Date: Thu, 13 Jun 1996 13:01:45 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: amcrae@cisco.com (Andrew McRae) Cc: nate@sneezy.sri.com, hackers@freebsd.org Subject: Re: gated & pccard don't get along Message-ID: <199606132001.NAA08899@phaeton.artisoft.com> In-Reply-To: <199606131337.GAA21556@doberman.cisco.com> from "Andrew McRae" at Jun 13, 96 06:37:33 am
next in thread | previous in thread | raw e-mail | index | archive | help
> Having thought long and hard about this, I have come to the > conclusion that having hot-swappable resources and interfaces > is a great idea in theory, but the kernel (and parts of the user-land > and daemons) generally assumes that devices are not going to > appear and disappear at random intervals. It is pretty scary to > think of the changes required to really make the system understand > this concept fully. The net code is a good example; whilst the > insert/remove scripts can already do some of these things (like > add default routes etc.), we are really working with a bit > of glue around the edges, and not tackling some of the core > problems. One issue is the way various bits get informed about > changes [e.g a card being pulled]. The need is for programs > to be started or stopped, signals sent, kernel tables to be > modified, daemons to be informed [e.g gated] etc. You have to move to message-based event processing, of course, instead of just signal-based with daemons. The point is to *not* have to run babysitting daemons. > Berny Goodheart and I were talking about this, and his > suggestion is to implement a registry scheme, I imagine with > a graph of dependancies and some IPC etc. Tandem (Berny's > employer) uses such a scheme to implement hot swap > in their high availability architecture. Having worked on such a scheme > myself, I appreciate the complexity. Unfortunately, you can't implement > just a *little* bit of the scheme. If you do *any* form of > hot swap, you have to go the whole hog. Cisco also support > hot-swap, and even when it's designed in from day one, it is > still a significant effort to make it work. Yes, but it needent involve a registry. The problems are on the order of those addressed by the 1.2 APM BIOS specification (www.intel.com, search, APM). The issue is to allow the system to be sufficiently modular so that you can add/revoke modules as a result of a plug/unplug event. For networking, this means transient mounts (mounts are the *most* persistent network connections) and some small changes to the FS code (which require devfs to keep them small). > So I guess I am saying that the little bit of glue around > the edges is a pretty good scheme for FreeBSD, unless some > serious effort is undertaken. Thus I would consider pccard > to not really offer hot swap, but more of a `user friendly > hardware bus'. Having said that, I think the glue holds > together as much as can be expected :-) I'd like to see a serious effort. The RAID hot swap is partially supported by SCSI reprobe, and other capabilities need to be similarly supported. > I am tempted to say, "It's not pccard's fault, but all the > rest of the system for not handling hot swap". That is a comforting, > but specious argument. I *am* glad to see it is being put > to good use. It *is* the rest of the system's fault. 8-). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199606132001.NAA08899>