summaryrefslogtreecommitdiff
path: root/fs/ceph/quota.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/ceph/quota.c')
-rw-r--r--fs/ceph/quota.c297
1 files changed, 241 insertions, 56 deletions
diff --git a/fs/ceph/quota.c b/fs/ceph/quota.c
index 03f4d24db8fe..d90eda19bcc4 100644
--- a/fs/ceph/quota.c
+++ b/fs/ceph/quota.c
@@ -3,19 +3,6 @@
* quota.c - CephFS quota
*
* Copyright (C) 2017-2018 SUSE
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License
- * as published by the Free Software Foundation; either version 2
- * of the License, or (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, see <http://www.gnu.org/licenses/>.
*/
#include <linux/statfs.h>
@@ -25,7 +12,7 @@
void ceph_adjust_quota_realms_count(struct inode *inode, bool inc)
{
- struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+ struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
if (inc)
atomic64_inc(&mdsc->quotarealms_count);
else
@@ -34,8 +21,20 @@ void ceph_adjust_quota_realms_count(struct inode *inode, bool inc)
static inline bool ceph_has_realms_with_quotas(struct inode *inode)
{
- struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
- return atomic64_read(&mdsc->quotarealms_count) > 0;
+ struct super_block *sb = inode->i_sb;
+ struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(sb);
+ struct inode *root = d_inode(sb->s_root);
+
+ if (atomic64_read(&mdsc->quotarealms_count) > 0)
+ return true;
+ /* if root is the real CephFS root, we don't have quota realms */
+ if (root && ceph_ino(root) == CEPH_INO_ROOT)
+ return false;
+ /* MDS stray dirs have no quota realms */
+ if (ceph_vino_is_reserved(ceph_inode(inode)->i_vino))
+ return false;
+ /* otherwise, we can't know for sure */
+ return true;
}
void ceph_handle_quota(struct ceph_mds_client *mdsc,
@@ -44,29 +43,28 @@ void ceph_handle_quota(struct ceph_mds_client *mdsc,
{
struct super_block *sb = mdsc->fsc->sb;
struct ceph_mds_quota *h = msg->front.iov_base;
+ struct ceph_client *cl = mdsc->fsc->client;
struct ceph_vino vino;
struct inode *inode;
struct ceph_inode_info *ci;
+ if (!ceph_inc_mds_stopping_blocker(mdsc, session))
+ return;
+
if (msg->front.iov_len < sizeof(*h)) {
- pr_err("%s corrupt message mds%d len %d\n", __func__,
- session->s_mds, (int)msg->front.iov_len);
+ pr_err_client(cl, "corrupt message mds%d len %d\n",
+ session->s_mds, (int)msg->front.iov_len);
ceph_msg_dump(msg);
- return;
+ goto out;
}
- /* increment msg sequence number */
- mutex_lock(&session->s_mutex);
- session->s_seq++;
- mutex_unlock(&session->s_mutex);
-
/* lookup inode */
vino.ino = le64_to_cpu(h->ino);
vino.snap = CEPH_NOSNAP;
inode = ceph_find_inode(sb, vino);
if (!inode) {
- pr_warn("Failed to find inode %llu\n", vino.ino);
- return;
+ pr_warn_client(cl, "failed to find inode %llx\n", vino.ino);
+ goto out;
}
ci = ceph_inode(inode);
@@ -79,48 +77,194 @@ void ceph_handle_quota(struct ceph_mds_client *mdsc,
spin_unlock(&ci->i_ceph_lock);
iput(inode);
+out:
+ ceph_dec_mds_stopping_blocker(mdsc);
+}
+
+static struct ceph_quotarealm_inode *
+find_quotarealm_inode(struct ceph_mds_client *mdsc, u64 ino)
+{
+ struct ceph_quotarealm_inode *qri = NULL;
+ struct rb_node **node, *parent = NULL;
+ struct ceph_client *cl = mdsc->fsc->client;
+
+ mutex_lock(&mdsc->quotarealms_inodes_mutex);
+ node = &(mdsc->quotarealms_inodes.rb_node);
+ while (*node) {
+ parent = *node;
+ qri = container_of(*node, struct ceph_quotarealm_inode, node);
+
+ if (ino < qri->ino)
+ node = &((*node)->rb_left);
+ else if (ino > qri->ino)
+ node = &((*node)->rb_right);
+ else
+ break;
+ }
+ if (!qri || (qri->ino != ino)) {
+ /* Not found, create a new one and insert it */
+ qri = kmalloc(sizeof(*qri), GFP_KERNEL);
+ if (qri) {
+ qri->ino = ino;
+ qri->inode = NULL;
+ qri->timeout = 0;
+ mutex_init(&qri->mutex);
+ rb_link_node(&qri->node, parent, node);
+ rb_insert_color(&qri->node, &mdsc->quotarealms_inodes);
+ } else
+ pr_warn_client(cl, "Failed to alloc quotarealms_inode\n");
+ }
+ mutex_unlock(&mdsc->quotarealms_inodes_mutex);
+
+ return qri;
}
/*
- * This function walks through the snaprealm for an inode and returns the
- * ceph_snap_realm for the first snaprealm that has quotas set (either max_files
- * or max_bytes). If the root is reached, return the root ceph_snap_realm
- * instead.
