[ofa-general] Re: [PATCH draft, untested] ehca srq emulation (for IPoIB CM)

[Michael S. Tsirkin]

> Quoting Sean Hefty <mshefty at ichips.intel.com>:
> Subject: Re: [ofa-general] Re: [PATCH draft,?untested] ehca srq emulation (for IPoIB CM)
> 
> >We have different definitions of "across layers": in my code everything is 
> >kept
> >inside ehca. I call it a maintenance nightmare when there's code in IPoIB 
> >that
> >only ehca owners can test.
> 
> I disagree with the concept of adding this code into the lower level 
> driver.  Posting a receive buffer onto a QP after it gets a receive 
> completion is something the ULP can and should do.
> 
> SRQ support is optional.  There's no reason why the no-SRQ code in IPoIB 
> can't be tested on all HCAs.  It's the SRQ code that requires specific 
> hardware.

Basically, I think that because of lack of SW level flow control,
generally IPoIB CM without SRQ does not make sense because of
the scalabilty problems.

However, the argument for adding this protocol revolves around the claim
that ehca (the only low level driver without SRQ that we have)
is used on systems with huge amount of memory and a small number
of nodes.

-- 
MST