[PATCH 0/12 v3] Moving i_dquot out of struct inode

[Jan Kara]

Hello,

this patch set moves i_dquot array from struct inode into filesystem private
part of the inode. Thus filesystems which don't need it save 2 pointers in
their inodes (would be 3 after we add project quota support into generic
quota).

The patch series also contains a change to quotactl so that each filesystem
can set quota types it supports. This is in the end unrelated change
(originally it was necessary for i_dquot moving but in the end I changed
things so that it's not anymore). I can move that into a separate series
but I was somewhat reluctant to do that since that would mean another 6
one-line patches to the same files we are changing here...

If people like the patches, I will queue this series into my tree for
the next merge window. For that I'd prefer to get acks from affected fs
maintainers (the changes are pretty trivial and I don't feel it's a must but
still I'd prefer fs maintainers to ack they are aware of the changes).

Honza

Changes since v1:
* Inode field names are now named enum
* Quota type masks now have names like QTYPE_MASK_{USR|GRP|PRJ} instead of
opencoding shifts.

Changes since v2:
* Use ->get_dquots callback instead of inode fields framework
* rebased on Linus' tree as of 3.18-rc1 + something.

[Jan Kara]

Currently all filesystems supporting VFS quota support user and group
quotas. With introduction of project quotas this is going to change so
make sure filesystem isn't called for quota type it doesn't support by
introduction of a bitmask determining which quota types each filesystem
supports.

Signed-off-by: Jan Kara <***@suse.cz>
---
fs/quota/quota.c | 13 +++++++++++--
fs/super.c | 7 +++++++
include/linux/quota.h | 6 ++++++
3 files changed, 24 insertions(+), 2 deletions(-)

diff --git a/fs/quota/quota.c b/fs/quota/quota.c
index 75621649dbd7..0f28eac6e638 100644
--- a/fs/quota/quota.c
+++ b/fs/quota/quota.c
@@ -47,8 +47,11 @@ static int check_quotactl_permission(struct super_block *sb, int type, int cmd,

static void quota_sync_one(struct super_block *sb, void *arg)
{
- if (sb->s_qcop && sb->s_qcop->quota_sync)
- sb->s_qcop->quota_sync(sb, *(int *)arg);
+ int type = *(int *)arg;
+
+ if (sb->s_qcop && sb->s_qcop->quota_sync &&
+ (sb->s_dquot.allowed_types & (1 << type)))
+ sb->s_qcop->quota_sync(sb, type);
}

static int quota_sync_all(int type)
@@ -297,8 +300,14 @@ static int do_quotactl(struct super_block *sb, int type, int cmd, qid_t id,

if (type >= (XQM_COMMAND(cmd) ? XQM_MAXQUOTAS : MAXQUOTAS))
return -EINVAL;
+ /*
+ * Quota not supported on this fs? Check this before allowed_types
+ * since they needn't be set if quota is not supported.
+ */
if (!sb->s_qcop)
return -ENOSYS;
+ if (!(sb->s_dquot.allowed_types & (1 << type)))
+ return -EINVAL;

ret = check_quotactl_permission(sb, type, cmd, id);
if (ret < 0)
diff --git a/fs/super.c b/fs/super.c
index eae088f6aaae..5e70cc327dae 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -218,6 +218,13 @@ static struct super_block *alloc_super(struct file_system_type *type, int flags)
atomic_set(&s->s_active, 1);
mutex_init(&s->s_vfs_rename_mutex);
lockdep_set_class(&s->s_vfs_rename_mutex, &type->s_vfs_rename_key);
+ /*
+ * For now MAXQUOTAS check in do_quotactl() will limit quota type
+ * appropriately. When each fs sets allowed_types, we can remove the
+ * line below
+ */
+ s->s_dquot.allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP |
+ QTYPE_MASK_PRJ;
mutex_init(&s->s_dquot.dqio_mutex);
mutex_init(&s->s_dquot.dqonoff_mutex);
s->s_maxbytes = MAX_NON_LFS;
diff --git a/include/linux/quota.h b/include/linux/quota.h
index 80d345a3524c..e0cfcce0d986 100644
--- a/include/linux/quota.h
+++ b/include/linux/quota.h
@@ -56,6 +56,11 @@ enum quota_type {
PRJQUOTA = 2, /* element used for project quotas */
};

+/* Masks for quota types when used as a bitmask */
+#define QTYPE_MASK_USR (1 << USRQUOTA)
+#define QTYPE_MASK_GRP (1 << GRPQUOTA)
+#define QTYPE_MASK_PRJ (1 << PRJQUOTA)
+
typedef __kernel_uid32_t qid_t; /* Type in which we store ids in memory */
typedef long long qsize_t; /* Type in which we store sizes */

@@ -388,6 +393,7 @@ static inline void quota_send_warning(struct kqid qid, dev_t dev,

struct quota_info {
unsigned int flags; /* Flags for diskquotas on this device */
+ unsigned int allowed_types; /* Bitmask of quota types this fs supports */
struct mutex dqio_mutex; /* lock device while I/O in progress */
struct mutex dqonoff_mutex; /* Serialize quotaon & quotaoff */
struct inode *files[MAXQUOTAS]; /* inodes of quotafiles */

[Christoph Hellwig]

Post by Jan Kara
Currently all filesystems supporting VFS quota support user and group
quotas. With introduction of project quotas this is going to change so
make sure filesystem isn't called for quota type it doesn't support by
introduction of a bitmask determining which quota types each filesystem
supports.

