xfs: remove bitfield based superblock updates
authorDave Chinner <dchinner@redhat.com>
Wed, 21 Jan 2015 22:10:26 +0000 (09:10 +1100)
committerDave Chinner <david@fromorbit.com>
Wed, 21 Jan 2015 22:10:26 +0000 (09:10 +1100)
When we log changes to the superblock, we first have to write them
to the on-disk buffer, and then log that. Right now we have a
complex bitfield based arrangement to only write the modified field
to the buffer before we log it.

This used to be necessary as a performance optimisation because we
logged the superblock buffer in every extent or inode allocation or
freeing, and so performance was extremely important. We haven't done
this for years, however, ever since the lazy superblock counters
pulled the superblock logging out of the transaction commit
fast path.

Hence we have a bunch of complexity that is not necessary that makes
writing the in-core superblock to disk much more complex than it
needs to be. We only need to log the superblock now during
management operations (e.g. during mount, unmount or quota control
operations) so it is not a performance critical path anymore.

As such, remove the complex field based logging mechanism and
replace it with a simple conversion function similar to what we use
for all other on-disk structures.

This means we always log the entirity of the superblock, but again
because we rarely modify the superblock this is not an issue for log
bandwidth or CPU time. Indeed, if we do log the superblock
frequently, delayed logging will minimise the impact of this
overhead.

[Fixed gquota/pquota inode sharing regression noticed by bfoster.]

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
fs/xfs/libxfs/xfs_attr_leaf.c
fs/xfs/libxfs/xfs_bmap.c
fs/xfs/libxfs/xfs_sb.c
fs/xfs/libxfs/xfs_sb.h
fs/xfs/xfs_fsops.c
fs/xfs/xfs_mount.c
fs/xfs/xfs_mount.h
fs/xfs/xfs_qm.c
fs/xfs/xfs_qm.h
fs/xfs/xfs_qm_syscalls.c
fs/xfs/xfs_super.c

index 5d38e8b8a913091a1b74bfac0fb3ab609d86a9df..c9144221798f2af2da33f79686d9e6ad1d624f14 100644 (file)
@@ -403,7 +403,7 @@ xfs_sbversion_add_attr2(xfs_mount_t *mp, xfs_trans_t *tp)
                if (!xfs_sb_version_hasattr2(&mp->m_sb)) {
                        xfs_sb_version_addattr2(&mp->m_sb);
                        spin_unlock(&mp->m_sb_lock);
-                       xfs_mod_sb(tp, XFS_SB_VERSIONNUM | XFS_SB_FEATURES2);
+                       xfs_mod_sb(tp);
                } else
                        spin_unlock(&mp->m_sb_lock);
        }
index b5eb4743f75a032c250662f5f8dcfe510db541bf..8c39cc852e4bd30431828c086fcaa46916c7d14b 100644 (file)
@@ -1221,22 +1221,20 @@ xfs_bmap_add_attrfork(
                goto bmap_cancel;
        if (!xfs_sb_version_hasattr(&mp->m_sb) ||
           (!xfs_sb_version_hasattr2(&mp->m_sb) && version == 2)) {
-               __int64_t sbfields = 0;
+               bool mod_sb = false;
 
                spin_lock(&mp->m_sb_lock);
                if (!xfs_sb_version_hasattr(&mp->m_sb)) {
                        xfs_sb_version_addattr(&mp->m_sb);
-                       sbfields |= XFS_SB_VERSIONNUM;
+                       mod_sb = true;
                }
                if (!xfs_sb_version_hasattr2(&mp->m_sb) && version == 2) {
                        xfs_sb_version_addattr2(&mp->m_sb);
-                       sbfields |= (XFS_SB_VERSIONNUM | XFS_SB_FEATURES2);
+                       mod_sb = true;
                }
-               if (sbfields) {
-                       spin_unlock(&mp->m_sb_lock);
-                       xfs_mod_sb(tp, sbfields);
-               } else
-                       spin_unlock(&mp->m_sb_lock);
+               spin_unlock(&mp->m_sb_lock);
+               if (mod_sb)
+                       xfs_mod_sb(tp);
        }
 
        error = xfs_bmap_finish(&tp, &flist, &committed);
index 752915fa775a5b53e7ec12c5676d469dd7c7069f..115a7cd3a6fbefc1f1c45ff6881d891ddc274ac5 100644 (file)
  * Physical superblock buffer manipulations. Shared with libxfs in userspace.
  */
 
