Re: [PATCH 13/42] xfs: perags need atomic operational state

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 2023-01-19 at 09:44 +1100, Dave Chinner wrote:
> From: Dave Chinner <dchinner@xxxxxxxxxx>
> 
> We currently don't have any flags or operational state in the
> xfs_perag except for the pagf_init and pagi_init flags. And the
> agflreset flag. Oh, there's also the pagf_metadata and pagi_inodeok
> flags, too.
> 
> For controlling per-ag operations, we are going to need some atomic
> state flags. Hence add an opstate field similar to what we already
> have in the mount and log, and convert all these state flags across
> to atomic bit operations.
> 
> Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
Seems like a reasonable conversion
Reviewed-by: Allison Henderson <allison.henderson@xxxxxxxxxx>

> ---
>  fs/xfs/libxfs/xfs_ag.h             | 27 ++++++++++++++----
>  fs/xfs/libxfs/xfs_alloc.c          | 23 ++++++++-------
>  fs/xfs/libxfs/xfs_alloc_btree.c    |  2 +-
>  fs/xfs/libxfs/xfs_bmap.c           |  2 +-
>  fs/xfs/libxfs/xfs_ialloc.c         | 14 ++++-----
>  fs/xfs/libxfs/xfs_ialloc_btree.c   |  4 +--
>  fs/xfs/libxfs/xfs_refcount_btree.c |  2 +-
>  fs/xfs/libxfs/xfs_rmap_btree.c     |  2 +-
>  fs/xfs/scrub/agheader_repair.c     | 28 +++++++++---------
>  fs/xfs/scrub/fscounters.c          |  9 ++++--
>  fs/xfs/scrub/repair.c              |  2 +-
>  fs/xfs/xfs_filestream.c            |  5 ++--
>  fs/xfs/xfs_super.c                 | 46 ++++++++++++++++++----------
> --
>  13 files changed, 101 insertions(+), 65 deletions(-)
> 
> diff --git a/fs/xfs/libxfs/xfs_ag.h b/fs/xfs/libxfs/xfs_ag.h
> index aeb21c8df201..187d30d9bb13 100644
> --- a/fs/xfs/libxfs/xfs_ag.h
> +++ b/fs/xfs/libxfs/xfs_ag.h
> @@ -35,13 +35,9 @@ struct xfs_perag {
>         atomic_t        pag_ref;        /* passive reference count */
>         atomic_t        pag_active_ref; /* active reference count */
>         wait_queue_head_t pag_active_wq;/* woken active_ref falls to
> zero */
> -       char            pagf_init;      /* this agf's entry is
> initialized */
> -       char            pagi_init;      /* this agi's entry is
> initialized */
> -       char            pagf_metadata;  /* the agf is preferred to be
> metadata */
> -       char            pagi_inodeok;   /* The agi is ok for inodes
> */
> +       unsigned long   pag_opstate;
>         uint8_t         pagf_levels[XFS_BTNUM_AGF];
>                                         /* # of levels in bno & cnt
> btree */
> -       bool            pagf_agflreset; /* agfl requires reset before
> use */
>         uint32_t        pagf_flcount;   /* count of blocks in
> freelist */
>         xfs_extlen_t    pagf_freeblks;  /* total free blocks */
>         xfs_extlen_t    pagf_longest;   /* longest free space */
> @@ -108,6 +104,27 @@ struct xfs_perag {
>  #endif /* __KERNEL__ */
>  };
>  
> +/*
> + * Per-AG operational state. These are atomic flag bits.