Why don't you keep this bitmask in the dquot.c instead of pushing it
to the caller? So far usage of s_dquot is mostly confined to dquot.c
(with a few leaks to the filesystems using it), so keeping it that
way seems like a good idea.

[Jan Kara]

Post by Christoph Hellwig

Post by Jan Kara
Currently all filesystems supporting VFS quota support user and group
quotas. With introduction of project quotas this is going to change so
make sure filesystem isn't called for quota type it doesn't support by
introduction of a bitmask determining which quota types each filesystem
supports.

Why don't you keep this bitmask in the dquot.c instead of pushing it
to the caller? So far usage of s_dquot is mostly confined to dquot.c
(with a few leaks to the filesystems using it), so keeping it that
way seems like a good idea.

So there are two reasons:
1) Currently if you call quotactl() with invalid quota type you'll get
EINVAL. To maintain this with addition of project quotas you need to check
the types early before calling check_quotactl_permission() and other
checks.

2) I didn't want filesystem quotactl callbacks to deal with quota types
they don't support. Sure each fs could do a type check in the callback but
this looked easier.

Now I see your point about s_dquot and I can move allowed_types out of
s_dquot if that makes you happier. But otherwise what I did still seems as
the best solution to me.

Honza

[Christoph Hellwig]

Post by Jan Kara
1) Currently if you call quotactl() with invalid quota type you'll get
EINVAL. To maintain this with addition of project quotas you need to check
the types early before calling check_quotactl_permission() and other
checks.
2) I didn't want filesystem quotactl callbacks to deal with quota types
they don't support. Sure each fs could do a type check in the callback but
this looked easier.
Now I see your point about s_dquot and I can move allowed_types out of
s_dquot if that makes you happier. But otherwise what I did still seems as
the best solution to me.

Moving it out of s_dquot seems very sensible to me.

[Jan Kara]

CC: linux-***@vger.kernel.org
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/ext3/ext3.h | 4 ++++
fs/ext3/super.c | 10 ++++++++++
2 files changed, 14 insertions(+)

diff --git a/fs/ext3/ext3.h b/fs/ext3/ext3.h
index fc3cdcf24aed..f483a80b3fe7 100644
--- a/fs/ext3/ext3.h
+++ b/fs/ext3/ext3.h
@@ -615,6 +615,10 @@ struct ext3_inode_info {
atomic_t i_sync_tid;
atomic_t i_datasync_tid;

+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
+
struct inode vfs_inode;
};

diff --git a/fs/ext3/super.c b/fs/ext3/super.c
index 7015db0bafd1..cfb338cc0263 100644
--- a/fs/ext3/super.c
+++ b/fs/ext3/super.c
@@ -485,6 +485,10 @@ static struct inode *ext3_alloc_inode(struct super_block *sb)
ei->vfs_inode.i_version = 1;
atomic_set(&ei->i_datasync_tid, 0);
atomic_set(&ei->i_sync_tid, 0);
+#ifdef CONFIG_QUOTA
+ memset(&ei->i_dquot, 0, sizeof(ei->i_dquot));
+#endif
+
return &ei->vfs_inode;
}

@@ -764,6 +768,10 @@ static ssize_t ext3_quota_read(struct super_block *sb, int type, char *data,
size_t len, loff_t off);
static ssize_t ext3_quota_write(struct super_block *sb, int type,
const char *data, size_t len, loff_t off);
+static struct dquot **ext3_get_dquots(struct inode *inode)
+{
+ return EXT3_I(inode)->i_dquot;
+}

static const struct dquot_operations ext3_quota_operations = {
.write_dquot = ext3_write_dquot,
@@ -803,6 +811,7 @@ static const struct super_operations ext3_sops = {
#ifdef CONFIG_QUOTA
.quota_read = ext3_quota_read,
.quota_write = ext3_quota_write,
+ .get_dquots = ext3_get_dquots,
#endif
.bdev_try_to_free_page = bdev_try_to_free_page,
};
@@ -2008,6 +2017,7 @@ static int ext3_fill_super (struct super_block *sb, void *data, int silent)
#ifdef CONFIG_QUOTA
sb->s_qcop = &ext3_qctl_operations;
sb->dq_op = &ext3_quota_operations;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
#endif
memcpy(sb->s_uuid, es->s_uuid, sizeof(es->s_uuid));
INIT_LIST_HEAD(&sbi->s_orphan); /* unlinked but open files */

[Jan Kara]

