On Aug 14, 12:42pm, thorpej@shagadelic.org (Jason Thorpe) wrote:
-- Subject: Re: CVS commit: src/sys/arch/x86/x86

| On Aug 14, 2006, at 12:40 PM, Christos Zoulas wrote:
| 
| > Have you tried booting from a disk with a regular disklabel on and
| > a kernel that has DKWEDGE_* turned on? What happens is that there
| > is only the non-wedge bootinfo information passed from the boot
| > blocks.  After that, the wedge code discovers all the wedges and
| > generates dkN from wd?x. Then the autoconfig mechanism tries to
| > open partitions wd?x from the old disklabel and it fails with EBUSY,
| > since that disk has wedges. How did you test?
| 
| How old are your boot blocks?

NetBSD/i386 BIOS Boot
Mon Dec 19 00:29:09 UTC 2005

Unfortunately I cannot get the version from strings...

| (Yes, I was doing specifically what you're talking about on my test  
| machine when I had it working...)

Well, now it works fine with old bootblocks too, so this can be used
as compat code.

Now that I have your ear:

1. How did you plan to deal with bsize,fsize,cpg info that used to
   be stored in the disklabel in the wedge world?
2. Do you have newfs/fsck/dump working with wedges?
3. What's the plan to convert existing disk drivers to use wedges,
   and moving the bulk of the old partitioning code to one place?
4. How is userland supposed to get geometry information now?
5. Are we planning to be able to mount by wedge name?
6. Are we going to have GPT NetBSD GUIDs? What is the list of GUIDs
   that we would need? How to generate them?

christos