xfs: remove the di_version field from struct icdinode
authorChristoph Hellwig <hch@lst.de>
Wed, 18 Mar 2020 15:15:11 +0000 (08:15 -0700)
committerDarrick J. Wong <darrick.wong@oracle.com>
Thu, 19 Mar 2020 15:48:47 +0000 (08:48 -0700)
We know the version is 3 if on a v5 file system.   For earlier file
systems formats we always upgrade the remaining v1 inodes to v2 and
thus only use v2 inodes.  Use the xfs_sb_version_has_large_dinode
helper to check if we deal with small or large dinodes, and thus
remove the need for the di_version field in struct icdinode.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Chandan Rajendra <chandanrlinux@gmail.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
fs/xfs/libxfs/xfs_inode_buf.c
fs/xfs/libxfs/xfs_inode_buf.h
fs/xfs/xfs_bmap_util.c
fs/xfs/xfs_inode.c
fs/xfs/xfs_inode_item.c
fs/xfs/xfs_ioctl.c
fs/xfs/xfs_iops.c
fs/xfs/xfs_itable.c
fs/xfs/xfs_log_recover.c

index 240d748403067eb4185647285051c66e33240806..39c5a6e24915ccc0ba875b595ef679c2c086c290 100644 (file)
@@ -194,16 +194,14 @@ xfs_inode_from_disk(
        struct xfs_icdinode     *to = &ip->i_d;
        struct inode            *inode = VFS_I(ip);
 
-
        /*
         * Convert v1 inodes immediately to v2 inode format as this is the
         * minimum inode version format we support in the rest of the code.
+        * They will also be unconditionally written back to disk as v2 inodes.
         */
-       to->di_version = from->di_version;
-       if (to->di_version == 1) {
+       if (unlikely(from->di_version == 1)) {
                set_nlink(inode, be16_to_cpu(from->di_onlink));
                to->di_projid = 0;
-               to->di_version = 2;
        } else {
                set_nlink(inode, be32_to_cpu(from->di_nlink));
                to->di_projid = (prid_t)be16_to_cpu(from->di_projid_hi) << 16 |
@@ -241,7 +239,7 @@ xfs_inode_from_disk(
        to->di_dmstate  = be16_to_cpu(from->di_dmstate);
        to->di_flags    = be16_to_cpu(from->di_flags);
 
-       if (to->di_version == 3) {
+       if (xfs_sb_version_has_v3inode(&ip->i_mount->m_sb)) {
                inode_set_iversion_queried(inode,
                                           be64_to_cpu(from->di_changecount));
                to->di_crtime.tv_sec = be32_to_cpu(from->di_crtime.t_sec);
@@ -263,7 +261,6 @@ xfs_inode_to_disk(
        to->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
        to->di_onlink = 0;
 
-       to->di_version = from->di_version;
        to->di_format = from->di_format;
        to->di_uid = cpu_to_be32(i_uid_read(inode));
        to->di_gid = cpu_to_be32(i_gid_read(inode));
@@ -292,7 +289,8 @@ xfs_inode_to_disk(
        to->di_dmstate = cpu_to_be16(from->di_dmstate);
        to->di_flags = cpu_to_be16(from->di_flags);
 
-       if (from->di_version == 3) {
+       if (xfs_sb_version_has_v3inode(&ip->i_mount->m_sb)) {
+               to->di_version = 3;
                to->di_changecount = cpu_to_be64(inode_peek_iversion(inode));
                to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.tv_sec);
                to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.tv_nsec);
@@ -304,6 +302,7 @@ xfs_inode_to_disk(
                uuid_copy(&to->di_uuid, &ip->i_mount->m_sb.sb_meta_uuid);
                to->di_flushiter = 0;
        } else {
+               to->di_version = 2;
                to->di_flushiter = cpu_to_be16(from->di_flushiter);
        }
 }
@@ -621,7 +620,6 @@ xfs_iread(
            xfs_sb_version_has_v3inode(&mp->m_sb) &&
            !(mp->m_flags & XFS_MOUNT_IKEEP)) {
                VFS_I(ip)->i_generation = prandom_u32();
-               ip->i_d.di_version = 3;
                return 0;
        }
 
@@ -663,7 +661,6 @@ xfs_iread(
                 * Partial initialisation of the in-core inode. Just the bits
                 * that xfs_ialloc won't overwrite or relies on being correct.
                 */
-               ip->i_d.di_version = dip->di_version;
                VFS_I(ip)->i_generation = be32_to_cpu(dip->di_gen);
                ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter);
 
@@ -677,7 +674,6 @@ xfs_iread(
                VFS_I(ip)->i_mode = 0;
        }
 
-       ASSERT(ip->i_d.di_version >= 2);
        ip->i_delayed_blks = 0;
 
