Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!news.unimelb.EDU.AU!munnari.OZ.AU!news.ecn.uoknor.edu!solace!umdac!newsfeed.sunet.se!news00.sunet.se!sunic!nntp.coast.net!chi-news.cic.net!news.synet.net!imdave From: imdave@synet.net (Dave Bodenstab) Newsgroups: comp.unix.bsd.freebsd.misc Subject: "hack hardware" drivers; was Linux vs. FreeBSD ... Date: 24 May 1996 21:04:42 GMT Organization: Dave Bodenstab's home machine Lines: 74 Sender: Dave Bodenstab Message-ID: <4o589a$t41@garuda.synet.net> References: <3188C1E2.45AE@onramp.net> <319BDC0B.75DDA671@lambert.org> <4o13i3$tp@news.siemens.at> <31A5367F.1FAD29B4@lambert.org> NNTP-Posting-Host: dial2.synet.net As a FreeBSD user, I would tend to agree with Mr. Molnar. I was involved with Unix from ~1982 (porting SVR2, and supporting/enhancing MP SVR3 and SVR4.) I chose FreeBSD over Linux due to its roots (Berkeley) which I expected to increase its stability vs. Linux. I've been extremely pleased with FreeBSD. I would like to recommend to anyone who asks that FreeBSD be their choice of OS. If their hack/cheap/etc hardware might be the deciding factor between Linux and FreeBSD, then it seems to me that if the lack of a "clean" driver prevents someone from using FreeBSD, then that is a problem. -- In article <31A5367F.1FAD29B4@lambert.org>, -- Terry Lambert <terry@lambert.org> wrote: -- >Ingo Molnar wrote: -- >] but it really doesnt render your OS useless or less clean ... -- >] there are alot of device drivers with code quality ranging from -- >] "perfect" to "poor". -- > -- >A hack driver may, in fact, render your kernel unbootable on -- >a large amount of hardware. -- > -- >Major examples include ATAPI CDROM drives, Lance ethernet, and -- >Floppy tapes (QIC 40/80/120 devices), and the PS/2 style mouse -- >(which hook to the keyboard controller, which is a notoriously -- >unstable interface anyway), all of which must be intrusively -- >probed. -- > -- >Intrusive probing is inherently evil. -- > -- >Any hardware that requires intrusive probing to identify it -- >is "hack hardware". Seems simple enough to me: don't probe! If the hardware in question cannot be probed non-intrusively, then don't. Just fall back to earlier days when the kernel had to be manually configured to exactly match the hardware configuration of the box on which it was to be run. Wouldn't this be win-win? "Hack" drivers would be included in the kernel not by default, but by the users that have the hardware and want the driver. *They* would enable the driver. The lack of a device driver would no longer necessarily be an impediment to one using FreeBSD or Linux. -- >There are only a *few* broken drivers in the FreeBSD kernel, -- >and you must explicitly go off and enable them by using the -- >-c argument at the boot prompt. Then no new precedent has to be set. -- > -- >] i would agree with you if we .... had to support 100000 buggy MMUs, -- >] which ... make it impossible to support 2 levels of paging ... this -- >] would be a compromise "right in the heart of the OS", which would -- >] be inacceptable. -- > -- >The problem is not in providing a driver for crappy hardware; -- >that's actually very easy. The problem is in using that driver -- >when the crappy hardware isn't present, and it causing perfectly -- >good hardware to not operate. Then don't include the driver in the GENERIC kernel. I honestly don't see that making FreeBSD useful to a wider audience would be a bad thing. (I'm being selfish here... if due to mass marketing FreeBSD garners fewer and fewer users to the point that the *wonderful* development effort is abandoned, then I'd lose too!) Respectfully, Dave Bodenstab imdave@synet.net