Re: [PATCH v2 24/27] xfsprogs: Add parent pointers to recreated inodes

From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: Allison Henderson <allison.henderson@oracle.com>
Cc: Amir Goldstein <amir73il@gmail.com>,
	linux-xfs <linux-xfs@vger.kernel.org>
Subject: Re: [PATCH v2 24/27] xfsprogs: Add parent pointers to recreated inodes
Date: Mon, 11 Jun 2018 10:31:59 -0700	[thread overview]
Message-ID: <20180611173159.GC22045@magnolia> (raw)
In-Reply-To: <cde33676-f0c7-4a67-c0b9-4e677ecf3d5e@oracle.com>

On Sun, Jun 10, 2018 at 10:19:48AM -0700, Allison Henderson wrote:
> On 06/10/2018 04:27 AM, Amir Goldstein wrote:
> > On Sun, Jun 10, 2018 at 8:07 AM, Allison Henderson
> > <allison.henderson@oracle.com> wrote:
> > > Inodes that are not recovered from the orphanage will
> > > be recreated, so we will need to re-add the parent pointers
> > > here too.
> > > 
> > > Signed-off-by: Allison Henderson <allison.henderson@oracle.com>
> > > ---
> > >   libxfs/libxfs_api_defs.h |  1 +
> > >   repair/phase6.c          | 40 +++++++++++++++++++++++++++-------------
> > >   2 files changed, 28 insertions(+), 13 deletions(-)
> > > 
> > > diff --git a/libxfs/libxfs_api_defs.h b/libxfs/libxfs_api_defs.h
> > > index 78daca0..4b47479 100644
> > > --- a/libxfs/libxfs_api_defs.h
> > > +++ b/libxfs/libxfs_api_defs.h
> > > @@ -152,5 +152,6 @@
> > >   #define xfs_rmap_compare               libxfs_rmap_compare
> > >   #define xfs_dir_get_ops                        libxfs_dir_get_ops
> > >   #define xfs_default_ifork_ops          libxfs_default_ifork_ops
> > > +#define xfs_parent_add                 libxfs_parent_add
> > > 
> > >   #endif /* __LIBXFS_API_DEFS_H__ */
> > > diff --git a/repair/phase6.c b/repair/phase6.c
> > > index 4fedb35..96ffade 100644
> > > --- a/repair/phase6.c
> > > +++ b/repair/phase6.c
> > > @@ -29,6 +29,7 @@
> > >   #include "dinode.h"
> > >   #include "progress.h"
> > >   #include "versions.h"
> > > +#include "xfs_parent.h"
> > > 
> > >   static struct cred             zerocr;
> > >   static struct fsxattr          zerofsx;
> > > @@ -962,19 +963,20 @@ mk_root_dir(xfs_mount_t *mp)
> > >   static xfs_ino_t
> > >   mk_orphanage(xfs_mount_t *mp)
> > >   {
> > > -       xfs_ino_t       ino;
> > > -       xfs_trans_t     *tp;
> > > -       xfs_inode_t     *ip;
> > > -       xfs_inode_t     *pip;
> > > -       xfs_fsblock_t   first;
> > > -       ino_tree_node_t *irec;
> > > -       int             ino_offset = 0;
> > > -       int             i;
> > > -       int             error;
> > > +       xfs_ino_t               ino;
> > > +       struct xfs_trans        *tp;
> > > +       struct xfs_inode        *ip;
> > > +       struct xfs_inode        *pip;
> > 
> > I guess you did some whitespace cleanups here while at it,
> > but why did those types change from xfs_inode_t etc?
> 
> Yes, removing the *_t types where appropriate was something that came up in
> a few of the earlier reviews.  I just assumed we are trying to clear them
> all out?  It does seem to kind of help distinguish whats a structure and
> whats a primitive data type.

Yes, the general practice (afaict) is to de-typedef-ify code when making
changes to them.

--D

> Allison
> 
> 
> > 
> > > +       xfs_fsblock_t           first;
> > > +       struct ino_tree_node    *irec;
> > > +       int                     ino_offset = 0;
> > > +       int                     i;
> > > +       int                     error;
> > >          struct xfs_defer_ops    dfops;
> > > -       const int       mode = 0755;
> > > -       int             nres;
> > > -       struct xfs_name xname;
> > > +       const int               mode = 0755;
> > > +       int                     nres;
> > > +       struct xfs_name         xname;
> > > +       xfs_dir2_dataptr_t      offset;
> > > 
> > >          /*
> > >           * check for an existing lost+found first, if it exists, return
> > > @@ -1061,7 +1063,7 @@ mk_orphanage(xfs_mount_t *mp)
> > >           * create the actual entry
> > >           */
> > >          error = -libxfs_dir_createname(tp, pip, &xname, ip->i_ino, &first,
> > > -                                       &dfops, nres, NULL);
> > > +                                       &dfops, nres, &offset);
> > >          if (error)
> > >                  do_error(
> > >                  _("can't make %s, createname error %d\n"),
> > > @@ -1083,6 +1085,18 @@ mk_orphanage(xfs_mount_t *mp)
> > >          libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
> > > 
> > >          libxfs_defer_ijoin(&dfops, ip);
> > > +
> > > +        if (xfs_sb_version_hasparent(&mp->m_sb)) {
> > > +                error = -libxfs_parent_add(pip, ip, &xname, offset,
> > > +                                      &first, &dfops);
> > > +                if (error)
> > > +                        do_error(_("Error creating parent pointer: %d\n"),
> > > +                                error);
> > > +               libxfs_trans_log_inode(tp, pip, XFS_ILOG_CORE);
> > > +               libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
> > > +               libxfs_defer_ijoin(&dfops, pip);
> > > +        }
> > > +
> > >          error = -libxfs_defer_finish(&tp, &dfops);
> > >          if (error) {
> > >                  do_error(_("%s directory creation failed -- bmapf error %d\n"),
> > > --
> > > 2.7.4
> > > 
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> > > the body of a message to majordomo@vger.kernel.org
> > > More majordomo info at  https://urldefense.proofpoint.com/v2/url?u=http-3A__vger.kernel.org_majordomo-2Dinfo.html&d=DwIBaQ&c=RoP1YumCXCgaWHvlZYR8PZh8Bv7qIrMUB65eapI_JnE&r=LHZQ8fHvy6wDKXGTWcm97burZH5sQKHRDMaY1UthQxc&m=G-6Ft7goE2Cf-mpLlEpFZl98e-5uBpekKLRKZzIJTXE&s=SbB24E7j4VibOio0puyqOZPpxpfyIMm0xJ2wF-rltDA&e=
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html