CC: Dave Kleikamp <***@oracle.com>
CC: jfs-***@lists.sourceforge.net
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/jfs/jfs_incore.h | 3 +++
fs/jfs/super.c | 9 +++++++++
2 files changed, 12 insertions(+)

diff --git a/fs/jfs/jfs_incore.h b/fs/jfs/jfs_incore.h
index cf47f09e8ac8..fa7e795bd8ae 100644
--- a/fs/jfs/jfs_incore.h
+++ b/fs/jfs/jfs_incore.h
@@ -94,6 +94,9 @@ struct jfs_inode_info {
unchar _inline_ea[128]; /* 128: inline extended attr */
} link;
} u;
+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
u32 dev; /* will die when we get wide dev_t */
struct inode vfs_inode;
};
diff --git a/fs/jfs/super.c b/fs/jfs/super.c
index 93e897e588a8..2a34332e889f 100644
--- a/fs/jfs/super.c
+++ b/fs/jfs/super.c
@@ -117,6 +117,9 @@ static struct inode *jfs_alloc_inode(struct super_block *sb)
jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
if (!jfs_inode)
return NULL;
+#ifdef CONFIG_QUOTA
+ memset(&jfs_inode->i_dquot, 0, sizeof(jfs_inode->i_dquot));
+#endif
return &jfs_inode->vfs_inode;
}

@@ -537,6 +540,7 @@ static int jfs_fill_super(struct super_block *sb, void *data, int silent)
#ifdef CONFIG_QUOTA
sb->dq_op = &dquot_operations;
sb->s_qcop = &dquot_quotactl_ops;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
#endif

/*
@@ -836,6 +840,10 @@ out:
return len - towrite;
}

+static struct dquot **jfs_get_dquots(struct inode *inode)
+{
+ return JFS_IP(inode)->i_dquot;
+}
#endif

static const struct super_operations jfs_super_operations = {
@@ -854,6 +862,7 @@ static const struct super_operations jfs_super_operations = {
#ifdef CONFIG_QUOTA
.quota_read = jfs_quota_read,
.quota_write = jfs_quota_write,
+ .get_dquots = jfs_get_dquots,
#endif
};

[Jan Kara]

All filesystems using VFS quotas are now converted to use their private
i_dquot fields. Remove the i_dquot field from generic inode structure.

Signed-off-by: Jan Kara <***@suse.cz>
---
fs/inode.c | 3 ---
fs/quota/dquot.c | 3 ---
fs/super.c | 7 -------
include/linux/fs.h | 3 ---
4 files changed, 16 deletions(-)

diff --git a/fs/inode.c b/fs/inode.c
index 26753ba7b6d6..2ed95f7caa4f 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -143,9 +143,6 @@ int inode_init_always(struct super_block *sb, struct inode *inode)
inode->i_blocks = 0;
inode->i_bytes = 0;
inode->i_generation = 0;
-#ifdef CONFIG_QUOTA
- memset(&inode->i_dquot, 0, sizeof(inode->i_dquot));
-#endif
inode->i_pipe = NULL;
inode->i_bdev = NULL;
inode->i_cdev = NULL;
diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index 8144ff2b561d..5bc5f254ac1f 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -895,9 +895,6 @@ EXPORT_SYMBOL(dqget);

static inline struct dquot **i_dquot(struct inode *inode)
{
- /* Temporary workaround until all filesystems are converted. */
- if (!inode->i_sb->s_op->get_dquots)
- return inode->i_dquot;
return inode->i_sb->s_op->get_dquots(inode);
}

diff --git a/fs/super.c b/fs/super.c
index 5e70cc327dae..eae088f6aaae 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -218,13 +218,6 @@ static struct super_block *alloc_super(struct file_system_type *type, int flags)
atomic_set(&s->s_active, 1);
mutex_init(&s->s_vfs_rename_mutex);
lockdep_set_class(&s->s_vfs_rename_mutex, &type->s_vfs_rename_key);
- /*
- * For now MAXQUOTAS check in do_quotactl() will limit quota type
- * appropriately. When each fs sets allowed_types, we can remove the
- * line below
- */
- s->s_dquot.allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP |
- QTYPE_MASK_PRJ;
mutex_init(&s->s_dquot.dqio_mutex);
mutex_init(&s->s_dquot.dqonoff_mutex);
s->s_maxbytes = MAX_NON_LFS;
diff --git a/include/linux/fs.h b/include/linux/fs.h
index b714e8e0415d..ab3513abf30f 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -593,9 +593,6 @@ struct inode {
const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
struct file_lock *i_flock;
struct address_space i_data;
-#ifdef CONFIG_QUOTA
- struct dquot *i_dquot[MAXQUOTAS];
-#endif
struct list_head i_devices;
union {
struct pipe_inode_info *i_pipe;

[Jan Kara]

CC: Mark Fasheh <***@suse.com>
CC: Joel Becker <***@evilplan.org>
CC: ocfs2-***@oss.oracle.com
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/ocfs2/inode.h | 4 ++++
fs/ocfs2/super.c | 10 ++++++++++
2 files changed, 14 insertions(+)

diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h
index a9b76de46047..1bb2e27eaad7 100644
--- a/fs/ocfs2/inode.h
+++ b/fs/ocfs2/inode.h
@@ -80,6 +80,10 @@ struct ocfs2_inode_info
*/
tid_t i_sync_tid;
tid_t i_datasync_tid;
+
+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
};

/*
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 93c85bc745e1..9a97986d54f5 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -143,6 +143,11 @@ static int ocfs2_susp_quotas(struct ocfs2_super *osb, int unsuspend);
static int ocfs2_enable_quotas(struct ocfs2_super *osb);
static void ocfs2_disable_quotas(struct ocfs2_super *osb);