> + */
> +#define XFS_AGSTATE_AGF_INIT           0
> +#define XFS_AGSTATE_AGI_INIT           1
> +#define XFS_AGSTATE_PREFERS_METADATA   2
> +#define XFS_AGSTATE_ALLOWS_INODES      3
> +#define XFS_AGSTATE_AGFL_NEEDS_RESET   4
> +
> +#define __XFS_AG_OPSTATE(name, NAME) \
> +static inline bool xfs_perag_ ## name (struct xfs_perag *pag) \
> +{ \
> +       return test_bit(XFS_AGSTATE_ ## NAME, &pag->pag_opstate); \
> +}
> +
> +__XFS_AG_OPSTATE(initialised_agf, AGF_INIT)
> +__XFS_AG_OPSTATE(initialised_agi, AGI_INIT)
> +__XFS_AG_OPSTATE(prefers_metadata, PREFERS_METADATA)
> +__XFS_AG_OPSTATE(allows_inodes, ALLOWS_INODES)
> +__XFS_AG_OPSTATE(agfl_needs_reset, AGFL_NEEDS_RESET)
> +
>  int xfs_initialize_perag(struct xfs_mount *mp, xfs_agnumber_t
> agcount,
>                         xfs_rfsblock_t dcount, xfs_agnumber_t
> *maxagi);
>  int xfs_initialize_perag_data(struct xfs_mount *mp, xfs_agnumber_t
> agno);
> diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
> index 9f26a9368eeb..246c2e7d9e7a 100644
> --- a/fs/xfs/libxfs/xfs_alloc.c
> +++ b/fs/xfs/libxfs/xfs_alloc.c
> @@ -2435,7 +2435,7 @@ xfs_agfl_reset(
>         struct xfs_mount        *mp = tp->t_mountp;
>         struct xfs_agf          *agf = agbp->b_addr;
>  
> -       ASSERT(pag->pagf_agflreset);
> +       ASSERT(xfs_perag_agfl_needs_reset(pag));
>         trace_xfs_agfl_reset(mp, agf, 0, _RET_IP_);
>  
>         xfs_warn(mp,
> @@ -2450,7 +2450,7 @@ xfs_agfl_reset(
>                                     XFS_AGF_FLCOUNT);
>  
>         pag->pagf_flcount = 0;
> -       pag->pagf_agflreset = false;
> +       clear_bit(XFS_AGSTATE_AGFL_NEEDS_RESET, &pag->pag_opstate);
>  }
>  
>  /*
> @@ -2605,7 +2605,7 @@ xfs_alloc_fix_freelist(
>         /* deferred ops (AGFL block frees) require permanent
> transactions */
>         ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
>  
> -       if (!pag->pagf_init) {
> +       if (!xfs_perag_initialised_agf(pag)) {
>                 error = xfs_alloc_read_agf(pag, tp, flags, &agbp);
>                 if (error) {
>                         /* Couldn't lock the AGF so skip this AG. */
> @@ -2620,7 +2620,8 @@ xfs_alloc_fix_freelist(
>          * somewhere else if we are not being asked to try harder at
> this
>          * point
>          */
> -       if (pag->pagf_metadata && (args->datatype &
> XFS_ALLOC_USERDATA) &&
> +       if (xfs_perag_prefers_metadata(pag) &&
> +           (args->datatype & XFS_ALLOC_USERDATA) &&
>             (flags & XFS_ALLOC_FLAG_TRYLOCK)) {
>                 ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
>                 goto out_agbp_relse;
> @@ -2646,7 +2647,7 @@ xfs_alloc_fix_freelist(
>         }
>  
>         /* reset a padding mismatched agfl before final free space
> check */
> -       if (pag->pagf_agflreset)
> +       if (xfs_perag_agfl_needs_reset(pag))
>                 xfs_agfl_reset(tp, agbp, pag);
>  
>         /* If there isn't enough total space or single-extent, reject
> it. */
> @@ -2803,7 +2804,7 @@ xfs_alloc_get_freelist(
>         if (be32_to_cpu(agf->agf_flfirst) == xfs_agfl_size(mp))
>                 agf->agf_flfirst = 0;
>  
> -       ASSERT(!pag->pagf_agflreset);
> +       ASSERT(!xfs_perag_agfl_needs_reset(pag));
>         be32_add_cpu(&agf->agf_flcount, -1);
>         pag->pagf_flcount--;
>  
> @@ -2892,7 +2893,7 @@ xfs_alloc_put_freelist(