-static const struct {
-       short offset;
-       short type;     /* 0 = integer
-                        * 1 = binary / string (no translation)
-                        */
-} xfs_sb_info[] = {
-       { offsetof(xfs_sb_t, sb_magicnum),      0 },
-       { offsetof(xfs_sb_t, sb_blocksize),     0 },
-       { offsetof(xfs_sb_t, sb_dblocks),       0 },
-       { offsetof(xfs_sb_t, sb_rblocks),       0 },
-       { offsetof(xfs_sb_t, sb_rextents),      0 },
-       { offsetof(xfs_sb_t, sb_uuid),          1 },
-       { offsetof(xfs_sb_t, sb_logstart),      0 },
-       { offsetof(xfs_sb_t, sb_rootino),       0 },
-       { offsetof(xfs_sb_t, sb_rbmino),        0 },
-       { offsetof(xfs_sb_t, sb_rsumino),       0 },
-       { offsetof(xfs_sb_t, sb_rextsize),      0 },
-       { offsetof(xfs_sb_t, sb_agblocks),      0 },
-       { offsetof(xfs_sb_t, sb_agcount),       0 },
-       { offsetof(xfs_sb_t, sb_rbmblocks),     0 },
-       { offsetof(xfs_sb_t, sb_logblocks),     0 },
-       { offsetof(xfs_sb_t, sb_versionnum),    0 },
-       { offsetof(xfs_sb_t, sb_sectsize),      0 },
-       { offsetof(xfs_sb_t, sb_inodesize),     0 },
-       { offsetof(xfs_sb_t, sb_inopblock),     0 },
-       { offsetof(xfs_sb_t, sb_fname[0]),      1 },
-       { offsetof(xfs_sb_t, sb_blocklog),      0 },
-       { offsetof(xfs_sb_t, sb_sectlog),       0 },
-       { offsetof(xfs_sb_t, sb_inodelog),      0 },
-       { offsetof(xfs_sb_t, sb_inopblog),      0 },
-       { offsetof(xfs_sb_t, sb_agblklog),      0 },
-       { offsetof(xfs_sb_t, sb_rextslog),      0 },
-       { offsetof(xfs_sb_t, sb_inprogress),    0 },
-       { offsetof(xfs_sb_t, sb_imax_pct),      0 },
-       { offsetof(xfs_sb_t, sb_icount),        0 },
-       { offsetof(xfs_sb_t, sb_ifree),         0 },
-       { offsetof(xfs_sb_t, sb_fdblocks),      0 },
-       { offsetof(xfs_sb_t, sb_frextents),     0 },
-       { offsetof(xfs_sb_t, sb_uquotino),      0 },
-       { offsetof(xfs_sb_t, sb_gquotino),      0 },
-       { offsetof(xfs_sb_t, sb_qflags),        0 },
-       { offsetof(xfs_sb_t, sb_flags),         0 },
-       { offsetof(xfs_sb_t, sb_shared_vn),     0 },
-       { offsetof(xfs_sb_t, sb_inoalignmt),    0 },
-       { offsetof(xfs_sb_t, sb_unit),          0 },
-       { offsetof(xfs_sb_t, sb_width),         0 },
-       { offsetof(xfs_sb_t, sb_dirblklog),     0 },
-       { offsetof(xfs_sb_t, sb_logsectlog),    0 },
-       { offsetof(xfs_sb_t, sb_logsectsize),   0 },
-       { offsetof(xfs_sb_t, sb_logsunit),      0 },
-       { offsetof(xfs_sb_t, sb_features2),     0 },
-       { offsetof(xfs_sb_t, sb_bad_features2), 0 },
-       { offsetof(xfs_sb_t, sb_features_compat),       0 },
-       { offsetof(xfs_sb_t, sb_features_ro_compat),    0 },
-       { offsetof(xfs_sb_t, sb_features_incompat),     0 },
-       { offsetof(xfs_sb_t, sb_features_log_incompat), 0 },
-       { offsetof(xfs_sb_t, sb_crc),           0 },
-       { offsetof(xfs_sb_t, sb_pad),           0 },
-       { offsetof(xfs_sb_t, sb_pquotino),      0 },
-       { offsetof(xfs_sb_t, sb_lsn),           0 },
-       { sizeof(xfs_sb_t),                     0 }
-};
-
 /*
  * Reference counting access wrappers to the perag structures.
  * Because we never free per-ag structures, the only thing we
@@ -461,58 +398,49 @@ xfs_sb_from_disk(
        __xfs_sb_from_disk(to, from, true);
 }
 
