atom feed32 messages in org.freebsd.freebsd-currentCurrent is Really Broken(tm)
FromSent OnAttachments
Alfred PerlsteinSep 23, 1998 1:38 am 
Jordan K. HubbardSep 23, 1998 2:29 am 
Alfred PerlsteinSep 23, 1998 3:42 am 
Peter WemmSep 23, 1998 4:24 am 
Jordan K. HubbardSep 23, 1998 4:35 am 
Bruce EvansSep 23, 1998 5:02 pm 
Terry LambertSep 23, 1998 6:03 pm 
David HollandSep 24, 1998 1:52 pm 
Archie CobbsSep 24, 1998 6:20 pm 
Bruce EvansSep 25, 1998 6:37 pm 
Poul-Henning KampSep 25, 1998 11:31 pm 
Justin T. GibbsSep 26, 1998 9:52 am 
Poul-Henning KampSep 26, 1998 10:32 am 
David HollandSep 26, 1998 7:45 pm 
David HollandSep 26, 1998 8:03 pm 
Poul-Henning KampSep 26, 1998 10:56 pm 
David HollandSep 26, 1998 11:59 pm 
Poul-Henning KampSep 27, 1998 12:21 am 
Greg LeheySep 27, 1998 1:11 am 
Greg LeheySep 27, 1998 1:26 am 
David HollandSep 27, 1998 1:32 am 
Terry LambertSep 27, 1998 1:58 pm 
Terry LambertSep 27, 1998 2:40 pm 
Jeremy LeaSep 28, 1998 7:19 am 
Eivind EklundSep 28, 1998 9:29 am 
Poul-Henning KampSep 28, 1998 10:49 am 
Julian ElischerSep 28, 1998 12:24 pm 
Larry S. LileSep 28, 1998 12:39 pm 
Poul-Henning KampSep 28, 1998 12:51 pm 
Larry S. LileSep 28, 1998 1:56 pm 
Terry LambertSep 28, 1998 4:47 pm 
David HollandSep 28, 1998 4:55 pm 
Subject:Current is Really Broken(tm)
From:Alfred Perlstein (bri@hotjobs.com)
Date:Sep 23, 1998 1:38:59 am
List:org.freebsd.freebsd-current

(if you are involved with devfs, libdisk or CAM it's important that you read this imo)

several major problems i've been smacked with the last couple of days:

i'm using CAM, softupdates, devfs, ipfw, posix_sched and mfs, this is a -current system of about mid sept 22nd (yesterday)

before you click your commiter teeth and say "don't run -current" let me say, i enjoy it, and try to give feedback on bad things like what i'm posting now. with the release date so close i hope someone has time to look at these issues:

1a) specifing the wrong "da" entry to mount and it will hang, you can ^C it though, then i tried to use disklabel on it, that hung and ^C didn't help. after that operations on the device reported "device busy" then i isuued a reboot, this caused a panic, which led to the next $@#$!@#$%!@ bug. 1b) after setting my dump device to /dev/wd0s1b after a crash freebsd happily wrote over a real partition with the system dump. this has happened twice, first my /usr on one box, then /var on another.

2a) the cam scsi devices (da) don't seem to appear (in devfs) until you mount them without devfs mounted, then unmount, then mount devfs, then remount, which leads us to our next bug 2b) after doing that, a ls in /dev (with it being devfs) causes a panic.

3) many applications still don't properly support DEVFS, sysinstall "can't find any devices" and disklabel returns an undefined error or hangs when accessing a partition

do i have bad hardware? no i think not as my pre-CAM/+SLICE kernel happily chugged away at several buildworlds before rebooting into my worst nightmare.

i noticed the "crash dumps were broken, but fixed now messages" let me assure you they are not.

with -current falling over on a ls, i hope someone has the time to contact me before a release.

my box at home is staying in this state for a couple of days if anyone wants me to test/do anything.

i'll cvsup in a week or so assuming that: a) someone fixed some/most of this b) noone gave a hoot

To Unsubscribe: send mail to majo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message