[c-nsp] RE: Unknown reason for High CPU 3620

[Gert Doering]

Hi,

On Fri, Nov 05, 2004 at 02:43:42PM -0000, Alex Foster wrote:
> Bridging is handled at interrupt level and is then process-switched

Dunno, but that's kind of contradicting itself.  Process switching
is what happens if you *cannot* forward the packet in the interrupt
routine.

> Routing (without CEF) is handled at process level and is then
> fast-switched

"fast-switched" is what happens if you have no CEF, and are *not* 
process switching - it's a route cache that enables the IRQ routine
to directly forward the packet.  Being a route *cache*, it has some 
issues.

> Routing (with CEF) is handled at interrupt level and is then
> fast-switched

If you do CEF, you do CEF, not "fast switching", which is a non-CEF
switching strategy.

> These may be generically incorrect - I understand some types of traffic
> are always handled at process-level ie: telnet, ping, protocol updates
> etc;

Packets being *sourced* by the router are done on process level (of
course).

gert

PS: people, please try to trim the postings before you reply to them - I
can't see any intrinsic value in receiving a 20-line article that
carries a full 650-line quote of all the thread before it in the footer,
especially if the article is only marginally connected to the thread.

-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de