+static struct dquot **ocfs2_get_dquots(struct inode *inode)
+{
+ return OCFS2_I(inode)->i_dquot;
+}
+
static const struct super_operations ocfs2_sops = {
.statfs = ocfs2_statfs,
.alloc_inode = ocfs2_alloc_inode,
@@ -155,6 +160,7 @@ static const struct super_operations ocfs2_sops = {
.show_options = ocfs2_show_options,
.quota_read = ocfs2_quota_read,
.quota_write = ocfs2_quota_write,
+ .get_dquots = ocfs2_get_dquots,
};

enum {
@@ -563,6 +569,9 @@ static struct inode *ocfs2_alloc_inode(struct super_block *sb)

oi->i_sync_tid = 0;
oi->i_datasync_tid = 0;
+#ifdef CONFIG_QUOTA
+ memset(&oi->i_dquot, 0, sizeof(oi->i_dquot));
+#endif

jbd2_journal_init_jbd_inode(&oi->ip_jinode, &oi->vfs_inode);
return &oi->vfs_inode;
@@ -2073,6 +2082,7 @@ static int ocfs2_initialize_super(struct super_block *sb,
sb->s_export_op = &ocfs2_export_ops;
sb->s_qcop = &ocfs2_quotactl_ops;
sb->dq_op = &ocfs2_quota_operations;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
sb->s_xattr = ocfs2_xattr_handlers;
sb->s_time_gran = 1;
sb->s_flags |= MS_NOATIME;

[Jan Kara]

CC: reiserfs-***@vger.kernel.org
CC: Jeff Mahoney <***@suse.de>
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/reiserfs/reiserfs.h | 4 ++++
fs/reiserfs/super.c | 11 +++++++++++
2 files changed, 15 insertions(+)

diff --git a/fs/reiserfs/reiserfs.h b/fs/reiserfs/reiserfs.h
index 1894d96ccb7c..bb79cddf0a1f 100644
--- a/fs/reiserfs/reiserfs.h
+++ b/fs/reiserfs/reiserfs.h
@@ -97,6 +97,10 @@ struct reiserfs_inode_info {
#ifdef CONFIG_REISERFS_FS_XATTR
struct rw_semaphore i_xattr_sem;
#endif
+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
+
struct inode vfs_inode;
};

diff --git a/fs/reiserfs/super.c b/fs/reiserfs/super.c
index f1376c92cf74..f9bbac270cb5 100644
--- a/fs/reiserfs/super.c
+++ b/fs/reiserfs/super.c
@@ -594,6 +594,10 @@ static struct inode *reiserfs_alloc_inode(struct super_block *sb)
return NULL;
atomic_set(&ei->openers, 0);
mutex_init(&ei->tailpack);
+#ifdef CONFIG_QUOTA
+ memset(&ei->i_dquot, 0, sizeof(ei->i_dquot));
+#endif
+
return &ei->vfs_inode;
}

@@ -750,6 +754,11 @@ static ssize_t reiserfs_quota_write(struct super_block *, int, const char *,
size_t, loff_t);
static ssize_t reiserfs_quota_read(struct super_block *, int, char *, size_t,
loff_t);
+
+static struct dquot **reiserfs_get_dquots(struct inode *inode)
+{
+ return REISERFS_I(inode)->i_dquot;
+}
#endif

static const struct super_operations reiserfs_sops = {
@@ -768,6 +777,7 @@ static const struct super_operations reiserfs_sops = {
#ifdef CONFIG_QUOTA
.quota_read = reiserfs_quota_read,
.quota_write = reiserfs_quota_write,
+ .get_dquots = reiserfs_get_dquots,
#endif
};

@@ -1633,6 +1643,7 @@ static int read_super_block(struct super_block *s, int offset)
#ifdef CONFIG_QUOTA
s->s_qcop = &reiserfs_qctl_operations;
s->dq_op = &reiserfs_quota_operations;
+ sb_dqopt(s)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
#endif

/*

[Jan Kara]

CC: linux-***@vger.kernel.org
CC: "Theodore Ts'o" <***@mit.edu>
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/ext4/ext4.h | 4 ++++
fs/ext4/super.c | 8 ++++++++
2 files changed, 12 insertions(+)

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index c55a1faaed58..db3f772e57ae 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -941,6 +941,10 @@ struct ext4_inode_info {
tid_t i_sync_tid;
tid_t i_datasync_tid;

+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
+
/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
__u32 i_csum_seed;
};
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 1eda6ab0ef9d..7207799e5939 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -892,6 +892,7 @@ static struct inode *ext4_alloc_inode(struct super_block *sb)
spin_lock_init(&(ei->i_block_reservation_lock));
#ifdef CONFIG_QUOTA
ei->i_reserved_quota = 0;
+ memset(&ei->i_dquot, 0, sizeof(ei->i_dquot));
#endif
ei->jinode = NULL;
INIT_LIST_HEAD(&ei->i_rsv_conversion_list);
@@ -1068,6 +1069,11 @@ static int ext4_quota_enable(struct super_block *sb, int type, int format_id,
unsigned int flags);
static int ext4_enable_quotas(struct super_block *sb);