-static inline void
+static void
 xfs_sb_quota_to_disk(
-       xfs_dsb_t       *to,
-       xfs_sb_t        *from,
-       __int64_t       *fields)
+       struct xfs_dsb  *to,
+       struct xfs_sb   *from)
 {
        __uint16_t      qflags = from->sb_qflags;
 
+       to->sb_uquotino = cpu_to_be64(from->sb_uquotino);
+       if (xfs_sb_version_has_pquotino(from)) {
+               to->sb_qflags = cpu_to_be16(from->sb_qflags);
+               to->sb_gquotino = cpu_to_be64(from->sb_gquotino);
+               to->sb_pquotino = cpu_to_be64(from->sb_pquotino);
+               return;
+       }
+
        /*
-        * We need to do these manipilations only if we are working
-        * with an older version of on-disk superblock.
+        * The in-core version of sb_qflags do not have XFS_OQUOTA_*
+        * flags, whereas the on-disk version does.  So, convert incore
+        * XFS_{PG}QUOTA_* flags to on-disk XFS_OQUOTA_* flags.
         */
-       if (xfs_sb_version_has_pquotino(from))
-               return;
+       qflags &= ~(XFS_PQUOTA_ENFD | XFS_PQUOTA_CHKD |
+                       XFS_GQUOTA_ENFD | XFS_GQUOTA_CHKD);
 
-       if (*fields & XFS_SB_QFLAGS) {
-               /*
-                * The in-core version of sb_qflags do not have
-                * XFS_OQUOTA_* flags, whereas the on-disk version
-                * does.  So, convert incore XFS_{PG}QUOTA_* flags
-                * to on-disk XFS_OQUOTA_* flags.
-                */
-               qflags &= ~(XFS_PQUOTA_ENFD | XFS_PQUOTA_CHKD |
-                               XFS_GQUOTA_ENFD | XFS_GQUOTA_CHKD);
-
-               if (from->sb_qflags &
-                               (XFS_PQUOTA_ENFD | XFS_GQUOTA_ENFD))
-                       qflags |= XFS_OQUOTA_ENFD;
-               if (from->sb_qflags &
-                               (XFS_PQUOTA_CHKD | XFS_GQUOTA_CHKD))
-                       qflags |= XFS_OQUOTA_CHKD;
-               to->sb_qflags = cpu_to_be16(qflags);
-               *fields &= ~XFS_SB_QFLAGS;
-       }
+       if (from->sb_qflags &
+                       (XFS_PQUOTA_ENFD | XFS_GQUOTA_ENFD))
+               qflags |= XFS_OQUOTA_ENFD;
+       if (from->sb_qflags &
+                       (XFS_PQUOTA_CHKD | XFS_GQUOTA_CHKD))
+               qflags |= XFS_OQUOTA_CHKD;
+       to->sb_qflags = cpu_to_be16(qflags);
 
        /*
-        * GQUOTINO and PQUOTINO cannot be used together in versions of
-        * superblock that do not have pquotino. from->sb_flags tells us which
-        * quota is active and should be copied to disk. If neither are active,
-        * make sure we write NULLFSINO to the sb_gquotino field as a quota
-        * inode value of "0" is invalid when the XFS_SB_VERSION_QUOTA feature
-        * bit is set.
+        * GQUOTINO and PQUOTINO cannot be used together in versions
+        * of superblock that do not have pquotino. from->sb_flags
+        * tells us which quota is active and should be copied to
+        * disk. If neither are active, we should NULL the inode.
         *
-        * Note that we don't need to handle the sb_uquotino or sb_pquotino here
-        * as they do not require any translation. Hence the main sb field loop
-        * will write them appropriately from the in-core superblock.
+        * In all cases, the separate pquotino must remain 0 because it
+        * it beyond the "end" of the valid non-pquotino superblock.
         */
-       if ((*fields & XFS_SB_GQUOTINO) &&
-                               (from->sb_qflags & XFS_GQUOTA_ACCT))
+       if (from->sb_qflags & XFS_GQUOTA_ACCT)
                to->sb_gquotino = cpu_to_be64(from->sb_gquotino);
-       else if ((*fields & XFS_SB_PQUOTINO) &&
-                               (from->sb_qflags & XFS_PQUOTA_ACCT))
+       else if (from->sb_qflags & XFS_PQUOTA_ACCT)
                to->sb_gquotino = cpu_to_be64(from->sb_pquotino);
        else {
                /*
@@ -526,63 +454,72 @@ xfs_sb_quota_to_disk(
                        to->sb_gquotino = cpu_to_be64(NULLFSINO);
        }
 
