Subject: Re: Sun 3/260 and 3/110 blues on last snapshot
To: None <nbsdbob@weedcon1.cropsci.ncsu.edu, toddpw@best.com>
From: Perry Hutchison <perryh@pluto.rain.com>
List: port-sun3
Date: 01/08/2000 13:20:55
> > I manually mounted the ramdisk.fs file, and here is what I see in /dev:
> > 
> > dagobah:11# ls /mnt/dev
> > console   fb        md0c      rcd0a     rsd0d     rsd2c     rxd0d     stdout
> > drum      kbd       mem       rcd0b     rsd1a     rsd2d     rxy0a     tty
> > eeprom    kd        mouse     rcd0c     rsd1b     rst0      rxy0b     ttya
> > enrst0    klog      nrst0     rcd0d     rsd1c     rst1      rxy0c     ttyb
> > enrst1    kmem      nrst1     rsd0a     rsd1d     rxd0a     rxy0d     zero
> > erst0     leds      null      rsd0b     rsd2a     rxd0b     stderr
> > erst1     md0       pipe|     rsd0c     rsd2b     rxd0c     stdin
> 
>                                 RAW       RAW       RAW      RAW      block?
> > 
> > So I am stumped, I don't know what I could be doing wrong here.
> 
> Does it need a set of block (sdxxx) as opposed to raw (rsdxxx) mounts?
> I don't remember offhand which it mounts, raw or block devices.

Mount requires a block device, however the mount of the initial root
file system is handled specially.  (It cannot use an entry in /dev,
since that's not accessible until the fs has been mounted.)

Going back to the start of this thread, the absence of block special
files in /dev cannot explain the observation of "no dev/console" as
the first evidence of a problem.  Surely it mounts the initial root fs
before trying to open /dev/console, and from the above it appears that
the /dev/console special file exists (although it would take an "ls -l"
to be sure that it contains the correct major and minor device numbers).
Maybe the open is failing due to a VME-related problem in the console
driver, which would explain why it works on non-VME machines.