diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c index 193a5b4909c5bd280d1ee64c746b5c87e8a228ee..350ad203b082b32b16d55ef8bf48d9310c180a5e 100644 --- a/fs/xfs/libxfs/xfs_alloc.c +++ b/fs/xfs/libxfs/xfs_alloc.c @@ -39,6 +39,9 @@ #include "xfs_buf_item.h" #include "xfs_log.h" #include "xfs_ag_resv.h" +#include "xfs_bmap.h" + +extern kmem_zone_t *xfs_bmap_free_item_zone; struct workqueue_struct *xfs_alloc_wq; @@ -2171,6 +2174,40 @@ xfs_agfl_reset( pag->pagf_agflreset = false; } +/* + * Defer an AGFL block free. This is effectively equivalent to + * xfs_bmap_add_free() with some special handling particular to AGFL blocks. + * + * Deferring AGFL frees helps prevent log reservation overruns due to too many + * allocation operations in a transaction. AGFL frees are prone to this problem + * because for one they are always freed one at a time. Further, an immediate + * AGFL block free can cause a btree join and require another block free before + * the real allocation can proceed. Deferring the free disconnects freeing up + * the AGFL slot from freeing the block. + */ +STATIC void +xfs_defer_agfl_block( + struct xfs_mount *mp, + struct xfs_defer_ops *dfops, + xfs_agnumber_t agno, + xfs_fsblock_t agbno, + struct xfs_owner_info *oinfo) +{ + struct xfs_extent_free_item *new; /* new element */ + + ASSERT(xfs_bmap_free_item_zone != NULL); + ASSERT(oinfo != NULL); + + new = kmem_zone_alloc(xfs_bmap_free_item_zone, KM_SLEEP); + new->xefi_startblock = XFS_AGB_TO_FSB(mp, agno, agbno); + new->xefi_blockcount = 1; + new->xefi_oinfo = *oinfo; + + trace_xfs_agfl_free_defer(mp, agno, 0, agbno, 1); + + xfs_defer_add(dfops, XFS_DEFER_OPS_TYPE_AGFL_FREE, &new->xefi_list); +} + /* * Decide whether to use this allocation group for this allocation. * If so, fix up the btree freelist's size. @@ -2275,10 +2312,16 @@ xfs_alloc_fix_freelist( if (error) goto out_agbp_relse; - error = xfs_free_agfl_block(tp, args->agno, bno, agbp, - &targs.oinfo); - if (error) - goto out_agbp_relse; + /* defer agfl frees if dfops is provided */ + if (tp->t_agfl_dfops) { + xfs_defer_agfl_block(mp, tp->t_agfl_dfops, args->agno, + bno, &targs.oinfo); + } else { + error = xfs_free_agfl_block(tp, args->agno, bno, agbp, + &targs.oinfo); + if (error) + goto out_agbp_relse; + } } targs.tp = tp; diff --git a/fs/xfs/libxfs/xfs_defer.h b/fs/xfs/libxfs/xfs_defer.h index 045beacdd37d81c9e01e0ab46ab2bbcbbb581fbb..e70725ba1f5f794d2b262084e44ac3f80680c7b5 100644 --- a/fs/xfs/libxfs/xfs_defer.h +++ b/fs/xfs/libxfs/xfs_defer.h @@ -55,6 +55,7 @@ enum xfs_defer_ops_type { XFS_DEFER_OPS_TYPE_REFCOUNT, XFS_DEFER_OPS_TYPE_RMAP, XFS_DEFER_OPS_TYPE_FREE, + XFS_DEFER_OPS_TYPE_AGFL_FREE, XFS_DEFER_OPS_TYPE_MAX, }; diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index 21d494f30ea724374af9b444ad4e86b2f3b4017b..24892259301eea204dcf712e54d430ac249828aa 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -2436,6 +2436,8 @@ DEFINE_DEFER_PENDING_EVENT(xfs_defer_pending_abort); #define DEFINE_BMAP_FREE_DEFERRED_EVENT DEFINE_PHYS_EXTENT_DEFERRED_EVENT DEFINE_BMAP_FREE_DEFERRED_EVENT(xfs_bmap_free_defer); DEFINE_BMAP_FREE_DEFERRED_EVENT(xfs_bmap_free_deferred); +DEFINE_BMAP_FREE_DEFERRED_EVENT(xfs_agfl_free_defer); +DEFINE_BMAP_FREE_DEFERRED_EVENT(xfs_agfl_free_deferred); /* rmap tracepoints */ DECLARE_EVENT_CLASS(xfs_rmap_class, diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c index d6d8f9d129a779eb5a53d6939e0edaf07719a54e..06adb1a3e31f6cd74dd6e467dcdbce10a59fe11e 100644 --- a/fs/xfs/xfs_trans.c +++ b/fs/xfs/xfs_trans.c @@ -31,6 +31,7 @@ #include "xfs_log.h" #include "xfs_trace.h" #include "xfs_error.h" +#include "xfs_defer.h" kmem_zone_t *xfs_trans_zone; kmem_zone_t *xfs_log_item_desc_zone; @@ -94,11 +95,11 @@ xfs_trans_free( * blocks. Locks and log items, however, are no inherited. They must * be added to the new transaction explicitly. */ -STATIC xfs_trans_t * +STATIC struct xfs_trans * xfs_trans_dup( - xfs_trans_t *tp) + struct