Re: [PATCH][10/28] e2fsprogs-uninit.patch

[Andreas Dilger <adilger@sun.com>]

On Mar 15, 2008  15:41 -0400, Theodore Ts'o wrote:
> What is the intended use of the SF_DO_CSUM flag?  I see where it is
> defined, and where it gets sets, but as far as I can tell nothing
> actually tests for it or uses it.

Probably a hold-over from a previous version of the code.  It sets the
SF_DO_LAZY flag if GDT_CSUM is set, causing ext2fs_get_next_inode_full()
to skip the BG_INODE_UNINT groups.  It's up to you if it would be better
to keep SF_DO_CSUM and check for it explicitly (possibly using it for
something else later), or to overload SF_DO_LAZY as we do currently.

> On Sat, Feb 02, 2008 at 01:34:44AM -0700, Andreas Dilger wrote:
> > Index: e2fsprogs-1.40.5/lib/ext2fs/ext2fs.h
> >
> > +#define EXT2_SF_DO_CSUM		0x0020
> >
> 
> > Index: e2fsprogs-1.40.5/lib/ext2fs/inode.c
> > ===================================================================
> > --- e2fsprogs-1.40.5.orig/lib/ext2fs/inode.c
> > +++ e2fsprogs-1.40.5/lib/ext2fs/inode.c
> > @@ -167,6 +167,9 @@ errcode_t ext2fs_open_inode_scan(ext2_fi
> >  	if (EXT2_HAS_COMPAT_FEATURE(fs->super, 
> >  				    EXT2_FEATURE_COMPAT_LAZY_BG))
> >  		scan->scan_flags |= EXT2_SF_DO_LAZY;
> > +	if (EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
> > +				       EXT4_FEATURE_RO_COMPAT_GDT_CSUM))
> > +		scan->scan_flags |= EXT2_SF_DO_LAZY | EXT2_SF_DO_CSUM;

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.