On Thu, Oct 17, 2019 at 08:55:12AM -0400, Brian Foster wrote:
> On Wed, Oct 09, 2019 at 09:48:47AM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> >
> > When we're discarding old btree blocks after a repair, only invalidate
> > the buffers for the ones that we're freeing -- if the metadata was
> > crosslinked with another data structure, we don't want to touch it.
> >
> > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> > ---
> > fs/xfs/scrub/repair.c | 76 +++++++++++++++++++++++--------------------------
> > fs/xfs/scrub/repair.h | 1 -
> > 2 files changed, 35 insertions(+), 42 deletions(-)
> >
> >
> > diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
> > index 3a58788e0bd8..e21faef6db5a 100644
> > --- a/fs/xfs/scrub/repair.c
> > +++ b/fs/xfs/scrub/repair.c
> ...
> > @@ -515,6 +477,35 @@ xrep_put_freelist(
> > return 0;
> > }
> >
> > +/* Try to invalidate the incore buffer for a block that we're about to free. */
> > +STATIC void
> > +xrep_reap_invalidate_block(
> > + struct xfs_scrub *sc,
> > + xfs_fsblock_t fsbno)
> > +{
> > + struct xfs_buf *bp;
> > +
> > + /*
> > + * For each block in each extent, see if there's an incore buffer for
> > + * exactly that block; if so, invalidate it. The buffer cache only
> > + * lets us look for one buffer at a time, so we have to look one block
> > + * at a time. Avoid invalidating AG headers and post-EOFS blocks
> > + * because we never own those; and if we can't TRYLOCK the buffer we
> > + * assume it's owned by someone else.
> > + */
> > + /* Skip AG headers and post-EOFS blocks */
>
> The comment doesn't seem to quite go with the implementation any longer.
> Also, there's probably no need to have two separate comments here.
> Otherwise looks Ok.
<nod> Fixed.
--D
> Brian
>
> > + if (!xfs_verify_fsbno(sc->mp, fsbno))
> > + return;
> > + bp = xfs_buf_incore(sc->mp->m_ddev_targp,
> > + XFS_FSB_TO_DADDR(sc->mp, fsbno),
> > + XFS_FSB_TO_BB(sc->mp, 1), XBF_TRYLOCK);
> > + if (!bp)
> > + return;
> > +
> > + xfs_trans_bjoin(sc->tp, bp);
> > + xfs_trans_binval(sc->tp, bp);
> > +}
> > +
> > /* Dispose of a single block. */
> > STATIC int
> > xrep_reap_block(
> > @@ -568,12 +559,15 @@ xrep_reap_block(
> > * blow on writeout, the filesystem will shut down, and the admin gets
> > * to run xfs_repair.
> > */
> > - if (has_other_rmap)
> > + if (has_other_rmap) {
> > error = xfs_rmap_free(sc->tp, agf_bp, agno, agbno, 1, oinfo);
> > - else if (resv == XFS_AG_RESV_AGFL)
> > + } else if (resv == XFS_AG_RESV_AGFL) {
> > + xrep_reap_invalidate_block(sc, fsbno);
> > error = xrep_put_freelist(sc, agbno);
> > - else
> > + } else {
> > + xrep_reap_invalidate_block(sc, fsbno);
> > error = xfs_free_extent(sc->tp, fsbno, 1, oinfo, resv);
> > + }
> > if (agf_bp != sc->sa.agf_bp)
> > xfs_trans_brelse(sc->tp, agf_bp);
> > if (error)
> > diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
> > index 60c61d7052a8..eab41928990f 100644
> > --- a/fs/xfs/scrub/repair.h
> > +++ b/fs/xfs/scrub/repair.h
> > @@ -31,7 +31,6 @@ int xrep_init_btblock(struct xfs_scrub *sc, xfs_fsblock_t fsb,
> > struct xfs_bitmap;
> >
> > int xrep_fix_freelist(struct xfs_scrub *sc, bool can_shrink);
> > -int xrep_invalidate_blocks(struct xfs_scrub *sc, struct xfs_bitmap *btlist);
> > int xrep_reap_extents(struct xfs_scrub *sc, struct xfs_bitmap *exlist,
> > const struct xfs_owner_info *oinfo, enum xfs_ag_resv_type type);
> >
> >
