OK, I'll go this way then, as it's the easiest to do, can easily be refactored 
in future if someone really cares, and it feels less like a hack than 
injecting "if transport == rdma" into client code directly.

Best regards,
Christian Schoenebeck