+ * This function will try to lookup a realm inode which isn't visible in the
+ * filesystem mountpoint. A list of these kind of inodes (not visible) is
+ * maintained in the mdsc and freed only when the filesystem is umounted.
+ *
+ * Note that these inodes are kept in this list even if the lookup fails, which
+ * allows to prevent useless lookup requests.
+ */
+static struct inode *lookup_quotarealm_inode(struct ceph_mds_client *mdsc,
+ struct super_block *sb,
+ struct ceph_snap_realm *realm)
+{
+ struct ceph_client *cl = mdsc->fsc->client;
+ struct ceph_quotarealm_inode *qri;
+ struct inode *in;
+
+ qri = find_quotarealm_inode(mdsc, realm->ino);
+ if (!qri)
+ return NULL;
+
+ mutex_lock(&qri->mutex);
+ if (qri->inode && ceph_is_any_caps(qri->inode)) {
+ /* A request has already returned the inode */
+ mutex_unlock(&qri->mutex);
+ return qri->inode;
+ }
+ /* Check if this inode lookup has failed recently */
+ if (qri->timeout &&
+ time_before_eq(jiffies, qri->timeout)) {
+ mutex_unlock(&qri->mutex);
+ return NULL;
+ }
+ if (qri->inode) {
+ /* get caps */
+ int ret = __ceph_do_getattr(qri->inode, NULL,
+ CEPH_STAT_CAP_INODE, true);
+ if (ret >= 0)
+ in = qri->inode;
+ else
+ in = ERR_PTR(ret);
+ } else {
+ in = ceph_lookup_inode(sb, realm->ino);
+ }
+
+ if (IS_ERR(in)) {
+ doutc(cl, "Can't lookup inode %llx (err: %ld)\n", realm->ino,
+ PTR_ERR(in));
+ qri->timeout = jiffies + secs_to_jiffies(60); /* XXX */
+ } else {
+ qri->timeout = 0;
+ qri->inode = in;
+ }
+ mutex_unlock(&qri->mutex);
+
+ return in;
+}
+
+void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc)
+{
+ struct ceph_quotarealm_inode *qri;
+ struct rb_node *node;
+
+ /*
+ * It should now be safe to clean quotarealms_inode tree without holding
+ * mdsc->quotarealms_inodes_mutex...
+ */
+ mutex_lock(&mdsc->quotarealms_inodes_mutex);
+ while (!RB_EMPTY_ROOT(&mdsc->quotarealms_inodes)) {
+ node = rb_first(&mdsc->quotarealms_inodes);
+ qri = rb_entry(node, struct ceph_quotarealm_inode, node);
+ rb_erase(node, &mdsc->quotarealms_inodes);
+ iput(qri->inode);
+ kfree(qri);
+ }
+ mutex_unlock(&mdsc->quotarealms_inodes_mutex);
+}
+
+/*
+ * This function walks through the snaprealm for an inode and set the
+ * realmp with the first snaprealm that has quotas set (max_files,
+ * max_bytes, or any, depending on the 'which_quota' argument). If the root is
+ * reached, set the realmp with the root ceph_snap_realm instead.
*
* Note that the caller is responsible for calling ceph_put_snap_realm() on the
* returned realm.
