Return to BSD News archive
Path: sserve!manuel!munnari.oz.au!mips!mips!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!ames!agate!soda.berkeley.edu!wjolitz From: wjolitz@soda.berkeley.edu (William F. Jolitz) Newsgroups: comp.unix.bsd Subject: Re: taunted by happy extract command Message-ID: <14g475INNknj@agate.berkeley.edu> Date: 21 Jul 92 04:35:49 GMT References: <1992Jul20.213814.10944@athena.mit.edu> Organization: U.C. Berkeley, CS Undergraduate Association Lines: 77 NNTP-Posting-Host: soda.berkeley.edu In article <1992Jul20.213814.10944@athena.mit.edu> ccount@boltzmann.mit.edu (Dr. Craig A. Counterman) writes: >I've been following 386BSD since March, but not too closely since I >had a scsi-only system and 0.1 would be out in another week... Sorry. But look on the bright side, if I was still at Berkeley you'd have been lucky to see it within the same year }| -} ! >But /dev/rwt0 doesnt seem to work (device not found or some such) I >didn't try this too hard though. My qic drive is at interrupt 7, >which is a bit unusual. It is getting polled on boot though, I think. I doubt it. As in the INSTALL.NOTES, only at the given address and IRQ/DRQ will it work (the bronx cheer you get on reset with a tape present has nothing to do with software -- every QIC board I've seen has no boot ROM! >But the show-stopper is this: During 'extract' of bin01, it crunches >away for a while, then panics and reboots. The panic (glimpsed >briefly on the third go-round) involved >'panic kmem_malloc kmem_map too small'. This is due to the fact that the buffer cache is allocated out of kmem_map, and is undersized for systems above 12MB (you lucky devil, I've only 8MB!). The number of such entries are statically defined, and a temporary fix for this is to double the number (500->1000), and will be in patch set 1 (if I can ever get finished with mail long enough to MAKE patch set 1...). >Of course, I didn't know what was going on the first two times. I >started it going, left the room, and came back in a few minutes to >find it giving me a login: prompt. But things weren't quite right, >fsck wouldn't run (something about /etc/fstab or disktab being bad), >and before long it'd crash complaining about a bad directory entry. As the notes say, start over if things go wrong with "happy fun extract". It's difficult to unravel at this point, the most *vulnerable* of all( it used to be worse, you'd have to hope that all floppies read successfully on 0.0, over quite a period of time...). This will improve. >So: any clues, offhand? My guess is that I have too much memory for >the dist kernal. Correct, bang on! >System: >486-50 clone AMI/OPTI with 20MB memory, 256K cache (how come everyone has bigger systems than we do ....) >AHA-1542B scsi, with two 200 meg disks >QIC-02 mountain 60meg tape drive >Prodesigner IIs >Logitech bus mouse (IRQ 5) Ah. A bus-mouse. People have been suspicious of these rodents on 386BSD... >Serial card with one 16550 >1.2 and 1.44 floppies > >I'm planning on adding a scsi CD-ROM, a soundblaster, and more disk. > >What, if any, problems are there with this setup? None, other than I don't as well ! Seriously, it looks just fine. You'll find that if you get a SCSI tape instead of QIC-02 it will stream, and run circles around the other. SCSI really makes quite a difference. >Another question: How can I easily switch from DOS to BSD? I've been >using fdisk to change the active partion, but I gather there's a >better way. At the moment, that's it. I've had mixed luck with BOOTANY. I've heard of something called SHOELACE, but don't have a copy (yet). OS/2's partition switcher has gotten mixed reports with 0.1. I expect that there will be some fine person who will make an excellent contribution here, as will someone write a good partition management program for 386BSD (hint, hint). Bill.