+static struct dquot **ext4_get_dquots(struct inode *inode)
+{
+ return EXT4_I(inode)->i_dquot;
+}
+
static const struct dquot_operations ext4_quota_operations = {
.get_reserved_space = ext4_get_reserved_space,
.write_dquot = ext4_write_dquot,
@@ -1117,6 +1123,7 @@ static const struct super_operations ext4_sops = {
#ifdef CONFIG_QUOTA
.quota_read = ext4_quota_read,
.quota_write = ext4_quota_write,
+ .get_dquots = ext4_get_dquots,
#endif
.bdev_try_to_free_page = bdev_try_to_free_page,
};
@@ -3928,6 +3935,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
sb->s_qcop = &ext4_qctl_sysfile_operations;
else
sb->s_qcop = &ext4_qctl_operations;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
#endif
memcpy(sb->s_uuid, es->s_uuid, sizeof(es->s_uuid));

[Jan Kara]

CC: linux-***@vger.kernel.org
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/ext2/ext2.h | 3 +++
fs/ext2/super.c | 10 ++++++++++
2 files changed, 13 insertions(+)

diff --git a/fs/ext2/ext2.h b/fs/ext2/ext2.h
index d9a17d0b124d..e4279ead4a05 100644
--- a/fs/ext2/ext2.h
+++ b/fs/ext2/ext2.h
@@ -689,6 +689,9 @@ struct ext2_inode_info {
struct mutex truncate_mutex;
struct inode vfs_inode;
struct list_head i_orphan; /* unlinked but open inodes */
+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
};

/*
diff --git a/fs/ext2/super.c b/fs/ext2/super.c
index 170dc41e8bf4..ef2a6cce75d8 100644
--- a/fs/ext2/super.c
+++ b/fs/ext2/super.c
@@ -166,6 +166,10 @@ static struct inode *ext2_alloc_inode(struct super_block *sb)
return NULL;
ei->i_block_alloc_info = NULL;
ei->vfs_inode.i_version = 1;
+#ifdef CONFIG_QUOTA
+ memset(&ei->i_dquot, 0, sizeof(ei->i_dquot));
+#endif
+
return &ei->vfs_inode;
}

@@ -303,6 +307,10 @@ static int ext2_show_options(struct seq_file *seq, struct dentry *root)
#ifdef CONFIG_QUOTA
static ssize_t ext2_quota_read(struct super_block *sb, int type, char *data, size_t len, loff_t off);
static ssize_t ext2_quota_write(struct super_block *sb, int type, const char *data, size_t len, loff_t off);
+static struct dquot **ext2_get_dquots(struct inode *inode)
+{
+ return EXT2_I(inode)->i_dquot;
+}
#endif

static const struct super_operations ext2_sops = {
@@ -320,6 +328,7 @@ static const struct super_operations ext2_sops = {
#ifdef CONFIG_QUOTA
.quota_read = ext2_quota_read,
.quota_write = ext2_quota_write,
+ .get_dquots = ext2_get_dquots,
#endif
};

@@ -1090,6 +1099,7 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent)
#ifdef CONFIG_QUOTA
sb->dq_op = &dquot_operations;
sb->s_qcop = &dquot_quotactl_ops;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
#endif

root = ext2_iget(sb, EXT2_ROOT_INO);

[Jan Kara]

We support user, group, and project quotas. Tell VFS about it.

CC: ***@oss.sgi.com
CC: Dave Chinner <***@fromorbit.com>
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/xfs/xfs_super.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 9f622feda6a4..b38e90997dd1 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -1425,6 +1425,8 @@ xfs_fs_fill_super(
sb->s_export_op = &xfs_export_operations;
#ifdef CONFIG_XFS_QUOTA
sb->s_qcop = &xfs_quotactl_operations;
+ sb->s_dquot.allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP |
+ QTYPE_MASK_PRJ;
#endif
sb->s_op = &xfs_super_operations;

[Jan Kara]

i_dquot array is used by relatively few filesystems (ext?, ocfs2, jfs,
reiserfs) so it is beneficial to move this array to fs-private part of
the inode. We cannot just pass quota pointers from filesystems to quota
functions because during quotaon and quotaoff we have to traverse list
of all inodes and manipulate i_dquot pointers for each inode. So we
provide a function which generic quota code can use to get pointer to
the i_dquot array from the filesystem.

Signed-off-by: Jan Kara <***@suse.cz>
---
fs/quota/dquot.c | 54 +++++++++++++++++++++++++++++++-----------------------
include/linux/fs.h | 1 +
2 files changed, 32 insertions(+), 23 deletions(-)

diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index 8a6b95e9bf3d..8144ff2b561d 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -893,6 +893,14 @@ out:
}
EXPORT_SYMBOL(dqget);

+static inline struct dquot **i_dquot(struct inode *inode)
+{
+ /* Temporary workaround until all filesystems are converted. */
+ if (!inode->i_sb->s_op->get_dquots)
+ return inode->i_dquot;
+ return inode->i_sb->s_op->get_dquots(inode);
+}
+
static int dqinit_needed(struct inode *inode, int type)
{
int cnt;
@@ -900,9 +908,9 @@ static int dqinit_needed(struct inode *inode, int type)
if (IS_NOQUOTA(inode))
return 0;
if (type != -1)
- return !inode->i_dquot[type];
+ return !i_dquot(inode)[type];
for (cnt = 0; cnt < MAXQUOTAS; cnt++)
- if (!inode->i_dquot[cnt])
+ if (!i_dquot(inode)[cnt])
return 1;
return 0;
}
@@ -965,9 +973,9 @@ static void add_dquot_ref(struct super_block *sb, int type)
static void remove_inode_dquot_ref(struct inode *inode, int type,
struct list_head *tofree_head)
{
- struct dquot *dquot = inode->i_dquot[type];
+ struct dquot *dquot = i_dquot(inode)[type];

