Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 28 Mar 1995 23:50:44 -0800
From:      Paul Traina <pst@shockwave.com>
To:        Bruce Evans <bde@zeta.org.au>
Cc:        jkh@freefall.cdrom.com, root@shockwave.com, freebsd-bugs@freefall.cdrom.com
Subject:   Re: kern/280: new slice manager totally confused about old slice disks 
Message-ID:  <199503290750.XAA00464@precipice.shockwave.com>
In-Reply-To: Your message of "Wed, 29 Mar 1995 17:12:22 %2B1000." <199503290712.RAA20923@godzilla.zeta.org.au> 

next in thread | previous in thread | raw e-mail | index | archive | help

  From: Bruce Evans <bde@zeta.org.au>
  Subject: Re: kern/280: new slice manager totally confused about old slice dis
>>ks
  >  Subject: Re: kern/280: new slice manager totally confused about old slice 
>>dis
  >>>ks 
  >  > 1) It looks like the implicit label stuff for unlabeled disks is
  >  >    non-functional?  I cannot ask disklable to read the pseudo-label off
  >  >    of sd1
  
  Did you install a current MAKEDEV and run it?

Yes, I just didn't understand what I did at the time... all is working fine
now, very very nice job overall.
  
  >  Maybe if you specify /dev/sd0d explicitly?  Works for me!
  
  It shouldn't work.  The d partition is no longer special.

Yep.
  
  >root@precipice$ disklabel /dev/sd1d
  >disklabel: /dev/sd1d: Invalid argument
  
  Correct.  Perhaps the errno's could be improved.  I seem to have broken
  ENXIO being returned for opens of nonexistent partitions (EINVAL is now
  returned) while fixing(?) sd's return of ENXIO for existent empty
  partitions (wd always allowed opening such partitions).  Both drives
  always returned the confusing errno EINVAL for nonexistent labels.

  >root@precipice$ disklabel sd1
  >disklabel: ioctl DIOCGDINFO: Invalid argument
  
  disklabel(8) translates sd1 to /dev/sd1c.  Apparently you have no label
  on sd1c.  `disklabel sd1' used to return a dummy in-core label if there
  is no label on the disk, but this didn't seem reasonable so I broke it.
  There is no-longer a dummy in-core label for each slice, and you
  wouldn't want one for non-BSD slices.  There is only a dummy in-core
  label for the whole disk.  The whole disk is the device /dev/sd1, which
  is a completely different device from sd1 and /dev/sd1c.

Yep, makes perfect sense.
  
  >sd1s1: start 32, end = 819199, size 819168: OK
  >sd1s2: start 819200, end = 2041855, size 1222656: OK
  >sd1s5: start 819232, end = 2041855, size 1222624: OK
  
  >  > 2) All of this slice stuff is really verbose and I think gratuitous.
  >  >    Note: I am using older boot blocks that don't support -v,  so it
  
  More is apparently required for debugging old broken installations :-(.

Actually, I just think judicious use of bootverbose would help here.

As far as documentation,  all I can say is document the shit out of this
in the 2.1 release notes in the area "READ AT LEAST THIS BEFORE YOU START" :-)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503290750.XAA00464>