        /*
index 66de5964045c33d65425ff7cac83b73a7c64a5f4..9b373dcf9e34d87553499eb74dac8e1f303bd385 100644 (file)
@@ -16,7 +16,6 @@ struct xfs_dinode;
  * format specific structures at the appropriate time.
  */
 struct xfs_icdinode {
-       int8_t          di_version;     /* inode version */
        int8_t          di_format;      /* format of di_c data */
        uint16_t        di_flushiter;   /* incremented on flush */
        uint32_t        di_projid;      /* owner's project id */
index 3df4d0af9f226d0573b635072e11d530d8ab3bb6..4f800f7fe88829c7644fef379badc298971d775d 100644 (file)
@@ -1449,12 +1449,12 @@ xfs_swap_extent_forks(
         * event of a crash. Set the owner change log flags now and leave the
         * bmbt scan as the last step.
         */
-       if (ip->i_d.di_version == 3 &&
-           ip->i_d.di_format == XFS_DINODE_FMT_BTREE)
-               (*target_log_flags) |= XFS_ILOG_DOWNER;
-       if (tip->i_d.di_version == 3 &&
-           tip->i_d.di_format == XFS_DINODE_FMT_BTREE)
-               (*src_log_flags) |= XFS_ILOG_DOWNER;
+       if (xfs_sb_version_has_v3inode(&ip->i_mount->m_sb)) {
+               if (ip->i_d.di_format == XFS_DINODE_FMT_BTREE)
+                       (*target_log_flags) |= XFS_ILOG_DOWNER;
+               if (tip->i_d.di_format == XFS_DINODE_FMT_BTREE)
+                       (*src_log_flags) |= XFS_ILOG_DOWNER;
+       }
 
        /*
         * Swap the data forks of the inodes
@@ -1489,7 +1489,7 @@ xfs_swap_extent_forks(
                (*src_log_flags) |= XFS_ILOG_DEXT;
                break;
        case XFS_DINODE_FMT_BTREE:
-               ASSERT(ip->i_d.di_version < 3 ||
+               ASSERT(!xfs_sb_version_has_v3inode(&ip->i_mount->m_sb) ||
                       (*src_log_flags & XFS_ILOG_DOWNER));
                (*src_log_flags) |= XFS_ILOG_DBROOT;
                break;
@@ -1501,7 +1501,7 @@ xfs_swap_extent_forks(
                break;
        case XFS_DINODE_FMT_BTREE:
                (*target_log_flags) |= XFS_ILOG_DBROOT;
-               ASSERT(tip->i_d.di_version < 3 ||
+               ASSERT(!xfs_sb_version_has_v3inode(&ip->i_mount->m_sb) ||
                       (*target_log_flags & XFS_ILOG_DOWNER));
                break;
        }
index dfb0a452a87dd74fb5f235a9eaa0dd17eb835773..14b922f2a6db71e0f612cf293393c62404f1ecfe 100644 (file)
@@ -801,15 +801,6 @@ xfs_ialloc(
                return error;
        ASSERT(ip != NULL);
        inode = VFS_I(ip);
-
-       /*
-        * We always convert v1 inodes to v2 now - we only support filesystems
-        * with >= v2 inode capability, so there is no reason for ever leaving
-        * an inode in v1 format.
-        */
-       if (ip->i_d.di_version == 1)
-               ip->i_d.di_version = 2;
-
        inode->i_mode = mode;
        set_nlink(inode, nlink);
        inode->i_uid = current_fsuid();
@@ -847,14 +838,13 @@ xfs_ialloc(
        ip->i_d.di_dmstate = 0;
        ip->i_d.di_flags = 0;
 
-       if (ip->i_d.di_version == 3) {
+       if (xfs_sb_version_has_v3inode(&mp->m_sb)) {
                inode_set_iversion(inode, 1);
                ip->i_d.di_flags2 = 0;
                ip->i_d.di_cowextsize = 0;
                ip->i_d.di_crtime = tv;
        }
 
-
        flags = XFS_ILOG_CORE;
        switch (mode & S_IFMT) {
        case S_IFIFO:
@@ -1115,7 +1105,6 @@ xfs_bumplink(
 {
        xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);
 
-       ASSERT(ip->i_d.di_version > 1);
        inc_nlink(VFS_I(ip));
        xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 }
@@ -3798,7 +3787,6 @@ xfs_iflush_int(
        ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
               ip->i_d.di_nextents > XFS_IFORK_MAXEXT(ip, XFS_DATA_FORK));
        ASSERT(iip != NULL && iip->ili_fields != 0);
-       ASSERT(ip->i_d.di_version > 1);
 
        /* set *dip = inode's place in the buffer */
        dip = xfs_buf_offset(bp, ip->i_imap.im_boffset);
@@ -3859,7 +3847,7 @@ xfs_iflush_int(
         * backwards compatibility with old kernels that predate logging all
         * inode changes.
         */
-       if (ip->i_d.di_version < 3)
+       if (!xfs_sb_version_has_v3inode(&mp->m_sb))
                ip->i_d.di_flushiter++;
 
