This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: network questions
- From: Andrew Lunn <andrew dot lunn at ascom dot ch>
- To: Andrea Acquaviva <aacquaviva at deis dot unibo dot it>
- Cc: eCos Disuss <ecos-discuss at sources dot redhat dot com>
- Date: Mon, 14 Jan 2002 09:31:06 +0100
- Subject: Re: [ECOS] network questions
- References: <3C41CEED.42745A33@deis.unibo.it>
On Sun, Jan 13, 2002 at 07:16:13PM +0100, Andrea Acquaviva wrote:
> Hi,
>
> I have some questions about network stack. Since I would like to change
> the processor frequency to reduce the power consumption, I'm wondering
> how it happens if the driver and the network stack are too slow. Is
> there any timeout which may cause a network failure when the driver is
> too slow while providing packet to the higher network layers?
Generally it should not be a problem, unless you go down to a really
slow clock speeds. TCP has timeouts in terms of minutes. Maybe the
most critical is SNMP. It should respond to a query in one second or
less.
> A related question is:
> what is the meaning of CYGPKG_NET_NUM_WAKEUP_EVENTS?
> Could this parameter be important when the driver became too slow?
No. This is an array of semaphores used to wake up 'user' space threads
when there is something on their socket.
Andrew