[ofa-general] [PATCHv2] IB/ipoib: S/G and HW checksum support

[Jason Gunthorpe]

On Thu, Sep 06, 2007 at 01:49:58PM -0500, Todd Rimmer wrote:

> Michael's proposal is a nice optimization for the direct host to host
> case.

Right, that is probably where it is best used, this is easy if the
option is RC/UC only and is negotiated. Gateway devices could just
never allow it to be negotiated on.

Unless Michael gets TSO and LRO working too, then having gateways,
which are more like offload capable VNICs now, supporting the offload
features would be a benifit.

> However as soon as a gateway/router (B above) is added there is a
> serious gap in the integrity domains.  A hardware problem (or software
> bug) in B could undetectably corrupt the packet, but it would be
> delivered to C with a valid checksum.  Hence an undetected data
> corruption for the overall network path A<->C.

The counter to this is IB already has lots of things like this. SDP,
VNIC (well, with TSO or csum offload), ISER (gateway'd to FC, SATA,
ethernet, etc), etc all lack true end to end integrity. All rely on
the gateway device to have enough internal error controls. Basically
all the IB gateway type apps except for IPoIB lack true end to end
checking. 

If that was really a problem you'd have a hard sell with FC gateways
and the like too :)

That is why I like the name peer to peer vnic for this kind of feature.

I didn't like it too at first, but the notion has grown on me  :)

Jason