Subject: Re: 2.0 RC4 softfloat build
To: None <port-mac68k@NetBSD.org>
From: Joel Rees <firstname.lastname@example.org>
Date: 11/13/2004 22:57:38
A little followup --
>>> You shouldn't do that, since you need the programs that are in /usr.
>> Yeah, I kind of remembered that. Actually, I keep hoping that sysinst
>> is somehow supposed to dodge that bullet. I guess not?
> sysinst mounts new filesystems under /mnt.
That's nice to know.
>> BTW, if I didn't mount them, sysinst didn't seem to think they needed
>> to be newfs-ed. At least, it stopped to tell me the results of the
>> newfs when I mounted them all first and didn't tell me about it when
>> I didn't. I'm not sure it mattered where I mounted them, however.
> Not sure I understand this. You generally newfs, then mount, unless
> you're using preexisting slices.
So, if I'm doing a brand-new install via ftp with sysinst, use the
booter to select the install kernel and boot, then drop out of sysinst
immediately and use the interactive mode of disklabel to set up the
mount /dev/sd0a /mnt
mount /dev/sd0g /mnt/usr
mount /dev/sd0e /mnt/var
mount /dev/sd0f /mnt/tmp
mount /dev/sd0d /mnt/home
use df to check that it's all mounted nicely and the expected amount of
free space, then
log in anonymous, cd to the RC4 softfloat directory and do an mget *,
and the download starts, gets the MD5 and checksums and gets about 25%
into the first kernel, freezes, and several minutes later reports a
SCSI timeout error like the one below and goes into a timeout loop, I
can assume hardware problems.
And if I grab another drive that I've been using with more performa
550, and try to boot the install kernel from that, and it hangs before
it asks me what kind of terminal, I can assume serious hardware
problems, probably on the mother board. And putting a full '040 in (if
I had the fifty bucks or so to spare) would be basically a SWAG. Might
work, might not.
This thing has an Apple brand TV card in it. I suppose I really should
have removed that before playing with netBSD.
>>> Use the /mnt mountpoint, and mount /mnt/usr there, and untargzip to
>> untargzip? tar -zxf what?
>> Is there a tarball that contains the contents of /usr? Or is the idea
>> to pipe tar to tar instead of using cp -p?
> When I'm installing manually, after I've got the partition table all
> set up, I newfs, mount under /mnt, /mnt/usr, et cetera, then mkdir
> /mnt/install/, ftp all of the sets to that directory, then, in /mnt,
> for file in install/*.tgz
> tar xzpf $file
> cd /mnt/dev ; ./MAKEDEV all
> Many of the sets put stuff in usr, so there's no set that just has usr.
Wish I could get far enough to do it by hand like this.
>> The timeout errors look something like this I copied by hand to the
>> list back in May:
>>> sd0(esp0:0:0:0) esp0: timed out [ecb 0x104b000 (flags 0x1, dleft
>>> 10000, stat 0)] <state 5, nexus 0x104b000, phase (l 3, c 0, p 0),
>>> resid 10000, msg (q 0, o 0) DMA active> AGAIN
>>> sd0(esp0:0:0:0) esp0: timed out [ecb 0x104b000 (flags 0x41,
>>> dleft 10000, stat 0)] <state 5, nexus 0x104b000, phase (l 3, c 0, p
>>> 0), resid 10000, msg (q 20, o 0) DMA active> AGAIN
>>> cd0: async, 8-bit transfers
>>> sd0: waiting for pack to spin up ...
> Have you tried playing with termination? A different drive?
Yeah, I just did that about half an hour ago. The results are as I
>> Wish I had more time to give this. But what I really wish is that I
>> could afford a full 68040 to plug into this beast.
> What kind of machine is this?
>> I should mention again that I have reason to believe the box has
>> hardware problems. In Mac OS, Netscape tends to bomb a lot,
>> CodeWarrior tends to trip out in mid-compile sometimes, that kind of
> A different drive might help.
Thanks. I guess I can be pretty sure at this point that it's hardware.
even though much of what I do is not sensible
it does make sense if you know why ...