>         if (be32_to_cpu(agf->agf_fllast) == xfs_agfl_size(mp))
>                 agf->agf_fllast = 0;
>  
> -       ASSERT(!pag->pagf_agflreset);
> +       ASSERT(!xfs_perag_agfl_needs_reset(pag));
>         be32_add_cpu(&agf->agf_flcount, 1);
>         pag->pagf_flcount++;
>  
> @@ -3099,7 +3100,7 @@ xfs_alloc_read_agf(
>                 return error;
>  
>         agf = agfbp->b_addr;
> -       if (!pag->pagf_init) {
> +       if (!xfs_perag_initialised_agf(pag)) {
>                 pag->pagf_freeblks = be32_to_cpu(agf->agf_freeblks);
>                 pag->pagf_btreeblks = be32_to_cpu(agf-
> >agf_btreeblks);
>                 pag->pagf_flcount = be32_to_cpu(agf->agf_flcount);
> @@ -3111,8 +3112,8 @@ xfs_alloc_read_agf(
>                 pag->pagf_levels[XFS_BTNUM_RMAPi] =
>                         be32_to_cpu(agf-
> >agf_levels[XFS_BTNUM_RMAPi]);
>                 pag->pagf_refcount_level = be32_to_cpu(agf-
> >agf_refcount_level);
> -               pag->pagf_init = 1;
> -               pag->pagf_agflreset = xfs_agfl_needs_reset(pag-
> >pag_mount, agf);
> +               if (xfs_agfl_needs_reset(pag->pag_mount, agf))
> +                       set_bit(XFS_AGSTATE_AGFL_NEEDS_RESET, &pag-
> >pag_opstate);
>  
>                 /*
>                  * Update the in-core allocbt counter. Filter out the
> rmapbt
> @@ -3127,6 +3128,8 @@ xfs_alloc_read_agf(
>                 if (allocbt_blks > 0)
>                         atomic64_add(allocbt_blks,
>                                         &pag->pag_mount-
> >m_allocbt_blks);
> +
> +               set_bit(XFS_AGSTATE_AGF_INIT, &pag->pag_opstate);
>         }
>  #ifdef DEBUG
>         else if (!xfs_is_shutdown(pag->pag_mount)) {
> diff --git a/fs/xfs/libxfs/xfs_alloc_btree.c
> b/fs/xfs/libxfs/xfs_alloc_btree.c
> index 549a3cba0234..0f29c7b1b39f 100644
> --- a/fs/xfs/libxfs/xfs_alloc_btree.c
> +++ b/fs/xfs/libxfs/xfs_alloc_btree.c
> @@ -315,7 +315,7 @@ xfs_allocbt_verify(
>         level = be16_to_cpu(block->bb_level);
>         if (bp->b_ops->magic[0] == cpu_to_be32(XFS_ABTC_MAGIC))
>                 btnum = XFS_BTNUM_CNTi;
> -       if (pag && pag->pagf_init) {
> +       if (pag && xfs_perag_initialised_agf(pag)) {
>                 if (level >= pag->pagf_levels[btnum])
>                         return __this_address;
>         } else if (level >= mp->m_alloc_maxlevels)
> diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> index f15d45af661f..6aad0ea5e606 100644
> --- a/fs/xfs/libxfs/xfs_bmap.c
> +++ b/fs/xfs/libxfs/xfs_bmap.c
> @@ -3147,7 +3147,7 @@ xfs_bmap_longest_free_extent(
>         int                     error = 0;
>  
>         pag = xfs_perag_get(mp, ag);
> -       if (!pag->pagf_init) {
> +       if (!xfs_perag_initialised_agf(pag)) {
>                 error = xfs_alloc_read_agf(pag, tp,
> XFS_ALLOC_FLAG_TRYLOCK,
>                                 NULL);
>                 if (error) {
> diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
> index c8d837d8876f..2a323ffa5ba9 100644
> --- a/fs/xfs/libxfs/xfs_ialloc.c
> +++ b/fs/xfs/libxfs/xfs_ialloc.c
> @@ -998,8 +998,8 @@ xfs_dialloc_ag_inobt(
>         int                     i, j;
>         int                     searchdistance = 10;
>  
> -       ASSERT(pag->pagi_init);
> -       ASSERT(pag->pagi_inodeok);
> +       ASSERT(xfs_perag_initialised_agi(pag));
> +       ASSERT(xfs_perag_allows_inodes(pag));
>         ASSERT(pag->pagi_freecount > 0);