        /* Check the inline fork data before we write out. */
index 451f9b6b2806324e56946672c3063c5edfbc75a8..4a3d13d4a022889cbb50819c2e7d4b08ee315064 100644 (file)
@@ -305,8 +305,6 @@ xfs_inode_to_log_dinode(
        struct inode            *inode = VFS_I(ip);
 
        to->di_magic = XFS_DINODE_MAGIC;
-
-       to->di_version = from->di_version;
        to->di_format = from->di_format;
        to->di_uid = i_uid_read(inode);
        to->di_gid = i_gid_read(inode);
@@ -339,7 +337,8 @@ xfs_inode_to_log_dinode(
        /* log a dummy value to ensure log structure is fully initialised */
        to->di_next_unlinked = NULLAGINO;
 
-       if (from->di_version == 3) {
+       if (xfs_sb_version_has_v3inode(&ip->i_mount->m_sb)) {
+               to->di_version = 3;
                to->di_changecount = inode_peek_iversion(inode);
                to->di_crtime.t_sec = from->di_crtime.tv_sec;
                to->di_crtime.t_nsec = from->di_crtime.tv_nsec;
@@ -351,6 +350,7 @@ xfs_inode_to_log_dinode(
                uuid_copy(&to->di_uuid, &ip->i_mount->m_sb.sb_meta_uuid);
                to->di_flushiter = 0;
        } else {
+               to->di_version = 2;
                to->di_flushiter = from->di_flushiter;
        }
 }
@@ -395,8 +395,6 @@ xfs_inode_item_format(
        struct xfs_log_iovec    *vecp = NULL;
        struct xfs_inode_log_format *ilf;
 
-       ASSERT(ip->i_d.di_version > 1);
-
        ilf = xlog_prepare_iovec(lv, &vecp, XLOG_REG_TYPE_IFORMAT);
        ilf->ilf_type = XFS_LI_INODE;
        ilf->ilf_ino = ip->i_ino;
index ad825ffa7e4c750d427f24d9d43ce837d8349e07..cdfb3cd9a25bdc523bca574926d5d9da30b10180 100644 (file)
@@ -1263,7 +1263,7 @@ xfs_ioctl_setattr_xflags(
 
        /* diflags2 only valid for v3 inodes. */
        di_flags2 = xfs_flags2diflags2(ip, fa->fsx_xflags);
-       if (di_flags2 && ip->i_d.di_version < 3)
+       if (di_flags2 && !xfs_sb_version_has_v3inode(&mp->m_sb))
                return -EINVAL;
 
        ip->i_d.di_flags = xfs_flags2diflags(ip, fa->fsx_xflags);
@@ -1601,7 +1601,6 @@ xfs_ioctl_setattr(
                        olddquot = xfs_qm_vop_chown(tp, ip,
                                                &ip->i_pdquot, pdqp);
                }
-               ASSERT(ip->i_d.di_version > 1);
                ip->i_d.di_projid = fa->fsx_projid;
        }
 
@@ -1614,7 +1613,7 @@ xfs_ioctl_setattr(
                ip->i_d.di_extsize = fa->fsx_extsize >> mp->m_sb.sb_blocklog;
        else
                ip->i_d.di_extsize = 0;
-       if (ip->i_d.di_version == 3 &&
+       if (xfs_sb_version_has_v3inode(&mp->m_sb) &&
            (ip->i_d.di_flags2 & XFS_DIFLAG2_COWEXTSIZE))
                ip->i_d.di_cowextsize = fa->fsx_cowextsize >>
                                mp->m_sb.sb_blocklog;
index 87093a05aad79e4a2774ab5f9ca5a677e39080f9..f7a99b3bbcf7a37e4ce4fba965b2c53a424c9b0a 100644 (file)
@@ -557,7 +557,7 @@ xfs_vn_getattr(
        stat->blocks =
                XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);
 
-       if (ip->i_d.di_version == 3) {
+       if (xfs_sb_version_has_v3inode(&mp->m_sb)) {
                if (request_mask & STATX_BTIME) {
                        stat->result_mask |= STATX_BTIME;
                        stat->btime = ip->i_d.di_crtime;
index d1066046988498d4b2172f49137153262b2622ce..ff2da28fed90e75751f378967fc123bfb90d8032 100644 (file)
@@ -110,7 +110,7 @@ xfs_bulkstat_one_int(
        buf->bs_forkoff = XFS_IFORK_BOFF(ip);
        buf->bs_version = XFS_BULKSTAT_VERSION_V5;
 
-       if (dic->di_version == 3) {
+       if (xfs_sb_version_has_v3inode(&mp->m_sb)) {
                if (dic->di_flags2 & XFS_DIFLAG2_COWEXTSIZE)
                        buf->bs_cowextsize_blks = dic->di_cowextsize;
        }
index 308cc5dcac14aa7e5983dd722a19a944ee8e0a26..11c3502b07b13c2b3f8ea03753b120a1b00cee9f 100644 (file)
@@ -2869,8 +2869,8 @@ xfs_recover_inode_owner_change(
                return -ENOMEM;
 
        /* instantiate the inode */
+       ASSERT(dip->di_version >= 3);
        xfs_inode_from_disk(ip, dip);
-       ASSERT(ip->i_d.di_version >= 3);
 
        error = xfs_iformat_fork(ip, dip);
        if (error)