Lustre devel - Mar 2007 - [Bug 11981] Support new ldiskfs2

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11981



9856, 9857, 9859 landed on b1_4.  I will now work on a version of this for HEAD.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11981



(In reply to comment #33)
> This looks good, but I wonder if the "premount" path can be
reduced -- is
> ldiskfs2 a superset of ldiskfs?  e.g. Can ldiskfs2 read all
ldiskfs-formatted
> disks?	If so (and if the ldiskfs2 module is always available when ldiskfs
is),
> then we should only try ldiskfs2 for the premount under 2.6.
I''m concerned that ldiskfs2 may (at some point) automatically enable
features at
mount time that prevent it from being used by ldiskfs on subsequent mounts.  So,
to be safe, I chose to make it premount using ldiskfs first.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11981



(In reply to comment #34)
> In fact, the value in the GOTO doesn''t affect what is returned by
the fn.  If
> you wanted rc2, you''d have to say GOTO(out_free, rc=rc2).  Just an
FYI, since I
> think it should be GOTO(out_free, rc) anyhow.
It''s GOTO(out_free, rc) in my copy now due to Andreas''s
suggestion.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11981



(In reply to comment #36)
> +fsfilt_ldiskfs2-objs := fsfilt-ldiskfs2.o
> does this mean ldiskfs2 gets build under 2.4 also?
No, that line will be ignored since we do not add the fsfilt_ldiskfs2 module to
MODULES= unless LDISKFS is enabled.
> And should this be "if LDISKFS2" instead of "if
LDISKFS" just for consistency?
> +if LDISKFS
> +modulefs_DATA += fsfilt_ldiskfs2$(KMODEXT)
No, because then we''d have to add ldiskfs2 support to ./configure, etc.
Right
now I just assume ldiskfs2 is enabled if ldiskfs is enabled.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11981



The following patches have been committed to HEAD:

HEAD: move quota support from ldiskfs to lvfs.
HEAD: add support for the ldiskfs2 fstype (v2)
HEAD: format ldiskfs2 backing filesystems with uninit_groups

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11981

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


"HEAD: format ldiskfs2 backing filesystems with uninit_groups" landed
on HEAD.

Resolving.