>  
>   restart_pagno:
> @@ -1592,10 +1592,10 @@ xfs_dialloc_good_ag(
>  
>         if (!pag)
>                 return false;
> -       if (!pag->pagi_inodeok)
> +       if (!xfs_perag_allows_inodes(pag))
>                 return false;
>  
> -       if (!pag->pagi_init) {
> +       if (!xfs_perag_initialised_agi(pag)) {
>                 error = xfs_ialloc_read_agi(pag, tp, NULL);
>                 if (error)
>                         return false;
> @@ -1606,7 +1606,7 @@ xfs_dialloc_good_ag(
>         if (!ok_alloc)
>                 return false;
>  
> -       if (!pag->pagf_init) {
> +       if (!xfs_perag_initialised_agf(pag)) {
>                 error = xfs_alloc_read_agf(pag, tp, flags, NULL);
>                 if (error)
>                         return false;
> @@ -2603,10 +2603,10 @@ xfs_ialloc_read_agi(
>                 return error;
>  
>         agi = agibp->b_addr;
> -       if (!pag->pagi_init) {
> +       if (!xfs_perag_initialised_agi(pag)) {
>                 pag->pagi_freecount = be32_to_cpu(agi-
> >agi_freecount);
>                 pag->pagi_count = be32_to_cpu(agi->agi_count);
> -               pag->pagi_init = 1;
> +               set_bit(XFS_AGSTATE_AGI_INIT, &pag->pag_opstate);
>         }
>  
>         /*
> diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.c
> b/fs/xfs/libxfs/xfs_ialloc_btree.c
> index d657af2ec350..3675a0d29310 100644
> --- a/fs/xfs/libxfs/xfs_ialloc_btree.c
> +++ b/fs/xfs/libxfs/xfs_ialloc_btree.c
> @@ -291,8 +291,8 @@ xfs_inobt_verify(
>          * Similarly, during log recovery we will have a perag
> structure
>          * attached, but the agi information will not yet have been
> initialised
>          * from the on disk AGI. We don't currently use any of this
> information,
> -        * but beware of the landmine (i.e. need to check pag-
> >pagi_init) if we
> -        * ever do.
> +        * but beware of the landmine (i.e. need to check
> +        * xfs_perag_initialised_agi(pag)) if we ever do.
>          */
>         if (xfs_has_crc(mp)) {
>                 fa = xfs_btree_sblock_v5hdr_verify(bp);
> diff --git a/fs/xfs/libxfs/xfs_refcount_btree.c
> b/fs/xfs/libxfs/xfs_refcount_btree.c
> index e1f789866683..d20abf0390fc 100644
> --- a/fs/xfs/libxfs/xfs_refcount_btree.c
> +++ b/fs/xfs/libxfs/xfs_refcount_btree.c
> @@ -227,7 +227,7 @@ xfs_refcountbt_verify(
>                 return fa;
>  
>         level = be16_to_cpu(block->bb_level);
> -       if (pag && pag->pagf_init) {
> +       if (pag && xfs_perag_initialised_agf(pag)) {
>                 if (level >= pag->pagf_refcount_level)
>                         return __this_address;
>         } else if (level >= mp->m_refc_maxlevels)
> diff --git a/fs/xfs/libxfs/xfs_rmap_btree.c
> b/fs/xfs/libxfs/xfs_rmap_btree.c
> index 7f83f62e51e0..d3285684bb5e 100644
> --- a/fs/xfs/libxfs/xfs_rmap_btree.c
> +++ b/fs/xfs/libxfs/xfs_rmap_btree.c
> @@ -313,7 +313,7 @@ xfs_rmapbt_verify(
>                 return fa;
>  
>         level = be16_to_cpu(block->bb_level);
> -       if (pag && pag->pagf_init) {
> +       if (pag && xfs_perag_initialised_agf(pag)) {
>                 if (level >= pag->pagf_levels[XFS_BTNUM_RMAPi])
>                         return __this_address;
>         } else if (level >= mp->m_rmap_maxlevels)
> diff --git a/fs/xfs/scrub/agheader_repair.c
> b/fs/xfs/scrub/agheader_repair.c
