>3.  Michael VanLoon reported problems with the July 23rd (I think)
>tar_files.  This isn't my experience, although I'm running with old
>bootblocks (1.19), and I'm only going to go to 4.4 inodes when I
>absolutely have to.  In fact, I had absolutely no trouble upgrading
>this week.

My experience is that you are running on borrowed time if you try to
run an old filesystem format under a new kernel.  Mine got quite
mangled, and I was only able to unmangle it by running an old kernel
and old fsck on it.  Once I updated it to the new "fsck -c 2" format,
though, everything seems to be working just peachy under the new
kernel.  Remember, though, that the fsck -c 2 step is a never-look-
back step, since the old kernel will cease to work once you upgrade
the filesystem.

My problem with the source tar files (the 17th, actually) wasn't a
source problem at all.  Rather, I had some old cruft in my kernel
build dir and libkern dir that wasn't getting removed by make clean.
If you update to kernel sources after being dormant for awhile, make
sure to remove all the .o and .depend files from libkern/ or
libkern/obj/ (depending on your setup), and remove all the .o, .h,
.depend, and genassym files from your kernel compile dir.  Doing this
made kernel builds finally work for me.  Oh, of course, don't forget
to install new includes first, and build a new config before
attempting a kernel build.

>9. Last time I produced one of these reports, this was the i386
>supported SCSI device list. Could someone please update me with
>any changes?
>
>bt742   eisa                 bt742a.c        deraadt@fsa.ca
>bt747   eisa 		     bt742a.c        michaelv@iastate.edu

As far as I know, all the BusLogic cards work with the standard bt
driver now.  I know that, at least, my bt747s does.

>[The Adaptec 2742 is not supported, and may not be in future due to
>Adaptec's refusal to release "proprietary information" about the board.]

Not only the 2742, but the entire Adaptec 2xxx line.

>(There has been some discussion over the best SCSI board to purchase. 
>General reasoning was that Adaptec support wasn't good for non-DOS
>problems, and re the 2742 issue.  The Ultrastore has jumpers for
>everything, which can be a pain.  The Bustek boards have had
>favourable reports.  Please note that I'm just condensing other
>people's opinions here, all you lawyers out there.)

For PCI people, the NCR 810 SCSI chip/board is supposed to be the next
great thing, and is significantly cheaper than anything else, as well.
I haven't used one in a NetBSD system, but I've used one in a 66MHz
Pentium Windoze box, and can vouch that they're several times faster
than a 1542c in a 90MHz Pentium of the same manufacturer (both
machines also had 32MB of RAM).  Haven't had a chance to compare one
to a bt946c, however.

>13.  For anyone a trifle nervous, having seen stories of not being
>able to boot their root filesystem etc, I'm taking quite a
>conservative line on this - I'm still using the 1.19 bootblocks, I've
>got 4.3 style inodes, and I'm running on SCSI disks.  I repeat that
>I've had no problems this week upgrading.  The problems that I have
>seen over the last few months (and which have all been fixed) are:

Be warned that you could see a huge mess next time you reboot.  You
may not see any problems until then, however.  Proceed without
upgrading at your own peril.

Thanks for posting the snapshots again, Alistair.  If nothing else, it
gives us a sounding board to rehash the previous week's development.

-----------------------------------------------------------------------------
 Michael L. VanLoon                 Iowa State University Computation Center
    michaelv@iastate.edu                    Project Vincent Systems Staff
  Free your mind and your machine -- NetBSD free Un*x for PC/Mac/Amiga/etc.
-----------------------------------------------------------------------------

------------------------------------------------------------------------------