-       *fields &= ~(XFS_SB_PQUOTINO | XFS_SB_GQUOTINO);
+       to->sb_pquotino = 0;
 }
 
-/*
- * Copy in core superblock to ondisk one.
- *
- * The fields argument is mask of superblock fields to copy.
- */
 void
 xfs_sb_to_disk(
-       xfs_dsb_t       *to,
-       xfs_sb_t        *from,
-       __int64_t       fields)
+       struct xfs_dsb  *to,
+       struct xfs_sb   *from)
 {
-       xfs_caddr_t     to_ptr = (xfs_caddr_t)to;
-       xfs_caddr_t     from_ptr = (xfs_caddr_t)from;
-       xfs_sb_field_t  f;
-       int             first;
-       int             size;
-
-       ASSERT(fields);
-       if (!fields)
-               return;
+       xfs_sb_quota_to_disk(to, from);
+
+       to->sb_magicnum = cpu_to_be32(from->sb_magicnum);
+       to->sb_blocksize = cpu_to_be32(from->sb_blocksize);
+       to->sb_dblocks = cpu_to_be64(from->sb_dblocks);
+       to->sb_rblocks = cpu_to_be64(from->sb_rblocks);
+       to->sb_rextents = cpu_to_be64(from->sb_rextents);
+       memcpy(&to->sb_uuid, &from->sb_uuid, sizeof(to->sb_uuid));
+       to->sb_logstart = cpu_to_be64(from->sb_logstart);
+       to->sb_rootino = cpu_to_be64(from->sb_rootino);
+       to->sb_rbmino = cpu_to_be64(from->sb_rbmino);
+       to->sb_rsumino = cpu_to_be64(from->sb_rsumino);
+       to->sb_rextsize = cpu_to_be32(from->sb_rextsize);
+       to->sb_agblocks = cpu_to_be32(from->sb_agblocks);
+       to->sb_agcount = cpu_to_be32(from->sb_agcount);
+       to->sb_rbmblocks = cpu_to_be32(from->sb_rbmblocks);
+       to->sb_logblocks = cpu_to_be32(from->sb_logblocks);
+       to->sb_versionnum = cpu_to_be16(from->sb_versionnum);
+       to->sb_sectsize = cpu_to_be16(from->sb_sectsize);
+       to->sb_inodesize = cpu_to_be16(from->sb_inodesize);
+       to->sb_inopblock = cpu_to_be16(from->sb_inopblock);
+       memcpy(&to->sb_fname, &from->sb_fname, sizeof(to->sb_fname));
+       to->sb_blocklog = from->sb_blocklog;
+       to->sb_sectlog = from->sb_sectlog;
+       to->sb_inodelog = from->sb_inodelog;
+       to->sb_inopblog = from->sb_inopblog;
+       to->sb_agblklog = from->sb_agblklog;
+       to->sb_rextslog = from->sb_rextslog;
+       to->sb_inprogress = from->sb_inprogress;
+       to->sb_imax_pct = from->sb_imax_pct;
+       to->sb_icount = cpu_to_be64(from->sb_icount);
+       to->sb_ifree = cpu_to_be64(from->sb_ifree);
+       to->sb_fdblocks = cpu_to_be64(from->sb_fdblocks);
+       to->sb_frextents = cpu_to_be64(from->sb_frextents);
 
