From owner-freebsd-questions Tue Apr 30 17:35:11 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id RAA26303 for questions-outgoing; Tue, 30 Apr 1996 17:35:11 -0700 (PDT) Received: from rocky.sri.MT.net (rocky.sri.MT.net [204.182.243.10]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id RAA26297 for ; Tue, 30 Apr 1996 17:35:03 -0700 (PDT) Received: (from nate@localhost) by rocky.sri.MT.net (8.6.12/8.6.12) id SAA05536; Tue, 30 Apr 1996 18:34:54 -0600 Date: Tue, 30 Apr 1996 18:34:54 -0600 From: Nate Williams Message-Id: <199605010034.SAA05536@rocky.sri.MT.net> To: dwhite@resnet.uoregon.edu Cc: Nate Williams , Denis Malyavin , questions@FreeBSD.org Subject: Re: Hello In-Reply-To: References: <199604300124.TAA01656@rocky.sri.MT.net> Sender: owner-questions@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > > > Do you think its worth to install FreeBSD on my system? > > > > > > It'll be tight, depending on what you put on. you need to have enough > > > space either on there or on another FreeBSD machine to build a new kernel > > > for. You _want_ to apply the Nomad PCMCIA patches. > > ^^^^^^ > > > > Hey, I take offense at that. The code in the Nomad's patches now exists > > almost completely in -current, plus there are lots of bug-fixes for the > > APM code in -current that doesn't exist in the Nomad patches. > > Yes, but most of us don't want to run -current, especially on a laptop. > We had to upgrade our Dell to the SNAP so that we could keep up with the > patches, which are based on it. But that thing won't run -current as long > as I say so. Maybe this desktop, yes, but the laptop is a toy, and we > can't spend our time fiddling with it. We spend enough time fiddling > with our machines as it is, and they're all on 2.1. Fair enough. However, realize that the Nomad patches contain lots of -current code in them. > > If people don't run the actual FreeBSD code we'll never get real working > > laptop support in a FreeBSD release. I emailed Hosokawa-san about this, > > and he agrees. The Nomad code will continue to be 'alpha' quality > > containing hacks and bad things that 'make things work', intended for > > proof of concept than actual implementations. > > Then put it in -stable. -Current is too UNstable for non-hackers to > run. The Nomad patches are too unstable for -stable. Heck, some of the patches are too unstable for -current, that's why I hacked them up. All of the 'stable' patches already exist in -stable. However, I don't want to make it unstable by adding in some of the bogus and possibly de-stabilizing patches just to get it working in -stable. Let me summarize what's been done, and what's left: - APM kernel/userland code in both is the same, and is hopefully good enough for all laptops. - PC-CARD kernel support is the same, and is almost exactly the same as the Nomad code (minus some minor hacks and formatting changes). -current -------- - The userland code is working, but slightly out of date (missing some nice features of the Nomad code). - There are some additional (useful), additional (redundant), and patches that -current doesn't contain that the Nomad patches contains. - there are some big hacks that exist in the Nomad kernel code that don't exist in -current. Some of them may be necessary for APM support on certain laptops, but I don't think so with the new patches I did which work around the *real* problem. -stable ------- - The user-land code doesn't exist - A patch to /sys/i386/i386/autoconf.c is necessary to enable the kernel PC-CARD code. - None of the drivers have been patched to add PC-CARD functionality. These patches are not -stable fodder, and although most of them are "OK", some are a bit dicey. The Nomad code simply takes the -current drivers and back-ports them to -stable, plus it contains all of the bad hacks that no longer exist in both -current and -stable. So, if you use the Nomad patches, you still have a chance of an unstable system. (Although the instabilities may not show up on laptops). What I'm trying to do is this: - Work on the understanding the user-land code. Thank goodness I waited until the end to work on it, else I would have gave up long ago. There are lots of hard-coded #'s, missing comments, and lots of magic that goes in in the user-land code. The kernel code is *much* cleaner. - Cleanup the drivers in -current, and add in the good code from the Nomad patches. Also, try to bring in the functionality of their patches into our current drivers w/out breaking them would be nice. :) - Backport all of the driver code to -stable, and provide it as a experimental patch so the next -stable release will have some patches to add this functionality. - Keep my sanity. :) In order to do this, I need *testers* who can tell me how things work, and if folks only use the Nomad code this isn't going to help me at all. Telling folks that 'You _want_ to apply the Nomad PCMCIA patches' implies that there is no other solution, or that the other solutions are somehow 'bad'. Please avoid making my job harder, as I need some help. Nate