1 files changed, 57 insertions, 48 deletions

diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 7f7be5f98f52..bb262c25c8de 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -896,7 +896,6 @@ xfs_dinode_to_disk(
         to->di_projid_lo = cpu_to_be16(from->di_projid_lo);
         to->di_projid_hi = cpu_to_be16(from->di_projid_hi);
         memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
-        to->di_flushiter = cpu_to_be16(from->di_flushiter);
         to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec);
         to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec);
         to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec);
@@ -924,6 +923,9 @@ xfs_dinode_to_disk(
                 to->di_lsn = cpu_to_be64(from->di_lsn);
                 memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2));
                 uuid_copy(&to->di_uuid, &from->di_uuid);
+                to->di_flushiter = 0;
+        } else {
+                to->di_flushiter = cpu_to_be16(from->di_flushiter);
         }
 }
 
@@ -1028,6 +1030,15 @@ xfs_dinode_calc_crc(
 
 /*
  * Read the disk inode attributes into the in-core inode structure.
+ *
+ * For version 5 superblocks, if we are initialising a new inode and we are not
+ * utilising the XFS_MOUNT_IKEEP inode cluster mode, we can simple build the new
+ * inode core with a random generation number. If we are keeping inodes around,
+ * we need to read the inode cluster to get the existing generation number off
+ * disk. Further, if we are using version 4 superblocks (i.e. v1/v2 inode
+ * format) then log recovery is dependent on the di_flushiter field being
+ * initialised from the current on-disk value and hence we must also read the
+ * inode off disk.
  */
 int
 xfs_iread(
@@ -1047,6 +1058,23 @@ xfs_iread(
         if (error)
                 return error;
 
+        /* shortcut IO on inode allocation if possible */
+        if ((iget_flags & XFS_IGET_CREATE) &&
+            xfs_sb_version_hascrc(&mp->m_sb) &&
+            !(mp->m_flags & XFS_MOUNT_IKEEP)) {
+                /* initialise the on-disk inode core */
+                memset(&ip->i_d, 0, sizeof(ip->i_d));
+                ip->i_d.di_magic = XFS_DINODE_MAGIC;
+                ip->i_d.di_gen = prandom_u32();
+                if (xfs_sb_version_hascrc(&mp->m_sb)) {
+                        ip->i_d.di_version = 3;
+                        ip->i_d.di_ino = ip->i_ino;
+                        uuid_copy(&ip->i_d.di_uuid, &mp->m_sb.sb_uuid);
+                } else
+                        ip->i_d.di_version = 2;
+                return 0;
+        }
+
         /*
          * Get pointers to the on-disk inode and the buffer containing it.
          */
@@ -1133,17 +1161,16 @@ xfs_iread(
         xfs_buf_set_ref(bp, XFS_INO_REF);
 
         /*
-         * Use xfs_trans_brelse() to release the buffer containing the
-         * on-disk inode, because it was acquired with xfs_trans_read_buf()
-         * in xfs_imap_to_bp() above.  If tp is NULL, this is just a normal
+         * Use xfs_trans_brelse() to release the buffer containing the on-disk
+         * inode, because it was acquired with xfs_trans_read_buf() in
+         * xfs_imap_to_bp() above.  If tp is NULL, this is just a normal
          * brelse().  If we're within a transaction, then xfs_trans_brelse()
          * will only release the buffer if it is not dirty within the
          * transaction.  It will be OK to release the buffer in this case,
-         * because inodes on disk are never destroyed and we will be
-         * locking the new in-core inode before putting it in the hash
-         * table where other processes can find it.  Thus we don't have
-         * to worry about the inode being changed just because we released
-         * the buffer.
+         * because inodes on disk are never destroyed and we will be locking the
+         * new in-core inode before putting it in the cache where other
+         * processes can find it.  Thus we don't have to worry about the inode
+         * being changed just because we released the buffer.
          */
  out_brelse:
         xfs_trans_brelse(tp, bp);
@@ -2028,8 +2055,6 @@ xfs_ifree(
         int                     error;
         int                     delete;
         xfs_ino_t               first_ino;
-        xfs_dinode_t            *dip;
-        xfs_buf_t               *ibp;
 
         ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
         ASSERT(ip->i_d.di_nlink == 0);
@@ -2042,14 +2067,13 @@ xfs_ifree(
          * Pull the on-disk inode from the AGI unlinked list.
          */
         error = xfs_iunlink_remove(tp, ip);
-        if (error != 0) {
+        if (error)
                 return error;
-        }
 
         error = xfs_difree(tp, ip->i_ino, flist, &delete, &first_ino);
-        if (error != 0) {
+        if (error)
                 return error;
-        }
+
         ip->i_d.di_mode = 0;            /* mark incore inode as free */
         ip->i_d.di_flags = 0;
         ip->i_d.di_dmevmask = 0;
@@ -2061,31 +2085,10 @@ xfs_ifree(
          * by reincarnations of this inode.
          */
         ip->i_d.di_gen++;
-
         xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 
-        error = xfs_imap_to_bp(ip->i_mount, tp, &ip->i_imap, &dip, &ibp,
-                               0, 0);
-        if (error)
-                return error;
-
-        /*
-        * Clear the on-disk di_mode. This is to prevent xfs_bulkstat
-        * from picking up this inode when it is reclaimed (its incore state
-        * initialzed but not flushed to disk yet). The in-core di_mode is
-        * already cleared  and a corresponding transaction logged.
-        * The hack here just synchronizes the in-core to on-disk
-        * di_mode value in advance before the actual inode sync to disk.
-        * This is OK because the inode is already unlinked and would never
-        * change its di_mode again for this inode generation.
-        * This is a temporary hack that would require a proper fix
-        * in the future.
-        */
-        dip->di_mode = 0;
-
-        if (delete) {
+        if (delete)
                 error = xfs_ifree_cluster(ip, tp, first_ino);
-        }
 
         return error;
 }
@@ -2160,8 +2163,8 @@ xfs_iroot_realloc(
                 np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
                                                      (int)new_size);
                 ifp->if_broot_bytes = (int)new_size;
-                ASSERT(ifp->if_broot_bytes <=
-                        XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ(ip));
+                ASSERT(XFS_BMAP_BMDR_SPACE(ifp->if_broot) <=
+                        XFS_IFORK_SIZE(ip, whichfork));
                 memmove(np, op, cur_max * (uint)sizeof(xfs_dfsbno_t));
                 return;
         }
@@ -2214,8 +2217,9 @@ xfs_iroot_realloc(
         kmem_free(ifp->if_broot);
         ifp->if_broot = new_broot;
         ifp->if_broot_bytes = (int)new_size;
-        ASSERT(ifp->if_broot_bytes <=
-                XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ(ip));
+        if (ifp->if_broot)
+                ASSERT(XFS_BMAP_BMDR_SPACE(ifp->if_broot) <=
+                        XFS_IFORK_SIZE(ip, whichfork));
         return;
 }
 
@@ -2526,9 +2530,8 @@ xfs_iflush_fork(
                 if ((iip->ili_fields & brootflag[whichfork]) &&
                     (ifp->if_broot_bytes > 0)) {
                         ASSERT(ifp->if_broot != NULL);
-                        ASSERT(ifp->if_broot_bytes <=
-                               (XFS_IFORK_SIZE(ip, whichfork) +
-                                XFS_BROOT_SIZE_ADJ(ip)));
+                        ASSERT(XFS_BMAP_BMDR_SPACE(ifp->if_broot) <=
+                                XFS_IFORK_SIZE(ip, whichfork));
                         xfs_bmbt_to_bmdr(mp, ifp->if_broot, ifp->if_broot_bytes,
                                 (xfs_bmdr_block_t *)cp,
                                 XFS_DFORK_SIZE(dip, mp, whichfork));
@@ -2886,12 +2889,18 @@ xfs_iflush_int(
                         __func__, ip->i_ino, ip->i_d.di_forkoff, ip);
                 goto corrupt_out;
         }
+
         /*
-         * bump the flush iteration count, used to detect flushes which
-         * postdate a log record during recovery. This is redundant as we now
-         * log every change and hence this can't happen. Still, it doesn't hurt.
+         * Inode item log recovery for v1/v2 inodes are dependent on the
+         * di_flushiter count for correct sequencing. We bump the flush
+         * iteration count so we can detect flushes which postdate a log record
+         * during recovery. This is redundant as we now log every change and
+         * hence this can't happen but we need to still do it to ensure
+         * backwards compatibility with old kernels that predate logging all
+         * inode changes.
          */
-        ip->i_d.di_flushiter++;
+        if (ip->i_d.di_version < 3)
+                ip->i_d.di_flushiter++;
 
         /*
          * Copy the dirty parts of the inode into the on-disk