Subject: Re: NetBSD/Alpha -- comments after initial install on a DEC 3000/500
To: Jeff Kellem <email@example.com>
From: Chris G Demetriou <Chris_G_Demetriou@LAGAVULIN.PDL.CS.CMU.EDU>
Date: 03/13/1995 17:47:16
> I spent a little time this weekend installing NetBSD/Alpha on a
> DEC Alpha 3000 Model 500 (aka Flamingo), starting with kernel build #3,
> dated 18 Feb 95, 7:22:06am.
> The first problem seemed to be with the RRD42 (DEC CD-ROM drive). The SCSI
> driver seemed to recognize the device -- it displayed the vendor string and
> realized it was a CD-ROM drive. After that, the following messages were
> [ ... ]
> At this point it just hung. Disconnecting the RRD42, got rid of those
> messages and allowed the boot process to continue. (Of course, I'm not
> concerned with the PMAF-FA boot message. ;)
yup, this is because the SCSI code _SUCKS_. Charles Hannum has
threatened to write a new version of the 5394 SCSI code (which should
be usable on bothe the SPARC and on the Alpha), but hasn't finished it
> One of the consistent messages reported during boot is:
> nfs_nhinit: bad size 352
> indicating that the nfsnode struct size is not what was expected. This
> has happened on each of the three generic kernels tried so far.
right; this is an annoying warning, that should probably be removed
from the source code. (Kirk McKusick removed the analogous warning in
the UFS code...)
> Another error seen during boot include:
> starting network
> le0: error: stat=a2f3<ERR,CERR,TINT,INTR,INEA,RXON,TXON,STRT,INIT>
i forget what this is, but i've never seen a boot without it, either. 8-)
> With each of the three kernels, including build #94, dated 12 Mar 95,
> 5:34:39am EST, it panics with the following:
> panic: kmem_malloc: kmem_map too small
> after various seemingly random periods of time (but usually not very long
> and with basically only system activity going on).
that's interesting; i've seen that error exactly once, i think.
how much RAM do you have? (i'd bet that you've got more than I do,
and the constants i picked are just too small...)
> There've only been a couple "panic: trap" messages throughout the weekend,
> though it's unknown what was happening at the time.
i don't see many of these at all. obviously, if you note the PC and
the RA, you can gdb /netbsd, and disass the addresses, to see where
they happened, when you reboot...
> I haven't had time to delve into the problems or the code, itself. This is
> just a note to indicate what has happened so far and what things might seem
> a little odd. If I'm lucky to have the machine up long enough when I have
> time to work on it, I might even get a chance to build a new kernel. ;)
I'm glad to hear that you got it installed and running. "warm