Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 8 Jul 2000 22:13:41 +0200
From:      Stefan Esser <se@freebsd.org>
To:        Stephen McKay <mckay@thehub.com.au>
Cc:        Alan Edmonds <aedmonds@digitalconvergence.com>, Bill Paul <wpaul@freebsd.org>, Chris Wasser <cwasser@v-wave.com>, freebsd-stable@freebsd.org, Stefan Esser <se@freebsd.org>
Subject:   Re: Strangeness with 4.0-S
Message-ID:  <20000708221341.B2104@StefanEsser.FreeBSD.org>
In-Reply-To: <200007041411.AAA18590@dungeon.home>; from mckay@thehub.com.au on Wed, Jul 05, 2000 at 12:11:08AM %2B1000
References:  <200007030749.RAA13446@dungeon.home> <20000704140131.A1734@StefanEsser.FreeBSD.org> <200007041411.AAA18590@dungeon.home>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2000-07-05 00:11 +1000, Stephen McKay <mckay@thehub.com.au> wrote:
> On Tuesday, 4th July 2000, Stefan Esser wrote:
> 
> >It is just not necessary to disable the optimization, since it 
> >will cost a few retransmissions (and the driver will know that 
> >the frame was not successfully sent and can retry immediately 
> >with the modified buffer setting).
> 
> On my systems (multiple affected by this sort of thing) I get a long
> and annoying pause as the card resets and renegotiates speed and half/full
> duplex with the switch.  It is very noticeable and quite frankly not
> acceptable.  I had to hack out all the clever auto fallback because
> otherwise I would have thrown my computer out the window.

Oh, there are renegotiations after each overrun ???
They should not be required at all. The Ethernet chip probably supports
writing a new prefetch limit into the register while fully active ...
I have looked at a number of Ethernet controller data sheets. There never
was a warning that the chip must be quiescent when the "early send" limit
is modified.

> That's why I think that setting the default ever-so-slightly slower, but
> without big hiccups, is better than the current situation.  Of course, add
> a "maximum performance" switch if you want, but no regular user will find
> fault with a default of store and forward.

Well, I'd rather have the driver changed to not require a re-negotiation 
of the transmission parameters.

But as long as this is not the case, store-and-forward will at least 
hide that there is a problem ;-)

Regards, STefan


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000708221341.B2104>