Subject: Re: File I/O in hardware device driver
To: Greywolf <email@example.com>
From: Matthew Jacob <firstname.lastname@example.org>
Date: 07/28/2000 11:55:19
On Fri, 28 Jul 2000, Greywolf wrote:
> On Fri, 28 Jul 2000, Matthew Jacob wrote:
> # > That is not a good thing to do. A driver should not be opening
> # > files. That's for applications to do.
> # For shame, Eduardo! And you know as well as I do that this is exactly
> # what the kobj modules do in Solaris when they read /etc/system (and the
> # .conf files).
> Not, mind you, that Solaris is a shining example of how to handle kernel
> loading, in my mind. Slow as hell. Good, I suppose, if you're determined
> to distribute without source access, but I really hope we're not headed
> that direction (/etc/system, .conf files and the whole mess (isn't that
> a direct hand-me-down from SVRx? (No, I'm not bitching about the SVRx,
> I just want to know WHERE the hell Solaris got it -- it's not original
> to Sun, is it?))).
Yes. It's from Bill Pittore and Joe Provino from Sun's East Coast division-
they developed it out of the modload stuff they did for RoadRunner (4.0 based
i386- 25Mhz!). Joe Provino, btw, worked with Bruce Cutler at DEC on RSX-11
before Cutler went off to do VMS.
Jeez- this whole thing could turn into another dump technology love fest, so I
won't answer your comments unless you want to chat via private email... :-)