Colin Wood wrote:
> Ken Nakata wrote:
> > Colin Wood wrote:
> > > I think that Hauke was right and the NetBSD/mac68k kernel doesn't
> > > recognize ext2fs filesystems as he states below:
> > >  
> > > > > The MacBSD kernel evaluates the native Macintosh partition table and fakes
> > > > > a NetBSD disklabel from it. There is definitely no code in there to label a
> > > > > partition as "ext2fs" or whatever, so if mount_ext2fs(8) looks at the
> > > > > partition type, it sees "4.2BSD" and you lose.
> > > > >
> > > > > Yet another reason to implement native disklabels...
> > 
> > I think this is somehow irrelevant as to why you cannot mount_ext2fs a
> > partition formatted by MacLinux.  At least, mount_* commands doesn't
> > seem to care what type the partition is labeled, nor does mount(2).
> 
> Really?  That seems a bit strange.....

I'm not sure if this is enough to convince you, but I could
successfully newfs and newfs_msdos an "unknown" partition.  This
partition of type "unknown" is actually a free area, and not marked
either as 4.2BSD or as MSDOS, yet I can mount it (FFS or MSDOS).
Unfortunately, there doesn't exist newfs_ext2fs on NetBSD, so my
experiment was limited... (I doubt MacLinux runs usefully on my boxes
just yet).

> According to a post from Manuel Bouyer to current-users, I think, the
> bi-endian ffs stuff was in the 1.3 branch, so it should be in 1.3.2
> (unless I was reading the posting wrong).

As Bill correctly pointed out, if the filesystem in question is in
fact big endian, it is impossible to mount the partition without
*modifying* NetBSD's ext2fs implementation (i.e. recompiling the
kernel).

Later,

Ken