xfs: rmap btree transaction reservations
authorDarrick J. Wong <darrick.wong@oracle.com>
Wed, 3 Aug 2016 01:37:10 +0000 (11:37 +1000)
committerDave Chinner <david@fromorbit.com>
Wed, 3 Aug 2016 01:37:10 +0000 (11:37 +1000)
The rmap btrees will use the AGFL as the block allocation source, so
we need to ensure that the transaction reservations reflect the fact
this tree is modified by allocation and freeing. Hence we need to
extend all the extent allocation/free reservations used in
transactions to handle this.

Note that this also gets rid of the unused XFS_ALLOCFREE_LOG_RES
macro, as we now do buffer reservations based on the number of
buffers logged via xfs_calc_buf_res(). Hence we only need the buffer
count calculation now.

[darrick: use rmap_maxlevels when calculating log block resv]

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
fs/xfs/libxfs/xfs_trans_resv.c
fs/xfs/libxfs/xfs_trans_resv.h

index 4c7eb9d..301ef2f 100644 (file)
@@ -63,6 +63,30 @@ xfs_calc_buf_res(
        return nbufs * (size + xfs_buf_log_overhead());
 }
 
+/*
+ * Per-extent log reservation for the btree changes involved in freeing or
+ * allocating an extent.  In classic XFS there were two trees that will be
+ * modified (bnobt + cntbt).  With rmap enabled, there are three trees
+ * (rmapbt).  The number of blocks reserved is based on the formula:
+ *
+ * num trees * ((2 blocks/level * max depth) - 1)
+ *
+ * Keep in mind that max depth is calculated separately for each type of tree.
+ */
+static uint
+xfs_allocfree_log_count(
+       struct xfs_mount *mp,
+       uint            num_ops)
+{
+       uint            blocks;
+
+       blocks = num_ops * 2 * (2 * mp->m_ag_maxlevels - 1);
+       if (xfs_sb_version_hasrmapbt(&mp->m_sb))
+               blocks += num_ops * (2 * mp->m_rmap_maxlevels - 1);
+
+       return blocks;
+}
+
 /*
  * Logging inodes is really tricksy. They are logged in memory format,
  * which means that what we write into the log doesn't directly translate into
@@ -126,7 +150,7 @@ xfs_calc_inode_res(
  */
 STATIC uint
 xfs_calc_finobt_res(
-       struct xfs_mount        *mp,
+       struct xfs_mount        *mp,
        int                     alloc,
        int                     modify)
 {
@@ -137,7 +161,7 @@ xfs_calc_finobt_res(
 
        res = xfs_calc_buf_res(mp->m_in_maxlevels, XFS_FSB_TO_B(mp, 1));
        if (alloc)
-               res += xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1), 
+               res += xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                        XFS_FSB_TO_B(mp, 1));
        if (modify)
                res += (uint)XFS_FSB_TO_B(mp, 1);
@@ -188,10 +212,10 @@ xfs_calc_write_reservation(
                     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK),
                                      XFS_FSB_TO_B(mp, 1)) +
                     xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 2),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 2),
                                      XFS_FSB_TO_B(mp, 1))),
                    (xfs_calc_buf_res(5, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 2),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 2),
                                      XFS_FSB_TO_B(mp, 1))));
 }
 
@@ -217,10 +241,10 @@ xfs_calc_itruncate_reservation(
                     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) + 1,
                                      XFS_FSB_TO_B(mp, 1))),
                    (xfs_calc_buf_res(9, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 4),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 4),
                                      XFS_FSB_TO_B(mp, 1)) +
                    xfs_calc_buf_res(5, 0) +
-                   xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+                   xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                     XFS_FSB_TO_B(mp, 1)) +
                    xfs_calc_buf_res(2 + mp->m_ialloc_blks +
                                     mp->m_in_maxlevels, 0)));
@@ -247,7 +271,7 @@ xfs_calc_rename_reservation(
                     xfs_calc_buf_res(2 * XFS_DIROP_LOG_COUNT(mp),
                                      XFS_FSB_TO_B(mp, 1))),
                    (xfs_calc_buf_res(7, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 3),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 3),
                                      XFS_FSB_TO_B(mp, 1))));
 }
 
@@ -286,7 +310,7 @@ xfs_calc_link_reservation(
                     xfs_calc_buf_res(XFS_DIROP_LOG_COUNT(mp),
                                      XFS_FSB_TO_B(mp, 1))),
                    (xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                      XFS_FSB_TO_B(mp, 1))));
 }
 
