[Ocfs2-tools-devel] [PATCH 4/9] libocfs2: Set and clear journal features.

[Mark Fasheh]

On Tue, Oct 14, 2008 at 06:19:43PM -0700, Joel Becker wrote:
> On Tue, Oct 14, 2008 at 06:06:39PM -0700, Mark Fasheh wrote:
> > On Tue, Sep 09, 2008 at 05:57:11PM -0700, Joel Becker wrote:
> > > This adds functions to set and clear feature bits in the journal
> > > superblock.  Features are checked against the known feature set.
> > > 
> > > Signed-off-by: Joel Becker <joel.becker at oracle.com>
> > > ---
> > >  include/ocfs2/ocfs2.h |    4 +++
> > >  libocfs2/mkjournal.c  |   55 +++++++++++++++++++++++++++++++++++++++++++++++++
> > >  2 files changed, 59 insertions(+), 0 deletions(-)
> > > 
> > > diff --git a/include/ocfs2/ocfs2.h b/include/ocfs2/ocfs2.h
> > > index bdeadb9..8a7e865 100644
> > > --- a/include/ocfs2/ocfs2.h
> > > +++ b/include/ocfs2/ocfs2.h
> > > @@ -289,6 +289,10 @@ errcode_t ocfs2_write_journal_superblock(ocfs2_filesys *fs, uint64_t blkno,
> > >  					 char *jsb_buf);
> > >  errcode_t ocfs2_make_journal(ocfs2_filesys *fs, uint64_t blkno,
> > >  			     uint32_t clusters);
> > > +errcode_t ocfs2_journal_clear_features(journal_superblock_t *jsb,
> > > +				       ocfs2_fs_options *features);
> > > +errcode_t ocfs2_journal_set_features(journal_superblock_t *jsb,
> > > +				     ocfs2_fs_options *features);
> > >  extern size_t ocfs2_journal_tag_bytes(journal_superblock_t *jsb);
> > >  extern uint64_t ocfs2_journal_tag_block(journal_block_tag_t *tag,
> > >  					size_t tag_bytes);
> > > diff --git a/libocfs2/mkjournal.c b/libocfs2/mkjournal.c
> > > index e14a34d..35835f1 100644
> > > --- a/libocfs2/mkjournal.c
> > > +++ b/libocfs2/mkjournal.c
> > > @@ -50,6 +50,61 @@ uint64_t ocfs2_journal_tag_block(journal_block_tag_t *tag, size_t tag_bytes)
> > >  	return blockno;
> > >  }
> > >  
> > > +/* Returns true if we support these journal features */
> > > +static int ocfs2_journal_check_available_features(journal_superblock_t *jsb,
> > > +						  ocfs2_fs_options *features)
> > > +{
> > > +	if (!features->opt_compat && !features->opt_ro_compat &&
> > > +	    !features->opt_incompat)
> > > +		return 1;
> > > +
> > > +	if (jsb->s_header.h_blocktype != JBD2_SUPERBLOCK_V2)
> > > +		return 0;
> > 
> > Shouldn't this also check for JBD2_SUPERBLOCK_V1 somehow?
> > 	--Mark
> 
> 	It does by returning false when anything but V2.  Basically,
> those lines of context you included are:
> 
> - Can we support the journal features in *features?
>   - if you aren't asking for any features at all, return "yes"
>   - if the header is not V2, return "no"
> 
> So, if you ask for anything but the default feature set, you're required
> to have a V2 superblock.  I think that's safe.

So, we won't ever get into a situation where there's no features set, but
the superblock is junk?
	--Mark

--
Mark Fasheh