- inode->i_dquot[type] = NULL;
+ i_dquot(inode)[type] = NULL;
if (!dquot)
return;

@@ -1402,7 +1410,7 @@ static void __dquot_initialize(struct inode *inode, int type)
* we check it without locking here to avoid unnecessary
* dqget()/dqput() calls.
*/
- if (inode->i_dquot[cnt])
+ if (i_dquot(inode)[cnt])
continue;
init_needed = 1;

@@ -1433,8 +1441,8 @@ static void __dquot_initialize(struct inode *inode, int type)
/* We could race with quotaon or dqget() could have failed */
if (!got[cnt])
continue;
- if (!inode->i_dquot[cnt]) {
- inode->i_dquot[cnt] = got[cnt];
+ if (!i_dquot(inode)[cnt]) {
+ i_dquot(inode)[cnt] = got[cnt];
got[cnt] = NULL;
/*
* Make quota reservation system happy if someone
@@ -1442,7 +1450,7 @@ static void __dquot_initialize(struct inode *inode, int type)
*/
rsv = inode_get_rsv_space(inode);
if (unlikely(rsv))
- dquot_resv_space(inode->i_dquot[cnt], rsv);
+ dquot_resv_space(i_dquot(inode)[cnt], rsv);
}
}
out_err:
@@ -1472,8 +1480,8 @@ static void __dquot_drop(struct inode *inode)

spin_lock(&dq_data_lock);
for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
- put[cnt] = inode->i_dquot[cnt];
- inode->i_dquot[cnt] = NULL;
+ put[cnt] = i_dquot(inode)[cnt];
+ i_dquot(inode)[cnt] = NULL;
}
spin_unlock(&dq_data_lock);
dqput_all(put);
@@ -1494,7 +1502,7 @@ void dquot_drop(struct inode *inode)
* add quota pointers back anyway.
*/
for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
- if (inode->i_dquot[cnt])
+ if (i_dquot(inode)[cnt])
break;
}

@@ -1595,7 +1603,7 @@ int __dquot_alloc_space(struct inode *inode, qsize_t number, int flags)
{
int cnt, ret = 0, index;
struct dquot_warn warn[MAXQUOTAS];
- struct dquot **dquots = inode->i_dquot;
+ struct dquot **dquots = i_dquot(inode);
int reserve = flags & DQUOT_SPACE_RESERVE;

if (!dquot_active(inode)) {
@@ -1647,7 +1655,7 @@ int dquot_alloc_inode(struct inode *inode)
{
int cnt, ret = 0, index;
struct dquot_warn warn[MAXQUOTAS];
- struct dquot * const *dquots = inode->i_dquot;
+ struct dquot * const *dquots = i_dquot(inode);

if (!dquot_active(inode))
return 0;
@@ -1696,14 +1704,14 @@ int dquot_claim_space_nodirty(struct inode *inode, qsize_t number)
spin_lock(&dq_data_lock);
/* Claim reserved quotas to allocated quotas */
for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
- if (inode->i_dquot[cnt])
- dquot_claim_reserved_space(inode->i_dquot[cnt],
+ if (i_dquot(inode)[cnt])
+ dquot_claim_reserved_space(i_dquot(inode)[cnt],
number);
}
/* Update inode bytes */
inode_claim_rsv_space(inode, number);
spin_unlock(&dq_data_lock);
- mark_all_dquot_dirty(inode->i_dquot);
+ mark_all_dquot_dirty(i_dquot(inode));
srcu_read_unlock(&dquot_srcu, index);
return 0;
}
@@ -1725,14 +1733,14 @@ void dquot_reclaim_space_nodirty(struct inode *inode, qsize_t number)
spin_lock(&dq_data_lock);
/* Claim reserved quotas to allocated quotas */
for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
- if (inode->i_dquot[cnt])
- dquot_reclaim_reserved_space(inode->i_dquot[cnt],
+ if (i_dquot(inode)[cnt])
+ dquot_reclaim_reserved_space(i_dquot(inode)[cnt],
number);
}
/* Update inode bytes */
inode_reclaim_rsv_space(inode, number);
spin_unlock(&dq_data_lock);
- mark_all_dquot_dirty(inode->i_dquot);
+ mark_all_dquot_dirty(i_dquot(inode));
srcu_read_unlock(&dquot_srcu, index);
return;
}
@@ -1745,7 +1753,7 @@ void __dquot_free_space(struct inode *inode, qsize_t number, int flags)
{
unsigned int cnt;
struct dquot_warn warn[MAXQUOTAS];
- struct dquot **dquots = inode->i_dquot;
+ struct dquot **dquots = i_dquot(inode);
int reserve = flags & DQUOT_SPACE_RESERVE, index;

