From: Darrick J. Wong <djwong@xxxxxxxxxx>
Ideally, we'd put all the metadata inodes in one place if we could, so
that the metadata all stay reasonably close together instead of
spreading out over the disk. Furthermore, if the log is internal we'd
probably prefer to keep the metadata near the log. Therefore, disable
AGI rotoring for metadata inode allocations.
Signed-off-by: Darrick J. Wong <djwong@xxxxxxxxxx>
---
fs/xfs/libxfs/xfs_ialloc.c | 56 ++++++++++++++++++++++++++++++--------------
fs/xfs/libxfs/xfs_ialloc.h | 2 +-
fs/xfs/libxfs/xfs_imeta.c | 4 ++-
fs/xfs/scrub/tempfile.c | 2 +-
fs/xfs/xfs_inode.c | 4 ++-
fs/xfs/xfs_symlink.c | 2 +-
6 files changed, 45 insertions(+), 25 deletions(-)
diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
index d58d700ed8a8b..96d22f9698a7a 100644
--- a/fs/xfs/libxfs/xfs_ialloc.c
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -1799,6 +1799,37 @@ xfs_dialloc_try_ag(
return error;
}
+/*
+ * Pick an AG for the new inode.
+ *
+ * Directories, symlinks, and regular files frequently allocate at least one
+ * block, so factor that potential expansion when we examine whether an AG has
+ * enough space for file creation. Try to keep metadata files all in the same
+ * AG.
+ */
+static inline xfs_agnumber_t
+xfs_dialloc_pick_ag(
+ struct xfs_mount *mp,
+ struct xfs_inode *dp,
+ umode_t mode)
+{
+ xfs_agnumber_t start_agno;
+
+ if (!dp)
+ return 0;
+ if (xfs_is_metadir_inode(dp))
+ return 0;
+
+ if (S_ISDIR(mode))
+ return (atomic_inc_return(&mp->m_agirotor) - 1) % mp->m_maxagi;
+
+ start_agno = XFS_INO_TO_AGNO(mp, dp->i_ino);
+ if (start_agno >= mp->m_maxagi)
+ start_agno = 0;
+
+ return start_agno;
+}
+
/*
* Allocate an on-disk inode.
*
@@ -1810,34 +1841,23 @@ xfs_dialloc_try_ag(
int
xfs_dialloc(
struct xfs_trans **tpp,
- xfs_ino_t parent,
+ struct xfs_inode *dp,
umode_t mode,
xfs_ino_t *new_ino)
{
struct xfs_mount *mp = (*tpp)->t_mountp;
- xfs_agnumber_t agno;
- int error = 0;
- xfs_agnumber_t start_agno;
struct xfs_perag *pag;
struct xfs_ino_geometry *igeo = M_IGEO(mp);
+ xfs_ino_t ino = NULLFSINO;
+ xfs_ino_t parent = dp ? dp->i_ino : 0;
+ xfs_agnumber_t agno;
+ xfs_agnumber_t start_agno;
bool ok_alloc = true;
bool low_space = false;
int flags;
- xfs_ino_t ino = NULLFSINO;
+ int error = 0;
- /*
- * Directories, symlinks, and regular files frequently allocate at least
- * one block, so factor that potential expansion when we examine whether
- * an AG has enough space for file creation.
- */
- if (S_ISDIR(mode))
- start_agno = (atomic_inc_return(&mp->m_agirotor) - 1) %
- mp->m_maxagi;
- else {
- start_agno = XFS_INO_TO_AGNO(mp, parent);
- if (start_agno >= mp->m_maxagi)
- start_agno = 0;
- }
+ start_agno = xfs_dialloc_pick_ag(mp, dp, mode);
/*
* If we have already hit the ceiling of inode blocks then clear
diff --git a/fs/xfs/libxfs/xfs_ialloc.h b/fs/xfs/libxfs/xfs_ialloc.h
index f1412183bb44b..9bfe2d8d84b1d 100644
--- a/fs/xfs/libxfs/xfs_ialloc.h
+++ b/fs/xfs/libxfs/xfs_ialloc.h
@@ -37,7 +37,7 @@ xfs_make_iptr(struct xfs_mount *mp, struct xfs_buf *b, int o)
* Allocate an inode on disk. Mode is used to tell whether the new inode will
* need space, and whether it is a directory.
