Re: Driver for via-velocity gigabit NIC

Marty Connor <[email protected]> Wed, 8 Mar 2006 08:29:14 -0500
Newsgroups gmane.network.etherboot.devel
Message-ID <[email protected]>
On Mar 7, 2006, at 10:17 PM, Timothy Legge wrote:
> Well, I found the issue or more correctly I ripped out the method I  
> was using for 64 byte alignment of the rings and buffers and  
> replaced it with the method I used with the r8169 driver.  Afters  
> hours of wasted effort, the tried and true method worked in just a  
> few minutes...

As much as I am glad to see a new driver, this troubles me.  The fact  
that the driver "works" is nice, but the fact that you don't know why  
it was failing is not good.  I don't mean to be overly harsh, but  
getting something to work is not nearly enough in this case.  The  
next person who looks at your driver and compares it to the Linux  
driver will wonder why you changed what you changed.

Assuming the code worked in the Linux driver you were porting, it  
should work in Etherboot.  If it does not, we should know why.

> Afters hours of wasted effort, the tried and true method worked in  
> just a few minutes...

The time would have been worth it if you were able to understand why  
the code that evidently worked in Linux failed in Etherboot.  Now we  
have code in Etherboot that seems to work, but is gratuitously  
different than what works in Linux.

Now I certainly don't have problem re-writing drivers.  Most of the  
ones I did are derived from Linux Becker drivers that I stripped all  
the way down and rebuilt a line at a time so I knew what was going  
on.  What I do have a problem with is code getting committed that is  
not well-understood.

Could you post the code you removed, the code you replaced it with so  
we can figure out why one works and one does not?

I think it's time we stop adding functionality to Etherboot without  
gaining an understanding of how and why things work and fail.  As it  
is, we have this amazingly functional piece that few of us left  
actually understand to any depth.  This is not good.  Let's start  
fixing it now.

Marty




-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642