netdev
[Top] [All Lists]

Re: Compile-time versus run-time

To: Andrew Morton <andrewm@xxxxxxxxxx>
Subject: Re: Compile-time versus run-time
From: Jeff Garzik <jgarzik@xxxxxxxxxxxxxxxx>
Date: Mon, 26 Mar 2001 10:19:43 -0500
Cc: "netdev@xxxxxxxxxxx" <netdev@xxxxxxxxxxx>
Organization: MandrakeSoft
References: <3ABF2F8E.B212A96B@uow.edu.au>
Sender: owner-netdev@xxxxxxxxxxx
Andrew Morton wrote:
> 
> CONFIG_8139TOO_TUNE_TWISTER
> 
>         This implements a function which might come in handy in case you
>         are using low quality on long cabling. It tries to match the
>         transceiver to the cable characteristics. This is experimental
>         since hardly documented by the manufacturer.  If unsure, say N.
> 
> Jeff,
> 
> don't you think this sort of thing should be a module option, and
> not a compile-time option?
> 
> It's OK for you and me - we compile kernels occasionally.  But
> for most people, unless the distributor turns this on, they simply
> won't be able to access it.
> 
> No?
> 
> (And wouldn't it be nice to be able to get the same functionality
> which module options give us when using a statically linked driver?)

hmm..  I guess we could make this something enabled via ioctl.

I agree with your general point, but this option is marked experimental
and defaults to no.  Potentially the twister tuning varies based on chip
(or even silicon or board revision), so while I want to provide the
capability to those who can experiment with it, I'm not sure I want to
provide it for general use..

        Jeff


-- 
Jeff Garzik       | May you have warm words on a cold evening,
Building 1024     | a full moon on a dark night,
MandrakeSoft      | and a smooth road all the way to your door.

<Prev in Thread] Current Thread [Next in Thread>