+ *
+ * Callers of this function need to hold mdsc->snap_rwsem. However, if there's
+ * a need to do an inode lookup, this rwsem will be temporarily dropped. Hence
+ * the 'retry' argument: if rwsem needs to be dropped and 'retry' is 'false'
+ * this function will return -EAGAIN; otherwise, the snaprealms walk-through
+ * will be restarted.
*/
-static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc,
- struct inode *inode)
+static int get_quota_realm(struct ceph_mds_client *mdsc, struct inode *inode,
+ enum quota_get_realm which_quota,
+ struct ceph_snap_realm **realmp, bool retry)
{
+ struct ceph_client *cl = mdsc->fsc->client;
struct ceph_inode_info *ci = NULL;
struct ceph_snap_realm *realm, *next;
struct inode *in;
bool has_quota;
+ if (realmp)
+ *realmp = NULL;
if (ceph_snap(inode) != CEPH_NOSNAP)
- return NULL;
+ return 0;
+restart:
realm = ceph_inode(inode)->i_snap_realm;
if (realm)
ceph_get_snap_realm(mdsc, realm);
else
- pr_err_ratelimited("get_quota_realm: ino (%llx.%llx) "
- "null i_snap_realm\n", ceph_vinop(inode));
+ pr_err_ratelimited_client(cl,
+ "%p %llx.%llx null i_snap_realm\n",
+ inode, ceph_vinop(inode));
while (realm) {
+ bool has_inode;
+
spin_lock(&realm->inodes_with_caps_lock);
- in = realm->inode ? igrab(realm->inode) : NULL;
+ has_inode = realm->inode;
+ in = has_inode ? igrab(realm->inode) : NULL;
spin_unlock(&realm->inodes_with_caps_lock);
- if (!in)
+ if (has_inode && !in)
break;
+ if (!in) {
+ up_read(&mdsc->snap_rwsem);
+ in = lookup_quotarealm_inode(mdsc, inode->i_sb, realm);
+ down_read(&mdsc->snap_rwsem);
+ if (IS_ERR_OR_NULL(in))
+ break;
+ ceph_put_snap_realm(mdsc, realm);
+ if (!retry)
+ return -EAGAIN;
+ goto restart;
+ }
ci = ceph_inode(in);
- has_quota = __ceph_has_any_quota(ci);
+ has_quota = __ceph_has_quota(ci, which_quota);
iput(in);
next = realm->parent;
- if (has_quota || !next)
- return realm;
+ if (has_quota || !next) {
+ if (realmp)
+ *realmp = realm;
+ return 0;
+ }
ceph_get_snap_realm(mdsc, next);
ceph_put_snap_realm(mdsc, realm);
@@ -129,18 +273,32 @@ static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc,
if (realm)
ceph_put_snap_realm(mdsc, realm);
- return NULL;
+ return 0;
}
bool ceph_quota_is_same_realm(struct inode *old, struct inode *new)
{
- struct ceph_mds_client *mdsc = ceph_inode_to_client(old)->mdsc;
+ struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(old->i_sb);
struct ceph_snap_realm *old_realm, *new_realm;
bool is_same;
-
+ int ret;
+
+restart:
+ /*
+ * We need to lookup 2 quota realms atomically, i.e. with snap_rwsem.
+ * However, get_quota_realm may drop it temporarily. By setting the
+ * 'retry' parameter to 'false', we'll get -EAGAIN if the rwsem was
+ * dropped and we can then restart the whole operation.