> index b80b9111e781..c37e6d72760b 100644
> --- a/fs/xfs/scrub/agheader_repair.c
> +++ b/fs/xfs/scrub/agheader_repair.c
> @@ -191,14 +191,15 @@ xrep_agf_init_header(
>         struct xfs_agf          *old_agf)
>  {
>         struct xfs_mount        *mp = sc->mp;
> +       struct xfs_perag        *pag = sc->sa.pag;
>         struct xfs_agf          *agf = agf_bp->b_addr;
>  
>         memcpy(old_agf, agf, sizeof(*old_agf));
>         memset(agf, 0, BBTOB(agf_bp->b_length));
>         agf->agf_magicnum = cpu_to_be32(XFS_AGF_MAGIC);
>         agf->agf_versionnum = cpu_to_be32(XFS_AGF_VERSION);
> -       agf->agf_seqno = cpu_to_be32(sc->sa.pag->pag_agno);
> -       agf->agf_length = cpu_to_be32(sc->sa.pag->block_count);
> +       agf->agf_seqno = cpu_to_be32(pag->pag_agno);
> +       agf->agf_length = cpu_to_be32(pag->block_count);
>         agf->agf_flfirst = old_agf->agf_flfirst;
>         agf->agf_fllast = old_agf->agf_fllast;
>         agf->agf_flcount = old_agf->agf_flcount;
> @@ -206,8 +207,8 @@ xrep_agf_init_header(
>                 uuid_copy(&agf->agf_uuid, &mp->m_sb.sb_meta_uuid);
>  
>         /* Mark the incore AGF data stale until we're done fixing
> things. */
> -       ASSERT(sc->sa.pag->pagf_init);
> -       sc->sa.pag->pagf_init = 0;
> +       ASSERT(xfs_perag_initialised_agf(pag));
> +       clear_bit(XFS_AGSTATE_AGF_INIT, &pag->pag_opstate);
>  }
>  
>  /* Set btree root information in an AGF. */
> @@ -333,7 +334,7 @@ xrep_agf_commit_new(
>         pag->pagf_levels[XFS_BTNUM_RMAPi] =
>                         be32_to_cpu(agf-
> >agf_levels[XFS_BTNUM_RMAPi]);
>         pag->pagf_refcount_level = be32_to_cpu(agf-
> >agf_refcount_level);
> -       pag->pagf_init = 1;
> +       set_bit(XFS_AGSTATE_AGF_INIT, &pag->pag_opstate);
>  
>         return 0;
>  }
> @@ -434,7 +435,7 @@ xrep_agf(
>  
>  out_revert:
>         /* Mark the incore AGF state stale and revert the AGF. */
> -       sc->sa.pag->pagf_init = 0;
> +       clear_bit(XFS_AGSTATE_AGF_INIT, &sc->sa.pag->pag_opstate);
>         memcpy(agf, &old_agf, sizeof(old_agf));
>         return error;
>  }
> @@ -618,7 +619,7 @@ xrep_agfl_update_agf(
>         xfs_force_summary_recalc(sc->mp);
>  
>         /* Update the AGF counters. */
> -       if (sc->sa.pag->pagf_init)
> +       if (xfs_perag_initialised_agf(sc->sa.pag))
>                 sc->sa.pag->pagf_flcount = flcount;
>         agf->agf_flfirst = cpu_to_be32(0);
>         agf->agf_flcount = cpu_to_be32(flcount);
> @@ -822,14 +823,15 @@ xrep_agi_init_header(
>         struct xfs_agi          *old_agi)
>  {
>         struct xfs_agi          *agi = agi_bp->b_addr;
> +       struct xfs_perag        *pag = sc->sa.pag;
>         struct xfs_mount        *mp = sc->mp;
>  
>         memcpy(old_agi, agi, sizeof(*old_agi));
>         memset(agi, 0, BBTOB(agi_bp->b_length));
>         agi->agi_magicnum = cpu_to_be32(XFS_AGI_MAGIC);
>         agi->agi_versionnum = cpu_to_be32(XFS_AGI_VERSION);
> -       agi->agi_seqno = cpu_to_be32(sc->sa.pag->pag_agno);
> -       agi->agi_length = cpu_to_be32(sc->sa.pag->block_count);
> +       agi->agi_seqno = cpu_to_be32(pag->pag_agno);
> +       agi->agi_length = cpu_to_be32(pag->block_count);
>         agi->agi_newino = cpu_to_be32(NULLAGINO);
>         agi->agi_dirino = cpu_to_be32(NULLAGINO);
>         if (xfs_has_crc(mp))
> @@ -840,8 +842,8 @@ xrep_agi_init_header(