-       /* We should never write the crc here, it's updated in the IO path */
-       fields &= ~XFS_SB_CRC;
-
-       xfs_sb_quota_to_disk(to, from, &fields);
-       while (fields) {
-               f = (xfs_sb_field_t)xfs_lowbit64((__uint64_t)fields);
-               first = xfs_sb_info[f].offset;
-               size = xfs_sb_info[f + 1].offset - first;
-
-               ASSERT(xfs_sb_info[f].type == 0 || xfs_sb_info[f].type == 1);
-
-               if (size == 1 || xfs_sb_info[f].type == 1) {
-                       memcpy(to_ptr + first, from_ptr + first, size);
-               } else {
-                       switch (size) {
-                       case 2:
-                               *(__be16 *)(to_ptr + first) =
-                                     cpu_to_be16(*(__u16 *)(from_ptr + first));
-                               break;
-                       case 4:
-                               *(__be32 *)(to_ptr + first) =
-                                     cpu_to_be32(*(__u32 *)(from_ptr + first));
-                               break;
-                       case 8:
-                               *(__be64 *)(to_ptr + first) =
-                                     cpu_to_be64(*(__u64 *)(from_ptr + first));
-                               break;
-                       default:
-                               ASSERT(0);
-                       }
-               }
 
-               fields &= ~(1LL << f);
+       to->sb_flags = from->sb_flags;
+       to->sb_shared_vn = from->sb_shared_vn;
+       to->sb_inoalignmt = cpu_to_be32(from->sb_inoalignmt);
+       to->sb_unit = cpu_to_be32(from->sb_unit);
+       to->sb_width = cpu_to_be32(from->sb_width);
+       to->sb_dirblklog = from->sb_dirblklog;
+       to->sb_logsectlog = from->sb_logsectlog;
+       to->sb_logsectsize = cpu_to_be16(from->sb_logsectsize);
+       to->sb_logsunit = cpu_to_be32(from->sb_logsunit);
+       to->sb_features2 = cpu_to_be32(from->sb_features2);
+       to->sb_bad_features2 = cpu_to_be32(from->sb_bad_features2);
+
+       if (xfs_sb_version_hascrc(from)) {
+               to->sb_features_compat = cpu_to_be32(from->sb_features_compat);
+               to->sb_features_ro_compat =
+                               cpu_to_be32(from->sb_features_ro_compat);
+               to->sb_features_incompat =
+                               cpu_to_be32(from->sb_features_incompat);
+               to->sb_features_log_incompat =
+                               cpu_to_be32(from->sb_features_log_incompat);
+               to->sb_pad = 0;
+               to->sb_lsn = cpu_to_be64(from->sb_lsn);
        }
 }
 
@@ -823,35 +760,13 @@ xfs_initialize_perag_data(
  * access.
  */
 void
-xfs_mod_sb(xfs_trans_t *tp, __int64_t fields)
+xfs_mod_sb(
+       struct xfs_trans        *tp)
 {
-       xfs_buf_t       *bp;
-       int             first;
-       int             last;
-       xfs_mount_t     *mp;
-       xfs_sb_field_t  f;
-
-       ASSERT(fields);
-       if (!fields)
-               return;
-       mp = tp->t_mountp;
-       bp = xfs_trans_getsb(tp, mp, 0);
-       first = sizeof(xfs_sb_t);
-       last = 0;
-
-       /* translate/copy */
-
-       xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb, fields);
-
-       /* find modified range */
-       f = (xfs_sb_field_t)xfs_highbit64((__uint64_t)fields);
-       ASSERT((1LL << f) & XFS_SB_MOD_BITS);
-       last = xfs_sb_info[f + 1].offset - 1;
-
-       f = (xfs_sb_field_t)xfs_lowbit64((__uint64_t)fields);
-       ASSERT((1LL << f) & XFS_SB_MOD_BITS);
-       first = xfs_sb_info[f].offset;
+       struct xfs_mount        *mp = tp->t_mountp;
+       struct xfs_buf          *bp = xfs_trans_getsb(tp, mp, 0);
 
+       xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb);
        xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SB_BUF);
-       xfs_trans_log_buf(tp, bp, first, last);
+       xfs_trans_log_buf(tp, bp, 0, sizeof(struct xfs_dsb));
 }
index 8eb1c54bafbf987e8b9bfd98abc858cd081cedd1..e193caa42988cbaa9c3ee261d2af7e0e6573e126 100644 (file)
@@ -27,11 +27,11 @@ extern struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *, xfs_agnumber_t,
 extern void    xfs_perag_put(struct xfs_perag *pag);
 extern int     xfs_initialize_perag_data(struct xfs_mount *, xfs_agnumber_t);
 
