Re: [bug] Can't resize a hurd owned ext2 partition

[Andrew Clausen]

On Tue, Aug 07, 2001 at 11:16:02PM -0400, Roland McGrath wrote:
> > On Mon, Aug 06, 2001 at 11:34:06PM +0200, Jeroen Dekkers wrote:
> > > If I resize a hurd-owned ext2 partition I get this error message:
> > > 
> > > block 295512 has no reference? Weird
> > > 
> > > When I looked it up with debugfs I found out that on that block a passive
> > > translator is stored and parted doesn't know what to handle it.
> > 
> > Ah, right.  It doesn't know how to handle it.  Nor do I.  Any docs?
> 
> I don't know of any documentation of the ext2fs format that explains this.
> (But then I don't personally know of any actual documentation of the ext2fs
> format at all.)

There is some really bad docs in the kernel hacker's guide.  I should
write some...

> But it is pretty easy to see what to do from how e2fsck
> handles it.

In this case, yes.

> If the superblock s_creator_os field is set to EXT2_OS_HURD (1), then the
> `reserved1' word in inodes is in fact the i_translator word, which is
> either zero or the block number containing the passive translator setting.
> If you use debugfs to display an inode on an ext2fs with creator_os set to
> hurd, it displays the translator block number clearly.

Thanks.

> This all seems like it should be pretty obvious from looking at e2fsprogs,
> even for someone who knows nothing at all about the Hurd.

It is... except one thing: does the translator block reference any
other inodes or blocks?

Andrew