>                         sizeof(agi->agi_unlinked));
>  
>         /* Mark the incore AGF data stale until we're done fixing
> things. */
> -       ASSERT(sc->sa.pag->pagi_init);
> -       sc->sa.pag->pagi_init = 0;
> +       ASSERT(xfs_perag_initialised_agi(pag));
> +       clear_bit(XFS_AGSTATE_AGI_INIT, &pag->pag_opstate);
>  }
>  
>  /* Set btree root information in an AGI. */
> @@ -928,7 +930,7 @@ xrep_agi_commit_new(
>         pag = sc->sa.pag;
>         pag->pagi_count = be32_to_cpu(agi->agi_count);
>         pag->pagi_freecount = be32_to_cpu(agi->agi_freecount);
> -       pag->pagi_init = 1;
> +       set_bit(XFS_AGSTATE_AGI_INIT, &pag->pag_opstate);
>  
>         return 0;
>  }
> @@ -993,7 +995,7 @@ xrep_agi(
>  
>  out_revert:
>         /* Mark the incore AGI state stale and revert the AGI. */
> -       sc->sa.pag->pagi_init = 0;
> +       clear_bit(XFS_AGSTATE_AGI_INIT, &sc->sa.pag->pag_opstate);
>         memcpy(agi, &old_agi, sizeof(old_agi));
>         return error;
>  }
> diff --git a/fs/xfs/scrub/fscounters.c b/fs/xfs/scrub/fscounters.c
> index ef97670970c3..f0c7f41897b9 100644
> --- a/fs/xfs/scrub/fscounters.c
> +++ b/fs/xfs/scrub/fscounters.c
> @@ -86,7 +86,8 @@ xchk_fscount_warmup(
>         for_each_perag(mp, agno, pag) {
>                 if (xchk_should_terminate(sc, &error))
>                         break;
> -               if (pag->pagi_init && pag->pagf_init)
> +               if (xfs_perag_initialised_agi(pag) &&
> +                   xfs_perag_initialised_agf(pag))
>                         continue;
>  
>                 /* Lock both AG headers. */
> @@ -101,7 +102,8 @@ xchk_fscount_warmup(
>                  * These are supposed to be initialized by the header
> read
>                  * function.
>                  */
> -               if (!pag->pagi_init || !pag->pagf_init) {
> +               if (!xfs_perag_initialised_agi(pag) ||
> +                   !xfs_perag_initialised_agf(pag)) {
>                         error = -EFSCORRUPTED;
>                         break;
>                 }
> @@ -220,7 +222,8 @@ xchk_fscount_aggregate_agcounts(
>                         break;
>  
>                 /* This somehow got unset since the warmup? */
> -               if (!pag->pagi_init || !pag->pagf_init) {
> +               if (!xfs_perag_initialised_agi(pag) ||
> +                   !xfs_perag_initialised_agf(pag)) {
>                         error = -EFSCORRUPTED;
>                         break;
>                 }
> diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
> index 4b92f9253ccd..d0b1644efb89 100644
> --- a/fs/xfs/scrub/repair.c
> +++ b/fs/xfs/scrub/repair.c
> @@ -206,7 +206,7 @@ xrep_calc_ag_resblks(
>                 return 0;
>  
>         pag = xfs_perag_get(mp, sm->sm_agno);
> -       if (pag->pagi_init) {
> +       if (xfs_perag_initialised_agi(pag)) {
>                 /* Use in-core icount if possible. */
>                 icount = pag->pagi_count;
>         } else {
> diff --git a/fs/xfs/xfs_filestream.c b/fs/xfs/xfs_filestream.c
> index 34b21a29c39b..7e8b25ab6c46 100644
> --- a/fs/xfs/xfs_filestream.c
> +++ b/fs/xfs/xfs_filestream.c
> @@ -125,7 +125,7 @@ xfs_filestream_pick_ag(
>  
>                 pag = xfs_perag_get(mp, ag);
>  
> -               if (!pag->pagf_init) {
> +               if (!xfs_perag_initialised_agf(pag)) {
>                         err = xfs_alloc_read_agf(pag, NULL, trylock,
