xfs
[Top] [All Lists]

Re: [PATCH 04/13] xfs_repair: use multibuffer read routines in attr_repa

To: Eric Sandeen <sandeen@xxxxxxxxxxx>
Subject: Re: [PATCH 04/13] xfs_repair: use multibuffer read routines in attr_repair.c
From: Brian Foster <bfoster@xxxxxxxxxx>
Date: Mon, 14 Sep 2015 15:30:23 -0400
Cc: xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <55F71F05.90905@xxxxxxxxxxx>
References: <1441827251-13128-1-git-send-email-sandeen@xxxxxxxxxxx> <1441827251-13128-5-git-send-email-sandeen@xxxxxxxxxxx> <20150914191816.GD34083@xxxxxxxxxxxxxxx> <55F71F05.90905@xxxxxxxxxxx>
User-agent: Mutt/1.5.23 (2014-03-12)
On Mon, Sep 14, 2015 at 02:24:53PM -0500, Eric Sandeen wrote:
> 
> 
> On 9/14/15 2:18 PM, Brian Foster wrote:
> > On Wed, Sep 09, 2015 at 02:34:02PM -0500, Eric Sandeen wrote:
> >> verify_da_path and traverse_int_dablock are similar to
> >> verify_dir2_path and traverse_int_dir2block, but one
> >> difference is that the dir2 code reads using the
> >> multibuffer capable da_read_buf() routine, whereas
> >> the attr code doesn't need to, and just calls
> >> libxfs_readbuf.
> >>
> >> The multibuffer code falls back just fine when the
> >> geometry indicates that it's not needed, so use that
> >> same code in the attribute routines, and remove
> >> another dir2 / da difference.  We make da_read_buf()
> >> non-static to facilitate this.
> >>
> >> Finally, add a local *geo to these routines,
> >> to make the code even more similar at this point.
> >> The geometry will get passed in later in the series.
> >>
> >> Signed-off-by: Eric Sandeen <sandeen@xxxxxxxxxx>
> >> Signed-off-by: Eric Sandeen <sandeen@xxxxxxxxxxx>
> >> ---
> >>  repair/attr_repair.c |   49 
> >> +++++++++++++++++++++++++++++++++----------------
> >>  repair/dir2.c        |   22 +++++++++++++---------
> >>  repair/dir2.h        |    8 ++++++++
> >>  3 files changed, 54 insertions(+), 25 deletions(-)
> >>
> >> diff --git a/repair/attr_repair.c b/repair/attr_repair.c
> >> index aba0782..fe81df4 100644
> >> --- a/repair/attr_repair.c
> >> +++ b/repair/attr_repair.c
> >> @@ -138,14 +138,20 @@ traverse_int_dablock(xfs_mount_t     *mp,
> >>            xfs_dablk_t             *rbno,
> >>            int                     whichfork)
> >>  {
> >> +  bmap_ext_t              *bmp;
> 
> Well, yeah - I thought about that.  The goal was to make it as close
> as possible to the other code, so I decided against the struct foo's.
> 
> I could do a "get rid of typedefs" as the last patch to the new util
> code if you like.
> 

Ah, right. Don't worry about this then.

> > struct bmap_ext
> > 
> > (and several other places below)
> > 
> >>    xfs_dablk_t             bno;
> >>    int                     i;
> >> +  int                     nex;
> >>    xfs_da_intnode_t        *node;
> >> +  bmap_ext_t              lbmp;
> >>    xfs_fsblock_t           fsbno;
> >>    xfs_buf_t               *bp;
> >> +  struct xfs_da_geometry  *geo;
> >>    struct xfs_da_node_entry *btree;
> >>    struct xfs_da3_icnode_hdr nodehdr;
> >>  
> >> +  geo = mp->m_attr_geo;
> >> +
> >>    /*
> >>     * traverse down left-side of tree until we hit the
> >>     * left-most leaf block setting up the btree cursor along
> >> @@ -160,13 +166,16 @@ traverse_int_dablock(xfs_mount_t     *mp,
> >>            /*
> >>             * read in each block along the way and set up cursor
> >>             */
> >> -          fsbno = blkmap_get(da_cursor->blkmap, bno);
> >> +          nex = blkmap_getn(da_cursor->blkmap, bno,
> >> +                          geo->fsbcount, &bmp, &lbmp);
> >>  
> > 
> > ...
> > attr_repair.c: In function âprocess_attributesâ:
> > attr_repair.c:185:5: warning: âfsbnoâ may be used uninitialized in this 
> > function [-Wmaybe-uninitialized]
> 
> hm, whoops.  Well, it goes away in the long run, but I can still fix
> that up.
> 

I figured it might (haven't got that far yet), but I'd still rather not
have regressions in the history if we can help it, even if transient.