-extern void    xfs_sb_calc_crc(struct xfs_buf  *);
-extern void    xfs_mod_sb(struct xfs_trans *, __int64_t);
-extern void    xfs_sb_mount_common(struct xfs_mount *, struct xfs_sb *);
-extern void    xfs_sb_from_disk(struct xfs_sb *, struct xfs_dsb *);
-extern void    xfs_sb_to_disk(struct xfs_dsb *, struct xfs_sb *, __int64_t);
+extern void    xfs_sb_calc_crc(struct xfs_buf *bp);
+extern void    xfs_mod_sb(struct xfs_trans *tp);
+extern void    xfs_sb_mount_common(struct xfs_mount *mp, struct xfs_sb *sbp);
+extern void    xfs_sb_from_disk(struct xfs_sb *to, struct xfs_dsb *from);
+extern void    xfs_sb_to_disk(struct xfs_dsb *to, struct xfs_sb *from);
 extern void    xfs_sb_quota_from_disk(struct xfs_sb *sbp);
 
 #endif /* __XFS_SB_H__ */
index fdc64220fcb0fe02f0baa8b42110df8d4acdd3b1..82af857405af73150e63f600f9a3d3b279e56dfc 100644 (file)
@@ -541,7 +541,7 @@ xfs_growfs_data_private(
                        saved_error = error;
                        continue;
                }
-               xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb, XFS_SB_ALL_BITS);
+               xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb);
 
                error = xfs_bwrite(bp);
                xfs_buf_relse(bp);
@@ -780,9 +780,7 @@ xfs_fs_log_dummy(
                xfs_trans_cancel(tp, 0);
                return error;
        }
-
-       /* log the UUID because it is an unchanging field */
-       xfs_mod_sb(tp, XFS_SB_UUID);
+       xfs_mod_sb(tp);
        xfs_trans_set_sync(tp);
        return xfs_trans_commit(tp, 0);
 }
index d3d38836f87fad07b3c815dd33ae234ddf785b03..2953d46be24991779a534e44f824327b0720d723 100644 (file)
@@ -613,7 +613,7 @@ xfs_mount_reset_sbqflags(
                return error;
        }
 
-       xfs_mod_sb(tp, XFS_SB_QFLAGS);
+       xfs_mod_sb(tp);
        return xfs_trans_commit(tp, 0);
 }
 
@@ -896,7 +896,7 @@ xfs_mountfs(
         * perform the update e.g. for the root filesystem.
         */
        if (mp->m_update_flags && !(mp->m_flags & XFS_MOUNT_RDONLY)) {
-               error = xfs_mount_log_sb(mp, mp->m_update_flags);
+               error = xfs_mount_log_sb(mp);
                if (error) {
                        xfs_warn(mp, "failed to write sb changes");
                        goto out_rtunmount;
@@ -1126,7 +1126,7 @@ xfs_log_sbcount(xfs_mount_t *mp)
                return error;
        }
 
-       xfs_mod_sb(tp, XFS_SB_IFREE | XFS_SB_ICOUNT | XFS_SB_FDBLOCKS);
+       xfs_mod_sb(tp);
        xfs_trans_set_sync(tp);
        error = xfs_trans_commit(tp, 0);
        return error;
@@ -1429,15 +1429,10 @@ xfs_freesb(
  */
 int
 xfs_mount_log_sb(
-       xfs_mount_t     *mp,
-       __int64_t       fields)
+       struct xfs_mount        *mp)
 {
-       xfs_trans_t     *tp;
-       int             error;
-
-       ASSERT(fields & (XFS_SB_UNIT | XFS_SB_WIDTH | XFS_SB_UUID |
-                        XFS_SB_FEATURES2 | XFS_SB_BAD_FEATURES2 |
-                        XFS_SB_VERSIONNUM));
+       struct xfs_trans        *tp;
+       int                     error;
 
        tp = xfs_trans_alloc(mp, XFS_TRANS_SB_UNIT);
        error = xfs_trans_reserve(tp, &M_RES(mp)->tr_sb, 0, 0);
@@ -1445,9 +1440,8 @@ xfs_mount_log_sb(
                xfs_trans_cancel(tp, 0);
                return error;
        }
-       xfs_mod_sb(tp, fields);
-       error = xfs_trans_commit(tp, 0);
-       return error;
+       xfs_mod_sb(tp);
+       return xfs_trans_commit(tp, 0);
 }
 