xfs_trans *tp) { - xfs_trans_t *ntp; + struct xfs_trans *ntp; ntp = kmem_zone_zalloc(xfs_trans_zone, KM_SLEEP); @@ -127,6 +128,7 @@ xfs_trans_dup( ntp->t_rtx_res = tp->t_rtx_res - tp->t_rtx_res_used; tp->t_rtx_res = tp->t_rtx_res_used; ntp->t_pflags = tp->t_pflags; + ntp->t_agfl_dfops = tp->t_agfl_dfops; xfs_trans_dup_dqinfo(tp, ntp); @@ -936,6 +938,9 @@ __xfs_trans_commit( int error = 0; int sync = tp->t_flags & XFS_TRANS_SYNC; + ASSERT(!tp->t_agfl_dfops || + !xfs_defer_has_unfinished_work(tp->t_agfl_dfops) || regrant); + /* * If there is nothing to be logged by the transaction, * then unlock all of the items associated with the diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h index 9d542dfe00524cf23aa99fe0d22f5a2dba9353ab..834388c2c9defa7da4ec72e6640642f1c8e12a6c 100644 --- a/fs/xfs/xfs_trans.h +++ b/fs/xfs/xfs_trans.h @@ -111,6 +111,7 @@ typedef struct xfs_trans { struct xlog_ticket *t_ticket; /* log mgr ticket */ struct xfs_mount *t_mountp; /* ptr to fs mount struct */ struct xfs_dquot_acct *t_dqinfo; /* acctg info for dquots */ + struct xfs_defer_ops *t_agfl_dfops; /* optional agfl fixup dfops */ unsigned int t_flags; /* misc flags */ int64_t t_icount_delta; /* superblock icount change */ int64_t t_ifree_delta; /* superblock ifree change */ diff --git a/fs/xfs/xfs_trans_extfree.c b/fs/xfs/xfs_trans_extfree.c index ab438647592a847dc8918dea378070c11b05eb38..f5620796ae259e3f15676dfdf71d7da90484ebf3 100644 --- a/fs/xfs/xfs_trans_extfree.c +++ b/fs/xfs/xfs_trans_extfree.c @@ -231,9 +231,79 @@ static const struct xfs_defer_op_type xfs_extent_free_defer_type = { .cancel_item = xfs_extent_free_cancel_item, }; +/* + * AGFL blocks are accounted differently in the reserve pools and are not + * inserted into the busy extent list. + */ +STATIC int +xfs_agfl_free_finish_item( + struct xfs_trans *tp, + struct xfs_defer_ops *dop, + struct list_head *item, + void *done_item, + void **state) +{ + struct xfs_mount *mp = tp->t_mountp; + struct xfs_efd_log_item *efdp = done_item; + struct xfs_extent_free_item *free; + struct xfs_extent *extp; + struct xfs_buf *agbp; + int error; + xfs_agnumber_t agno; + xfs_agblock_t agbno; + uint next_extent; + + free = container_of(item, struct xfs_extent_free_item, xefi_list); + ASSERT(free->xefi_blockcount == 1); + agno = XFS_FSB_TO_AGNO(mp, free->xefi_startblock); + agbno = XFS_FSB_TO_AGBNO(mp, free->xefi_startblock); + + trace_xfs_agfl_free_deferred(mp, agno, 0, agbno, free->xefi_blockcount); + + error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp); + if (!error) + error = xfs_free_agfl_block(tp, agno, agbno, agbp, + &free->xefi_oinfo); + + /* + * Mark the transaction dirty, even on error. This ensures the + * transaction is aborted, which: + * + * 1.) releases the EFI and frees the EFD + * 2.) shuts down the filesystem + */ + tp->t_flags |= XFS_TRANS_DIRTY; + efdp->efd_item.li_desc->lid_flags |= XFS_LID_DIRTY; + + next_extent = efdp->efd_next_extent; + ASSERT(next_extent < efdp->efd_format.efd_nextents); + extp = &(efdp->efd_format.efd_extents[next_extent]); + extp->ext_start = free->xefi_startblock; + extp->ext_len = free->xefi_blockcount; + efdp->efd_next_extent++; + + kmem_free(free); + return error; +} + + +/* sub-type with special handling for AGFL deferred frees */ +static const struct xfs_defer_op_type xfs_agfl_free_defer_type = { + .type = XFS_DEFER_OPS_TYPE_AGFL_FREE, + .max_items = XFS_EFI_MAX_FAST_EXTENTS, + .diff_items = xfs_extent_free_diff_items, + .create_intent = xfs_extent_free_create_intent, + .abort_intent = xfs_extent_free_abort_intent, + .log_item = xfs_extent_free_log_item, + .create_done = xfs_extent_free_create_done, + .finish_item = xfs_agfl_free_finish_item, + .cancel_item = xfs_extent_free_cancel_item, +}; + /* Register the deferred op type. */ void xfs_extent_free_init_defer_op(void) { xfs_defer_init_op_type(&xfs_extent_free_defer_type); + xfs_defer_init_op_type(&xfs_agfl_free_defer_type); }