Re: kernel > 2.1.36 & nfs

Linus Torvalds (torvalds@transmeta.com)
Tue, 3 Jun 1997 16:12:21 -0700 (PDT)


On Tue, 3 Jun 1997, Alan Cox wrote:
>
> > reception is definitly a multi-CPU issue (one CPU takes the actual
> > interrupt and creates the packet, but it's very possible that the packet
> > will be used on another CPU).
>
> It is far better at least on Intel that we don't do that - the cross cache
> copy is expensive.

What cross-cache copy?

When we're talking really high performance, we're talking DMA and cards
that do the checksum on the card themselves. In that case the data never
even touches the caches on the CPU that takes the interrupt - or at least
only the headers for finding out what the packet is all about.

> The header can be in multiple parts. Also stuff like the NFS rpc code would
> need a rewrite to cope with fragmented rpc headers

I certainly agree on the need to have the headers in one fragment, but
that still doesn't mean we need to reassemble _all_ fragments.

Linus