if (!dquot_active(inode)) {
@@ -1788,7 +1796,7 @@ void dquot_free_inode(struct inode *inode)
{
unsigned int cnt;
struct dquot_warn warn[MAXQUOTAS];
- struct dquot * const *dquots = inode->i_dquot;
+ struct dquot * const *dquots = i_dquot(inode);
int index;

if (!dquot_active(inode))
@@ -1865,7 +1873,7 @@ int __dquot_transfer(struct inode *inode, struct dquot **transfer_to)
if (!sb_has_quota_active(inode->i_sb, cnt))
continue;
is_valid[cnt] = 1;
- transfer_from[cnt] = inode->i_dquot[cnt];
+ transfer_from[cnt] = i_dquot(inode)[cnt];
ret = check_idq(transfer_to[cnt], 1, &warn_to[cnt]);
if (ret)
goto over_quota;
@@ -1901,7 +1909,7 @@ int __dquot_transfer(struct inode *inode, struct dquot **transfer_to)
dquot_incr_space(transfer_to[cnt], cur_space);
dquot_resv_space(transfer_to[cnt], rsv_space);

- inode->i_dquot[cnt] = transfer_to[cnt];
+ i_dquot(inode)[cnt] = transfer_to[cnt];
}
spin_unlock(&dq_data_lock);

diff --git a/include/linux/fs.h b/include/linux/fs.h
index a957d4366c24..b714e8e0415d 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1565,6 +1565,7 @@ struct super_operations {
#ifdef CONFIG_QUOTA
ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
+ struct dquot **(*get_dquots)(struct inode *);
#endif
int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
long (*nr_cached_objects)(struct super_block *, int);

[Jan Kara]

We don't use const through VFS too much so just remove it from quota
function declarations.

Signed-off-by: Jan Kara <***@suse.cz>
---
fs/quota/dquot.c | 4 ++--
include/linux/quotaops.h | 8 ++++----
2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index 8b663b2d9562..8a6b95e9bf3d 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -1643,7 +1643,7 @@ EXPORT_SYMBOL(__dquot_alloc_space);
/*
* This operation can block, but only after everything is updated
*/
-int dquot_alloc_inode(const struct inode *inode)
+int dquot_alloc_inode(struct inode *inode)
{
int cnt, ret = 0, index;
struct dquot_warn warn[MAXQUOTAS];
@@ -1784,7 +1784,7 @@ EXPORT_SYMBOL(__dquot_free_space);
/*
* This operation can block, but only after everything is updated
*/
-void dquot_free_inode(const struct inode *inode)
+void dquot_free_inode(struct inode *inode)
{
unsigned int cnt;
struct dquot_warn warn[MAXQUOTAS];
diff --git a/include/linux/quotaops.h b/include/linux/quotaops.h
index 1d3eee594cd6..f23538a6e411 100644
--- a/include/linux/quotaops.h
+++ b/include/linux/quotaops.h
@@ -64,10 +64,10 @@ void dquot_destroy(struct dquot *dquot);
int __dquot_alloc_space(struct inode *inode, qsize_t number, int flags);
void __dquot_free_space(struct inode *inode, qsize_t number, int flags);

-int dquot_alloc_inode(const struct inode *inode);
+int dquot_alloc_inode(struct inode *inode);

int dquot_claim_space_nodirty(struct inode *inode, qsize_t number);
-void dquot_free_inode(const struct inode *inode);
+void dquot_free_inode(struct inode *inode);
void dquot_reclaim_space_nodirty(struct inode *inode, qsize_t number);

int dquot_disable(struct super_block *sb, int type, unsigned int flags);
@@ -213,12 +213,12 @@ static inline void dquot_drop(struct inode *inode)
{
}

-static inline int dquot_alloc_inode(const struct inode *inode)
+static inline int dquot_alloc_inode(struct inode *inode)
{
return 0;
}

-static inline void dquot_free_inode(const struct inode *inode)
+static inline void dquot_free_inode(struct inode *inode)
{
}

[Jan Kara]

We support user and group quotas. Tell vfs about it.

Acked-by: Steven Whitehouse <***@redhat.com>
CC: cluster-***@redhat.com
Signed-off-by: Jan Kara <***@suse.cz>
---
fs/gfs2/ops_fstype.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c
index d3eae244076e..23854d24eb50 100644
--- a/fs/gfs2/ops_fstype.c
+++ b/fs/gfs2/ops_fstype.c
@@ -1083,6 +1083,7 @@ static int fill_super(struct super_block *sb, struct gfs2_args *args, int silent
sb->s_xattr = gfs2_xattr_handlers;
sb->s_qcop = &gfs2_quotactl_ops;
sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
sb->s_time_gran = 1;
sb->s_maxbytes = MAX_LFS_FILESIZE;

[Joel Becker]

Post by Jan Kara
---
fs/ocfs2/inode.h | 4 ++++
fs/ocfs2/super.c | 10 ++++++++++
2 files changed, 14 insertions(+)
diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h
index a9b76de46047..1bb2e27eaad7 100644
--- a/fs/ocfs2/inode.h
+++ b/fs/ocfs2/inode.h
@@ -80,6 +80,10 @@ struct ocfs2_inode_info
*/
tid_t i_sync_tid;
tid_t i_datasync_tid;
+
+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
};
/*
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 93c85bc745e1..9a97986d54f5 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -143,6 +143,11 @@ static int ocfs2_susp_quotas(struct ocfs2_super *osb, int unsuspend);
static int ocfs2_enable_quotas(struct ocfs2_super *osb);
static void ocfs2_disable_quotas(struct ocfs2_super *osb);
+static struct dquot **ocfs2_get_dquots(struct inode *inode)
+{
+ return OCFS2_I(inode)->i_dquot;
+}

Call me silly, but given that ocfs2_inode_info->i_dquot; is wrapped in
CONFIG_QUOTA, shouldn't this accessor be as well? Is GCC really smart
enough to ignore the undefined member because ocfs2_get_dquots() isn't
called?