Brian

> Thanks,
> -Eric
> 
> >      do_warn(
> > ...
> > 
> > Brian
> > 
> >> -          if (fsbno == NULLFSBLOCK)
> >> +          if (nex == 0)
> >>                    goto error_out;
> >>  
> >> -          bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
> >> -                          XFS_FSB_TO_BB(mp, 1), 0, &xfs_da3_node_buf_ops);
> >> +          bp = da_read_buf(mp, nex, bmp, &xfs_da3_node_buf_ops);
> >> +          if (bmp != &lbmp)
> >> +                  free(bmp);
> >> +
> >>            if (!bp) {
> >>                    if (whichfork == XFS_DATA_FORK)
> >>                            do_warn(
> >> @@ -192,12 +201,10 @@ traverse_int_dablock(xfs_mount_t     *mp,
> >>                    goto error_out;
> >>            }
> >>  
> >> -          if (nodehdr.count > mp->m_attr_geo->node_ents)  {
> >> +          if (nodehdr.count > geo->node_ents)  {
> >>                    do_warn(_("bad record count in inode %" PRIu64 ", "
> >>                              "count = %d, max = %d\n"),
> >> -                          da_cursor->ino,
> >> -                          nodehdr.count,
> >> -                          mp->m_attr_geo->node_ents);
> >> +                          da_cursor->ino, nodehdr.count, geo->node_ents);
> >>                    libxfs_putbuf(bp);
> >>                    goto error_out;
> >>            }
> >> @@ -492,9 +499,15 @@ verify_da_path(xfs_mount_t    *mp,
> >>    int                     bad;
> >>    int                     entry;
> >>    int                     this_level = p_level + 1;
> >> +  bmap_ext_t              *bmp;
> >> +  int                     nex;
> >> +  bmap_ext_t              lbmp;
> >> +  struct xfs_da_geometry  *geo;
> >>    struct xfs_da_node_entry *btree;
> >>    struct xfs_da3_icnode_hdr nodehdr;
> >>  
> >> +  geo = mp->m_attr_geo;
> >> +
> >>    /*
> >>     * index is currently set to point to the entry that
> >>     * should be processed now in this level.
> >> @@ -536,17 +549,21 @@ verify_da_path(xfs_mount_t   *mp,
> >>             */
> >>            dabno = nodehdr.forw;
> >>            ASSERT(dabno != 0);
> >> -          fsbno = blkmap_get(cursor->blkmap, dabno);
> >> -
> >> -          if (fsbno == NULLFSBLOCK) {
> >> -                  do_warn(_("can't get map info for block %u "
> >> -                            "of directory inode %" PRIu64 "\n"),
> >> +          nex = blkmap_getn(cursor->blkmap, dabno, geo->fsbcount,
> >> +                  &bmp, &lbmp);
> >> +          if (nex == 0) {
> >> +                  do_warn(
> >> +_("can't get map info for block %u of directory inode %" PRIu64 "\n"),
> >>                            dabno, cursor->ino);
> >>                    return(1);
> >>            }
> >>  
> >> -          bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
> >> -                          XFS_FSB_TO_BB(mp, 1), 0, &xfs_da3_node_buf_ops);
> >> +          fsbno = bmp[0].startblock;
> >> +
> >> +          bp = da_read_buf(mp, nex, bmp, &xfs_da3_node_buf_ops);
> >> +          if (bmp != &lbmp)
> >> +                  free(bmp);
> >> +
> >>            if (!bp) {
> >>                    do_warn(
> >>    _("can't read block %u (%" PRIu64 ") for directory inode %" PRIu64 
> >> "\n"),
> >> @@ -577,7 +594,7 @@ verify_da_path(xfs_mount_t     *mp,
> >>                            dabno, fsbno, cursor->ino);
> >>                    bad++;
> >>            }
> >> -          if (nodehdr.count > mp->m_attr_geo->node_ents) {
> >> +          if (nodehdr.count > geo->node_ents) {
> >>                    do_warn(
> >>    _("entry count %d too large in block %u (%" PRIu64 ") for directory 
> >> inode %" PRIu64 "\n"),
> >>                            nodehdr.count,
> >> diff --git a/repair/dir2.c b/repair/dir2.c
> >> index 54c49eb..44367c6 100644
> >> --- a/repair/dir2.c
> >> +++ b/repair/dir2.c
> >> @@ -92,7 +92,7 @@ namecheck(char *name, int length)
> >>   * Multibuffer handling.
> >>   * V2 directory blocks can be noncontiguous, needing multiple buffers.
> >>   */
> >> -static struct xfs_buf *
> >> +struct xfs_buf *
> >>  da_read_buf(
