On Thu, 2008-03-27 at 19:11 -0500, Matt Mackall wrote:

In the 486 era, when CPU performance was close to 1:1 with memory,
branches were more expensive than sequential memory fetches, and
registers were scarce, inlining made a fair amount of sense.

But now we've moved very far away from that indeed:

Systems have certainly improved but Linux is used in a
wide variety of CPU Hz, memory & register architectures.

Some of those systems haven't changed at all.
Some of those systems have sufficient cache for a
networking stack.

I think this change could negatively impact some of these
different uses and systems.

In the case of this patch, removing 60-100k from the network stack means
we're almost certainly avoiding a lot of cache misses in the big picture
while taking a few cycle hit per packet in the smallest scale.

I think the quantities of big v small are instance dependent
and it might be prudent to have the capability to keep these
functions inline.