Joel

Post by Jan Kara
+
static const struct super_operations ocfs2_sops = {
.statfs = ocfs2_statfs,
.alloc_inode = ocfs2_alloc_inode,
@@ -155,6 +160,7 @@ static const struct super_operations ocfs2_sops = {
.show_options = ocfs2_show_options,
.quota_read = ocfs2_quota_read,
.quota_write = ocfs2_quota_write,
+ .get_dquots = ocfs2_get_dquots,
};
enum {
@@ -563,6 +569,9 @@ static struct inode *ocfs2_alloc_inode(struct super_block *sb)
oi->i_sync_tid = 0;
oi->i_datasync_tid = 0;
+#ifdef CONFIG_QUOTA
+ memset(&oi->i_dquot, 0, sizeof(oi->i_dquot));
+#endif
jbd2_journal_init_jbd_inode(&oi->ip_jinode, &oi->vfs_inode);
return &oi->vfs_inode;
@@ -2073,6 +2082,7 @@ static int ocfs2_initialize_super(struct super_block *sb,
sb->s_export_op = &ocfs2_export_ops;
sb->s_qcop = &ocfs2_quotactl_ops;
sb->dq_op = &ocfs2_quota_operations;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
sb->s_xattr = ocfs2_xattr_handlers;
sb->s_time_gran = 1;
sb->s_flags |= MS_NOATIME;
--
1.8.1.4

[Jan Kara]

Post by Joel Becker

Post by Jan Kara
---
fs/ocfs2/inode.h | 4 ++++
fs/ocfs2/super.c | 10 ++++++++++
2 files changed, 14 insertions(+)
diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h
index a9b76de46047..1bb2e27eaad7 100644
--- a/fs/ocfs2/inode.h
+++ b/fs/ocfs2/inode.h
@@ -80,6 +80,10 @@ struct ocfs2_inode_info
*/
tid_t i_sync_tid;
tid_t i_datasync_tid;
+
+#ifdef CONFIG_QUOTA
+ struct dquot *i_dquot[MAXQUOTAS];
+#endif
};
/*
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 93c85bc745e1..9a97986d54f5 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -143,6 +143,11 @@ static int ocfs2_susp_quotas(struct ocfs2_super *osb, int unsuspend);
static int ocfs2_enable_quotas(struct ocfs2_super *osb);
static void ocfs2_disable_quotas(struct ocfs2_super *osb);
+static struct dquot **ocfs2_get_dquots(struct inode *inode)
+{
+ return OCFS2_I(inode)->i_dquot;
+}

Call me silly, but given that ocfs2_inode_info->i_dquot; is wrapped in
CONFIG_QUOTA, shouldn't this accessor be as well? Is GCC really smart
enough to ignore the undefined member because ocfs2_get_dquots() isn't
called?

Actually, ocfs2 selects QUOTA so it cannot happen that CONFIG_QUOTA isn't
defined for ocfs2. I'll remove the CONFIG_QUOTA checks from ocfs2 patch.

Honza

Post by Joel Becker
Joel

Post by Jan Kara
+
static const struct super_operations ocfs2_sops = {
.statfs = ocfs2_statfs,
.alloc_inode = ocfs2_alloc_inode,
@@ -155,6 +160,7 @@ static const struct super_operations ocfs2_sops = {
.show_options = ocfs2_show_options,
.quota_read = ocfs2_quota_read,
.quota_write = ocfs2_quota_write,
+ .get_dquots = ocfs2_get_dquots,
};
enum {
@@ -563,6 +569,9 @@ static struct inode *ocfs2_alloc_inode(struct super_block *sb)
oi->i_sync_tid = 0;
oi->i_datasync_tid = 0;
+#ifdef CONFIG_QUOTA
+ memset(&oi->i_dquot, 0, sizeof(oi->i_dquot));
+#endif
jbd2_journal_init_jbd_inode(&oi->ip_jinode, &oi->vfs_inode);
return &oi->vfs_inode;
@@ -2073,6 +2082,7 @@ static int ocfs2_initialize_super(struct super_block *sb,
sb->s_export_op = &ocfs2_export_ops;
sb->s_qcop = &ocfs2_quotactl_ops;
sb->dq_op = &ocfs2_quota_operations;
+ sb_dqopt(sb)->allowed_types = QTYPE_MASK_USR | QTYPE_MASK_GRP;
sb->s_xattr = ocfs2_xattr_handlers;
sb->s_time_gran = 1;
sb->s_flags |= MS_NOATIME;
--
1.8.1.4

--
"I'm so tired of being tired,
Sure as night will follow day.
Most things I worry about
Never happen anyway."
http://www.jlbec.org/