@@ -324,7 +348,7 @@ xfs_calc_remove_reservation(
                     xfs_calc_buf_res(XFS_DIROP_LOG_COUNT(mp),
                                      XFS_FSB_TO_B(mp, 1))),
                    (xfs_calc_buf_res(4, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 2),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 2),
                                      XFS_FSB_TO_B(mp, 1))));
 }
 
@@ -371,7 +395,7 @@ xfs_calc_create_resv_alloc(
                mp->m_sb.sb_sectsize +
                xfs_calc_buf_res(mp->m_ialloc_blks, XFS_FSB_TO_B(mp, 1)) +
                xfs_calc_buf_res(mp->m_in_maxlevels, XFS_FSB_TO_B(mp, 1)) +
-               xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+               xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                 XFS_FSB_TO_B(mp, 1));
 }
 
@@ -399,7 +423,7 @@ xfs_calc_icreate_resv_alloc(
        return xfs_calc_buf_res(2, mp->m_sb.sb_sectsize) +
                mp->m_sb.sb_sectsize +
                xfs_calc_buf_res(mp->m_in_maxlevels, XFS_FSB_TO_B(mp, 1)) +
-               xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+               xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                 XFS_FSB_TO_B(mp, 1)) +
                xfs_calc_finobt_res(mp, 0, 0);
 }
@@ -483,7 +507,7 @@ xfs_calc_ifree_reservation(
                xfs_calc_buf_res(1, 0) +
                xfs_calc_buf_res(2 + mp->m_ialloc_blks +
                                 mp->m_in_maxlevels, 0) +
-               xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+               xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                 XFS_FSB_TO_B(mp, 1)) +
                xfs_calc_finobt_res(mp, 0, 1);
 }
@@ -513,7 +537,7 @@ xfs_calc_growdata_reservation(
        struct xfs_mount        *mp)
 {
        return xfs_calc_buf_res(3, mp->m_sb.sb_sectsize) +
-               xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+               xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                 XFS_FSB_TO_B(mp, 1));
 }
 
@@ -535,7 +559,7 @@ xfs_calc_growrtalloc_reservation(
                xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK),
                                 XFS_FSB_TO_B(mp, 1)) +
                xfs_calc_inode_res(mp, 1) +
-               xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+               xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                 XFS_FSB_TO_B(mp, 1));
 }
 
@@ -611,7 +635,7 @@ xfs_calc_addafork_reservation(
                xfs_calc_buf_res(1, mp->m_dir_geo->blksize) +
                xfs_calc_buf_res(XFS_DAENTER_BMAP1B(mp, XFS_DATA_FORK) + 1,
                                 XFS_FSB_TO_B(mp, 1)) +
-               xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 1),
+               xfs_calc_buf_res(xfs_allocfree_log_count(mp, 1),
                                 XFS_FSB_TO_B(mp, 1));
 }
 
@@ -634,7 +658,7 @@ xfs_calc_attrinval_reservation(
                    xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK),
                                     XFS_FSB_TO_B(mp, 1))),
                   (xfs_calc_buf_res(9, mp->m_sb.sb_sectsize) +
-                   xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 4),
+                   xfs_calc_buf_res(xfs_allocfree_log_count(mp, 4),
                                     XFS_FSB_TO_B(mp, 1))));
 }
 
@@ -701,7 +725,7 @@ xfs_calc_attrrm_reservation(
                                        XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK)) +
                     xfs_calc_buf_res(XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK), 0)),
                    (xfs_calc_buf_res(5, mp->m_sb.sb_sectsize) +
-                    xfs_calc_buf_res(XFS_ALLOCFREE_LOG_COUNT(mp, 2),
+                    xfs_calc_buf_res(xfs_allocfree_log_count(mp, 2),
                                      XFS_FSB_TO_B(mp, 1))));
 }
 
index 7978150..0eb46ed 100644 (file)
@@ -67,16 +67,6 @@ struct xfs_trans_resv {
 /* shorthand way of accessing reservation structure */
 #define M_RES(mp)      (&(mp)->m_resv)
 
-/*
- * Per-extent log reservation for the allocation btree changes
- * involved in freeing or allocating an extent.
- * 2 trees * (2 blocks/level * max depth - 1) * block size
- */
-#define        XFS_ALLOCFREE_LOG_RES(mp,nx) \
-       ((nx) * (2 * XFS_FSB_TO_B((mp), 2 * (mp)->m_ag_maxlevels - 1)))
-#define        XFS_ALLOCFREE_LOG_COUNT(mp,nx) \
-       ((nx) * (2 * (2 * (mp)->m_ag_maxlevels - 1)))
-
 /*
  * Per-directory log reservation for any directory change.
  * dir blocks: (1 btree block per level + data block + free block) * dblock size