On Fri, Apr 08, 2005 at 02:25:34PM +0200, Thomas Hood wrote:
> First let me say that I don't think that the current system is ideal.
> However, it does work, and it is the best that we can do so long as we
> can't make changes to the ifupdown package.
>
>
> On Thu, 2005-04-07 at 18:03 -0700, Jean Tourrilhes wrote:
> > A) The current solution only deal with events generated at
> > step 6, therefore is incomplete. Events generated at 2 are always
> > lost, events generated at 4 are usually lost.
>
>
> I don't think that that is true. Anything that causes a hotplug event
> will start a net.ifup process. Please explain why you think that events
> get lost.
Compile network driver statically in the kernel (i.e. not as
module). Boot kernel. Verify that event is lost. That's number 2.
For number 4, I did not check exactly what goes into the
initrd, as I am personally not using that.
> > B) Forking a script adds overhead, this overhead occur in
> > every case, even if ifup is available. Most often, ifup is already up
> > and this overhead should not occur.
>
>
> The overhead is insignificant.
Every little count, that's the death by a thousand cuts. I can
tell you that on my 400MHz box, Testing is booting much slower than
Stable. Obviously, It's very hard to pinpoint the blame, but I can see
that hotplug seems to take some time.
The point is that the current solution has overhead, while my
proposal has not.
> > C) A process is an expensive way to keep a simple state, each
> > waiting process consume memory during the boot process. Let's not even
> > talk of the busy wait.
>
>
> Polling is an easy way to enforce synchronization.
Synchronisation is not necessary.
> > o net.agent drop/ignore events if called before net.agent,
> > otherwise calls directly ifup (no forking).
>
>
> ... if called before net.enable is created?
Yes.
Actually, I forgot that ifup already verify if it's enabled or
not. So, you could just call ifup in every case, and remove net.enable
altogether. That would simplify things even more.
> Cheers!
> --
> Thomas Hood <[EMAIL PROTECTED]>
Jean
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
