[ On Wednesday, May 31, 2000 at 11:49:06 (-0400), NetBSD Bob wrote: ]
> Subject: Re: Continuing sun3 nbsd142 funzies.....drive issues
>
> OK, is there anything about DEC drives that anyone knows about on
> sun3 hardware with NetBSD.  Whatever sunos is doing, it is spinning
> the drives up.  Whatever NetBSD is doing, it is not spinning up my'
> drives in question.  I have to attribute that to software and not
> hardware.  Where in the code were you looking?

src/sys/dev/scsipi/scsiconf.c

since other folks have said their drives do spin up when probed (and
indeed I've seen this), it must mean NetBSD is indeed sending the "START
UNIT" command.

I don't know what else it might be doing, or what SunOS does that might
be different.....

> OK, this still has me curious, though.  It is occurring when the bootup
> does the first scsi probing.  Sunos spins up the drives then.  NetBSD
> does not.  Apparently NetBSD is sending some code down the bus, but
> not what is spinning the drives.  After about 10-15 minutes, it times
> out and returns a REQ not found or something like that error to the
> console, and then abruptly halts.  I don't have my notes handy, so
> can't report the exact error.

You mean that under SunOS they all spin up after the second stage boot
loader is loaded but before the kernel is loaded?  I.e. the point where
in NetBSD it would have printed the boot loader revision number with
this printf from src/sys/arch/sun3/stand/ufsboot/boot.c?

        printf(">> NetBSD ufsboot [%s]\n", version);

If so then it's entirely possible that the SunOS bootloader is resetting
the SCSI bus and that's what's triggering your drives to spin up.

You should also have a read of src/sys/arch/sun3/stand/README too.

From what I can see most everything is still done with the PROM drivers,
so maybe there's another call to the prom that NetBSD could make which
would convince it to do the bus reset.

> Should NetBSD be sending something like a bus reset to force all
> drives to spin up, then do the probing?  Maybe sunos is doing that
> sort of thing, and thus, can spin up the drives where NetBSD does not?

mabye, maybe not.... I think it should be adequate to just warn folks to
always jumper their drives to spin on power up, but if the bus reset
isn't too much of a pain (it's rather yucky if you have an Exabyte),
then that's the safest thing to do.....

> I recall that when sunos boots, you can hear the tape resetting, several
> times, possibly indicative of a bus reset.  I don't recall NetBSD doing
> that, but I will have to check that to be sure.

Hmm... yes, if the tape drive does something after the second stage
bootloader loads then it's almost certainly due to a bus reset.
 
-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>