*/
-int xfs_dialloc(struct xfs_trans **tpp, xfs_ino_t parent, umode_t mode,
+int xfs_dialloc(struct xfs_trans **tpp, struct xfs_inode *dp, umode_t mode,
xfs_ino_t *new_ino);
int xfs_difree(struct xfs_trans *tp, struct xfs_perag *pag,
diff --git a/fs/xfs/libxfs/xfs_imeta.c b/fs/xfs/libxfs/xfs_imeta.c
index 5c5a801386a29..e3651c7bba2b3 100644
--- a/fs/xfs/libxfs/xfs_imeta.c
+++ b/fs/xfs/libxfs/xfs_imeta.c
@@ -230,7 +230,7 @@ xfs_imeta_sb_create(
return -EEXIST;
/* Create a new inode and set the sb pointer. */
- error = xfs_dialloc(&upd->tp, 0, mode, &ino);
+ error = xfs_dialloc(&upd->tp, NULL, mode, &ino);
if (error)
return error;
error = xfs_icreate(upd->tp, ino, &args, &upd->ip);
@@ -662,7 +662,7 @@ xfs_imeta_dir_create(
* entry pointing to them, but a directory also the "." entry
* pointing to itself.
*/
- error = xfs_dialloc(&upd->tp, upd->dp->i_ino, mode, &ino);
+ error = xfs_dialloc(&upd->tp, upd->dp, mode, &ino);
if (error)
return error;
error = xfs_icreate(upd->tp, ino, &args, &upd->ip);
diff --git a/fs/xfs/scrub/tempfile.c b/fs/xfs/scrub/tempfile.c
index 6d207559df989..db3a9a47e1f82 100644
--- a/fs/xfs/scrub/tempfile.c
+++ b/fs/xfs/scrub/tempfile.c
@@ -87,7 +87,7 @@ xrep_tempfile_create(
goto out_release_dquots;
/* Allocate inode, set up directory. */
- error = xfs_dialloc(&tp, dp->i_ino, mode, &ino);
+ error = xfs_dialloc(&tp, dp, mode, &ino);
if (error)
goto out_trans_cancel;
error = xfs_icreate(tp, ino, &args, &sc->tempip);
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 7f5ad390b6c7c..66aa8bbcb1045 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -768,7 +768,7 @@ xfs_create(
* entry pointing to them, but a directory also the "." entry
* pointing to itself.
*/
- error = xfs_dialloc(&tp, dp->i_ino, args->mode, &ino);
+ error = xfs_dialloc(&tp, dp, args->mode, &ino);
if (!error)
error = xfs_icreate(tp, ino, args, &du.ip);
if (error)
@@ -879,7 +879,7 @@ xfs_create_tmpfile(
if (error)
goto out_release_dquots;
- error = xfs_dialloc(&tp, dp->i_ino, args->mode, &ino);
+ error = xfs_dialloc(&tp, dp, args->mode, &ino);
if (!error)
error = xfs_icreate(tp, ino, args, &ip);
if (error)
diff --git a/fs/xfs/xfs_symlink.c b/fs/xfs/xfs_symlink.c
index e7936858dfcf5..973be5e8b14be 100644
--- a/fs/xfs/xfs_symlink.c
+++ b/fs/xfs/xfs_symlink.c
@@ -169,7 +169,7 @@ xfs_symlink(
/*
* Allocate an inode for the symlink.
*/
- error = xfs_dialloc(&tp, dp->i_ino, S_IFLNK, &ino);
+ error = xfs_dialloc(&tp, dp, S_IFLNK, &ino);
if (!error)
error = xfs_icreate(tp, ino, &args, &du.ip);
if (error)