> >>    xfs_mount_t     *mp,
> >>    int             nex,
> >> @@ -143,9 +143,12 @@ traverse_int_dir2block(xfs_mount_t    *mp,
> >>    int                     nex;
> >>    xfs_da_intnode_t        *node;
> >>    bmap_ext_t              lbmp;
> >> +  struct xfs_da_geometry  *geo;
> >>    struct xfs_da_node_entry *btree;
> >>    struct xfs_da3_icnode_hdr nodehdr;
> >>  
> >> +  geo = mp->m_dir_geo;
> >> +
> >>    /*
> >>     * traverse down left-side of tree until we hit the
> >>     * left-most leaf block setting up the btree cursor along
> >> @@ -161,7 +164,7 @@ traverse_int_dir2block(xfs_mount_t     *mp,
> >>             * read in each block along the way and set up cursor
> >>             */
> >>            nex = blkmap_getn(da_cursor->blkmap, bno,
> >> -                          mp->m_dir_geo->fsbcount, &bmp, &lbmp);
> >> +                          geo->fsbcount, &bmp, &lbmp);
> >>  
> >>            if (nex == 0)
> >>                    goto error_out;
> >> @@ -207,13 +210,11 @@ _("corrupt tree block %u for directory inode %" 
> >> PRIu64 "\n"),
> >>                    goto error_out;
> >>            }
> >>            btree = M_DIROPS(mp)->node_tree_p(node);
> >> -          if (nodehdr.count > mp->m_dir_geo->node_ents)  {
> >> -                  libxfs_putbuf(bp);
> >> +          if (nodehdr.count > geo->node_ents)  {
> >>                    do_warn(
> >>  _("bad record count in inode %" PRIu64 ", count = %d, max = %d\n"),
> >> -                          da_cursor->ino,
> >> -                          nodehdr.count,
> >> -                          mp->m_dir_geo->node_ents);
> >> +                          da_cursor->ino, nodehdr.count, geo->node_ents);
> >> +                  libxfs_putbuf(bp);
> >>                    goto error_out;
> >>            }
> >>            /*
> >> @@ -488,9 +489,12 @@ verify_dir2_path(xfs_mount_t  *mp,
> >>    bmap_ext_t              *bmp;
> >>    int                     nex;
> >>    bmap_ext_t              lbmp;
> >> +  struct xfs_da_geometry  *geo;
> >>    struct xfs_da_node_entry *btree;
> >>    struct xfs_da3_icnode_hdr nodehdr;
> >>  
> >> +  geo = mp->m_dir_geo;
> >> +
> >>    /*
> >>     * index is currently set to point to the entry that
> >>     * should be processed now in this level.
> >> @@ -532,7 +536,7 @@ verify_dir2_path(xfs_mount_t   *mp,
> >>             */
> >>            dabno = nodehdr.forw;
> >>            ASSERT(dabno != 0);
> >> -          nex = blkmap_getn(cursor->blkmap, dabno, 
> >> mp->m_dir_geo->fsbcount,
> >> +          nex = blkmap_getn(cursor->blkmap, dabno, geo->fsbcount,
> >>                    &bmp, &lbmp);
> >>            if (nex == 0) {
> >>                    do_warn(
> >> @@ -574,7 +578,7 @@ _("bad back pointer in block %u for directory inode %" 
> >> PRIu64 "\n"),
> >>                            dabno, cursor->ino);
> >>                    bad++;
> >>            }
> >> -          if (nodehdr.count > mp->m_dir_geo->node_ents)  {
> >> +          if (nodehdr.count > geo->node_ents)  {
> >>                    do_warn(
> >>  _("entry count %d too large in block %u for directory inode %" PRIu64 
> >> "\n"),
> >>                            nodehdr.count,
> >> diff --git a/repair/dir2.h b/repair/dir2.h
> >> index 3cc1941..186633f 100644
> >> --- a/repair/dir2.h
> >> +++ b/repair/dir2.h
> >> @@ -58,6 +58,14 @@ typedef struct dir2_bt_cursor  {
> >>    struct blkmap           *blkmap;
> >>  } dir2_bt_cursor_t;
> >>  
> >> +#include "bmap.h" /* Goes away in later refactoring */
> >> +struct xfs_buf *
> >> +da_read_buf(
> >> +  xfs_mount_t     *mp,
> >> +  int             nex,
> >> +  bmap_ext_t      *bmp,
> >> +  const struct xfs_buf_ops *ops);
> >> +
> >>  int
> >>  process_dir2(
> >>    xfs_mount_t     *mp,
> >> -- 
> >> 1.7.1
> >>
> >> _______________________________________________
> >> xfs mailing list
> >> xfs@xxxxxxxxxxx
> >> http://oss.sgi.com/mailman/listinfo/xfs
> > 
> 
> _______________________________________________
> xfs mailing list
> xfs@xxxxxxxxxxx
> http://oss.sgi.com/mailman/listinfo/xfs

<Prev in Thread] Current Thread [Next in Thread>