Date: Sat, 2 Jan 1999 12:41:12 +0100 From: J Wunsch <j@uriah.heep.sax.de> To: freebsd-arch@FreeBSD.ORG Subject: Re: DEVFS, the time has come... Message-ID: <19990102124112.29898@uriah.heep.sax.de> In-Reply-To: <199901020438.VAA15410@mt.sri.com>; from Nate Williams on Fri, Jan 01, 1999 at 09:38:50PM -0700 References: <199901020116.RAA03885@dingo.cdrom.com> <199901020312.UAA09044@harmony.village.org> <199901020438.VAA15410@mt.sri.com>
next in thread | previous in thread | raw e-mail | index | archive | help
As Nate Williams wrote: > To counter Joerg's 'not bothered by non-persistent DEVFS', I'd have to > say that my experience with Solaris has been less than enthusiastic. > Note, it only dealt with one device (the tape drive), but it was a pain > in the butt. Well, that's funny. It's not funny that you had problems with Solaris, but it's funny since in some recent discussion in -core, I quoted Solaris as an example of a persistent DEVFS that has been causing me major grief _due to its persistence_. (Jordan's reply was that the grief is mainly caused by a poor implementation, not by the fact that they've implemented persistence. He's probably right on this.) I can't follow you on this, btw, and have just tried it again. The Solaris implementation _is_ persistent, and they store their persistent data in the root f/s (which I believe is terribly wrong). I have been able to chmod something, and to reboot, and everything was as previously. Their persistence is sticky until you boot -r (or run drvconfig(1m), and even then, old devices under /devices and /dev are remembered forever unless you manually rm the entries there (sounds like AIX, doesn't it? :). I always found this terrible, since there's no clean way to tell the system to ``go to hell and restart from scratch''. My bad experience with them was an attempt to migrate a disk from an Ultra 10 machine with a SymBios Logic SCSI card to a Sun AXi OEM-board based clone with an onboard SymBios Logic. Although both architectures are fairly similar, it has proven to be plain impossible to do this, short of a reboot from a CD-ROM after the migration, mounting the disk's root f/s, rm -rf'ing all the /dev and /devices cruft, and copying it over from the dynamically created /dev and /devices that exist in the CD-ROM environment. This is a good example of something I'd never like to happen to a FreeBSD disk. But again, that's probably more a point to prove that the Solaris _implementation_ is poor, not that persistence itself is something that sucks. But maybe then, it could be that any possible implementation of a persistent DEVFS sucks similarly. :.) (After working with some really large server machine, I began to understand why they did some things the way they did, still I believe it could have been done better.) -- cheers, J"org joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE Never trust an operating system you don't have sources for. ;-) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990102124112.29898>