 /*
index 22ccf69d4d3c27f6732c5c08ec41fb4f9f001c4c..28b341bf1555a21292c24051e77cd7b61b8ae7f8 100644 (file)
@@ -378,7 +378,7 @@ extern void xfs_unmountfs(xfs_mount_t *);
 extern int     xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int64_t, int);
 extern int     xfs_mod_incore_sb_batch(xfs_mount_t *, xfs_mod_sb_t *,
                        uint, int);
-extern int     xfs_mount_log_sb(xfs_mount_t *, __int64_t);
+extern int     xfs_mount_log_sb(xfs_mount_t *);
 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
 extern int     xfs_readsb(xfs_mount_t *, int);
 extern void    xfs_freesb(xfs_mount_t *);
index 79fb19dd9c833de5fefc8b44a332f6e47250e13f..c815a8060b59fcadedd642add87bb2f39447b4c9 100644 (file)
@@ -714,7 +714,6 @@ STATIC int
 xfs_qm_qino_alloc(
        xfs_mount_t     *mp,
        xfs_inode_t     **ip,
-       __int64_t       sbfields,
        uint            flags)
 {
        xfs_trans_t     *tp;
@@ -777,11 +776,6 @@ xfs_qm_qino_alloc(
        spin_lock(&mp->m_sb_lock);
        if (flags & XFS_QMOPT_SBVERSION) {
                ASSERT(!xfs_sb_version_hasquota(&mp->m_sb));
-               ASSERT((sbfields & (XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO |
-                       XFS_SB_GQUOTINO | XFS_SB_PQUOTINO | XFS_SB_QFLAGS)) ==
-                               (XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO |
-                                XFS_SB_GQUOTINO | XFS_SB_PQUOTINO |
-                                XFS_SB_QFLAGS));
 
                xfs_sb_version_addquota(&mp->m_sb);
                mp->m_sb.sb_uquotino = NULLFSINO;
@@ -798,7 +792,7 @@ xfs_qm_qino_alloc(
        else
                mp->m_sb.sb_pquotino = (*ip)->i_ino;
        spin_unlock(&mp->m_sb_lock);
-       xfs_mod_sb(tp, sbfields);
+       xfs_mod_sb(tp);
 
        if ((error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES))) {
                xfs_alert(mp, "%s failed (error %d)!", __func__, error);
@@ -1451,7 +1445,7 @@ xfs_qm_mount_quotas(
        spin_unlock(&mp->m_sb_lock);
 
        if (sbf != (mp->m_qflags & XFS_MOUNT_QUOTA_ALL)) {
-               if (xfs_qm_write_sb_changes(mp, XFS_SB_QFLAGS)) {
+               if (xfs_qm_write_sb_changes(mp)) {
                        /*
                         * We could only have been turning quotas off.
                         * We aren't in very good shape actually because
@@ -1482,7 +1476,6 @@ xfs_qm_init_quotainos(
        struct xfs_inode        *gip = NULL;
        struct xfs_inode        *pip = NULL;
        int                     error;
-       __int64_t               sbflags = 0;
        uint                    flags = 0;
 
        ASSERT(mp->m_quotainfo);
@@ -1517,9 +1510,6 @@ xfs_qm_init_quotainos(
                }
        } else {
                flags |= XFS_QMOPT_SBVERSION;
-               sbflags |= (XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO |
-                           XFS_SB_GQUOTINO | XFS_SB_PQUOTINO |
-                           XFS_SB_QFLAGS);
        }
 
        /*
@@ -1530,7 +1520,6 @@ xfs_qm_init_quotainos(
         */
        if (XFS_IS_UQUOTA_ON(mp) && uip == NULL) {
                error = xfs_qm_qino_alloc(mp, &uip,
-                                             sbflags | XFS_SB_UQUOTINO,
                                              flags | XFS_QMOPT_UQUOTA);
                if (error)
                        goto error_rele;
@@ -1539,7 +1528,6 @@ xfs_qm_init_quotainos(
        }
        if (XFS_IS_GQUOTA_ON(mp) && gip == NULL) {
                error = xfs_qm_qino_alloc(mp, &gip,
-                                         sbflags | XFS_SB_GQUOTINO,
                                          flags | XFS_QMOPT_GQUOTA);
                if (error)
                        goto error_rele;
@@ -1548,7 +1536,6 @@ xfs_qm_init_quotainos(
        }
        if (XFS_IS_PQUOTA_ON(mp) && pip == NULL) {
                error = xfs_qm_qino_alloc(mp, &pip,
-                                         sbflags | XFS_SB_PQUOTINO,
                                          flags | XFS_QMOPT_PQUOTA);
                if (error)
                        goto error_rele;
@@ -1593,8 +1580,7 @@ xfs_qm_dqfree_one(
  */
 int
 xfs_qm_write_sb_changes(
-       xfs_mount_t     *mp,
-       __int64_t       flags)
+       struct xfs_mount *mp)
 {
        xfs_trans_t     *tp;
        int             error;
@@ -1606,10 +1592,8 @@ xfs_qm_write_sb_changes(
                return error;
        }
 