> NULL);
>                         if (err) {
>                                 if (err != -EAGAIN) {
> @@ -159,7 +159,8 @@ xfs_filestream_pick_ag(
>                                 xfs_ag_resv_needed(pag,
> XFS_AG_RESV_NONE));
>                 if (((minlen && longest >= minlen) ||
>                      (!minlen && pag->pagf_freeblks >= minfree)) &&
> -                   (!pag->pagf_metadata || !(flags &
> XFS_PICK_USERDATA) ||
> +                   (!xfs_perag_prefers_metadata(pag) ||
> +                    !(flags & XFS_PICK_USERDATA) ||
>                      (flags & XFS_PICK_LOWSPACE))) {
>  
>                         /* Break out, retaining the reference on the
> AG. */
> diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> index 96375b5622fd..2479b5cbd75e 100644
> --- a/fs/xfs/xfs_super.c
> +++ b/fs/xfs/xfs_super.c
> @@ -247,6 +247,32 @@ xfs_fs_show_options(
>         return 0;
>  }
>  
> +static bool
> +xfs_set_inode_alloc_perag(
> +       struct xfs_perag        *pag,
> +       xfs_ino_t               ino,
> +       xfs_agnumber_t          max_metadata)
> +{
> +       if (!xfs_is_inode32(pag->pag_mount)) {
> +               set_bit(XFS_AGSTATE_ALLOWS_INODES, &pag-
> >pag_opstate);
> +               clear_bit(XFS_AGSTATE_PREFERS_METADATA, &pag-
> >pag_opstate);
> +               return false;
> +       }
> +
> +       if (ino > XFS_MAXINUMBER_32) {
> +               clear_bit(XFS_AGSTATE_ALLOWS_INODES, &pag-
> >pag_opstate);
> +               clear_bit(XFS_AGSTATE_PREFERS_METADATA, &pag-
> >pag_opstate);
> +               return false;
> +       }
> +
> +       set_bit(XFS_AGSTATE_ALLOWS_INODES, &pag->pag_opstate);
> +       if (pag->pag_agno < max_metadata)
> +               set_bit(XFS_AGSTATE_PREFERS_METADATA, &pag-
> >pag_opstate);
> +       else
> +               clear_bit(XFS_AGSTATE_PREFERS_METADATA, &pag-
> >pag_opstate);
> +       return true;
> +}
> +
>  /*
>   * Set parameters for inode allocation heuristics, taking into
> account
>   * filesystem size and inode32/inode64 mount options; i.e.
> specifically
> @@ -310,24 +336,8 @@ xfs_set_inode_alloc(
>                 ino = XFS_AGINO_TO_INO(mp, index, agino);
>  
>                 pag = xfs_perag_get(mp, index);
> -
> -               if (xfs_is_inode32(mp)) {
> -                       if (ino > XFS_MAXINUMBER_32) {
> -                               pag->pagi_inodeok = 0;
> -                               pag->pagf_metadata = 0;
> -                       } else {
> -                               pag->pagi_inodeok = 1;
> -                               maxagi++;
> -                               if (index < max_metadata)
> -                                       pag->pagf_metadata = 1;
> -                               else
> -                                       pag->pagf_metadata = 0;
> -                       }
> -               } else {
> -                       pag->pagi_inodeok = 1;
> -                       pag->pagf_metadata = 0;
> -               }
> -
> +               if (xfs_set_inode_alloc_perag(pag, ino,
> max_metadata))
> +                       maxagi++;
>                 xfs_perag_put(pag);
>         }
>  





[Index of Archives]     [XFS Filesystem Development (older mail)]     [Linux Filesystem Development]     [Linux Audio Users]     [Yosemite Trails]     [Linux Kernel]     [Linux RAID]     [Linux SCSI]


  Powered by Linux