On Mon, 17 Feb 1997 15:26:27 -0800 (PST),
Paul Goyette <paul@pgoyette.bdt.com> wrote:
> OK, folks, I got the latest sup from today (including the very latest
> version of vm/vnode_pager.c) and rebuilt my kernel.  Everything builds
> fine, but the system hangs during multi-user startup, right in the middle
> of running fsck!

I've been experiencing this since the end of last year, I think.  More
precisely, fsck -p does hang, but fsck without -p doesn't.  Hence the
following behavior you got (thanks to Scott Reynolds for pointing it
out to me ;-):

> I can fsck the disks individually, and it works OK.  But when the
> automatic fsck runs, it hangs right after completing the second partition
> (ie, after the first partition in pass #2, per /etc/fstab).

Anyway, Scott suggested I change the option MAXBSIZE in the config file
as follows:

options		MAXBSIZE=0x8000

This unfortunately has lead to three kernels that wouldn't even boot
(panicked in the middle).  Frankly I'm at a loss.  Scott's latest
GENERIC-23 kernels exhibit the same locking-fsck-p behavior.

> Also, I booted this same kernel in single-user, and mounted /usr ('cuz I
> wanted to edit /etc/fstab with vi).  This worked OK, but when I cd'd back
> to / and tried to umount /usr, it claimed that the device was busy!  I
> _know_ for sure that I didn't have any files open on it, and my cwd wasn't
> pointing to it,so why couldn't I unmount it?

I haven't had this...  What does fstat say?

> For now, I'm back to a kernel built using last Tuesday's -current and all
> is well....

Hmm...

Ken