-       xfs_mod_sb(tp, flags);
-       error = xfs_trans_commit(tp, 0);
-
-       return error;
+       xfs_mod_sb(tp);
+       return xfs_trans_commit(tp, 0);
 }
 
 
index 3a07a937e232a7e51bf089f981a664f782f5a75f..bddd23fda9ce3cfdf6a4cbca6d619e54659959e8 100644 (file)
@@ -157,7 +157,7 @@ struct xfs_dquot_acct {
 #define XFS_QM_RTBWARNLIMIT    5
 
 extern void            xfs_qm_destroy_quotainfo(struct xfs_mount *);
-extern int             xfs_qm_write_sb_changes(struct xfs_mount *, __int64_t);
+extern int             xfs_qm_write_sb_changes(struct xfs_mount *);
 
 /* dquot stuff */
 extern void            xfs_qm_dqpurge_all(struct xfs_mount *, uint);
index 74fca68e43b6b4e98a0d203204d4ce2f646ff8fa..8d7e5f0688039c968d183131752dc52fa1b871d6 100644 (file)
@@ -92,8 +92,7 @@ xfs_qm_scall_quotaoff(
                mutex_unlock(&q->qi_quotaofflock);
 
                /* XXX what to do if error ? Revert back to old vals incore ? */
-               error = xfs_qm_write_sb_changes(mp, XFS_SB_QFLAGS);
-               return error;
+               return xfs_qm_write_sb_changes(mp);
        }
 
        dqtype = 0;
@@ -314,7 +313,6 @@ xfs_qm_scall_quotaon(
 {
        int             error;
        uint            qf;
-       __int64_t       sbflags;
 
        flags &= (XFS_ALL_QUOTA_ACCT | XFS_ALL_QUOTA_ENFD);
        /*
@@ -322,8 +320,6 @@ xfs_qm_scall_quotaon(
         */
        flags &= ~(XFS_ALL_QUOTA_ACCT);
 
-       sbflags = 0;
-
        if (flags == 0) {
                xfs_debug(mp, "%s: zero flags, m_qflags=%x",
                        __func__, mp->m_qflags);
@@ -370,11 +366,10 @@ xfs_qm_scall_quotaon(
        /*
         * There's nothing to change if it's the same.
         */
-       if ((qf & flags) == flags && sbflags == 0)
+       if ((qf & flags) == flags)
                return -EEXIST;
-       sbflags |= XFS_SB_QFLAGS;
 
-       if ((error = xfs_qm_write_sb_changes(mp, sbflags)))
+       if ((error = xfs_qm_write_sb_changes(mp)))
                return error;
        /*
         * If we aren't trying to switch on quota enforcement, we are done.
@@ -801,7 +796,7 @@ xfs_qm_log_quotaoff(
        mp->m_sb.sb_qflags = (mp->m_qflags & ~(flags)) & XFS_MOUNT_QUOTA_ALL;
        spin_unlock(&mp->m_sb_lock);
 
-       xfs_mod_sb(tp, XFS_SB_QFLAGS);
+       xfs_mod_sb(tp);
 
        /*
         * We have to make sure that the transaction is secure on disk before we
index 19cbda196369021cf48fb1f7cab29b866d4ba16c..6fb298963d1b78ff9bc28590aa28d21e94dee229 100644 (file)
@@ -1258,7 +1258,7 @@ xfs_fs_remount(
                 * might have some superblock changes to update.
                 */
                if (mp->m_update_flags) {
-                       error = xfs_mount_log_sb(mp, mp->m_update_flags);
+                       error = xfs_mount_log_sb(mp);
                        if (error) {
                                xfs_warn(mp, "failed to write sb changes");
                                return error;