+ */
down_read(&mdsc->snap_rwsem);
- old_realm = get_quota_realm(mdsc, old);
- new_realm = get_quota_realm(mdsc, new);
+ get_quota_realm(mdsc, old, QUOTA_GET_ANY, &old_realm, true);
+ ret = get_quota_realm(mdsc, new, QUOTA_GET_ANY, &new_realm, false);
+ if (ret == -EAGAIN) {
+ up_read(&mdsc->snap_rwsem);
+ if (old_realm)
+ ceph_put_snap_realm(mdsc, old_realm);
+ goto restart;
+ }
is_same = (old_realm == new_realm);
up_read(&mdsc->snap_rwsem);
@@ -168,7 +326,8 @@ enum quota_check_op {
static bool check_quota_exceeded(struct inode *inode, enum quota_check_op op,
loff_t delta)
{
- struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+ struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
+ struct ceph_client *cl = mdsc->fsc->client;
struct ceph_inode_info *ci;
struct ceph_snap_realm *realm, *next;
struct inode *in;
@@ -179,19 +338,32 @@ static bool check_quota_exceeded(struct inode *inode, enum quota_check_op op,
return false;
down_read(&mdsc->snap_rwsem);
+restart:
realm = ceph_inode(inode)->i_snap_realm;
if (realm)
ceph_get_snap_realm(mdsc, realm);
else
- pr_err_ratelimited("check_quota_exceeded: ino (%llx.%llx) "
- "null i_snap_realm\n", ceph_vinop(inode));
+ pr_err_ratelimited_client(cl,
+ "%p %llx.%llx null i_snap_realm\n",
+ inode, ceph_vinop(inode));
while (realm) {
+ bool has_inode;
+
spin_lock(&realm->inodes_with_caps_lock);
- in = realm->inode ? igrab(realm->inode) : NULL;
+ has_inode = realm->inode;
+ in = has_inode ? igrab(realm->inode) : NULL;
spin_unlock(&realm->inodes_with_caps_lock);
- if (!in)
+ if (has_inode && !in)
break;
-
+ if (!in) {
+ up_read(&mdsc->snap_rwsem);
+ in = lookup_quotarealm_inode(mdsc, inode->i_sb, realm);
+ down_read(&mdsc->snap_rwsem);
+ if (IS_ERR_OR_NULL(in))
+ break;
+ ceph_put_snap_realm(mdsc, realm);
+ goto restart;
+ }
ci = ceph_inode(in);
spin_lock(&ci->i_ceph_lock);
if (op == QUOTA_CHECK_MAX_FILES_OP) {
@@ -204,8 +376,6 @@ static bool check_quota_exceeded(struct inode *inode, enum quota_check_op op,
spin_unlock(&ci->i_ceph_lock);
switch (op) {
case QUOTA_CHECK_MAX_FILES_OP:
- exceeded = (max && (rvalue >= max));
- break;
case QUOTA_CHECK_MAX_BYTES_OP:
exceeded = (max && (rvalue + delta > max));
break;
@@ -225,7 +395,7 @@ static bool check_quota_exceeded(struct inode *inode, enum quota_check_op op,
break;
default:
/* Shouldn't happen */
- pr_warn("Invalid quota check op (%d)\n", op);
+ pr_warn_client(cl, "Invalid quota check op (%d)\n", op);
exceeded = true; /* Just break the loop */
}
iput(in);
@@ -259,7 +429,7 @@ bool ceph_quota_is_max_files_exceeded(struct inode *inode)
WARN_ON(!S_ISDIR(inode->i_mode));
- return check_quota_exceeded(inode, QUOTA_CHECK_MAX_FILES_OP, 0);
+ return check_quota_exceeded(inode, QUOTA_CHECK_MAX_FILES_OP, 1);
}
/*
@@ -327,7 +497,8 @@ bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, struct kstatfs *buf)
bool is_updated = false;
down_read(&mdsc->snap_rwsem);
- realm = get_quota_realm(mdsc, d_inode(fsc->sb->s_root));
+ get_quota_realm(mdsc, d_inode(fsc->sb->s_root), QUOTA_GET_MAX_BYTES,
+ &realm, true);
up_read(&mdsc->snap_rwsem);
if (!realm)
return false;
@@ -341,10 +512,24 @@ bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, struct kstatfs *buf)
if (ci->i_max_bytes) {
total = ci->i_max_bytes >> CEPH_BLOCK_SHIFT;
used = ci->i_rbytes >> CEPH_BLOCK_SHIFT;
+ /* For quota size less than 4MB, use 4KB block size */
+ if (!total) {
+ total = ci->i_max_bytes >> CEPH_4K_BLOCK_SHIFT;
+ used = ci->i_rbytes >> CEPH_4K_BLOCK_SHIFT;
+ buf->f_frsize = 1 << CEPH_4K_BLOCK_SHIFT;
+ }
/* It is possible for a quota to be exceeded.
* Report 'zero' in that case
*/
free = total > used ? total - used : 0;
+ /* For quota size less than 4KB, report the
+ * total=used=4KB,free=0 when quota is full
+ * and total=free=4KB, used=0 otherwise */
+ if (!total) {
+ total = 1;
+ free = ci->i_max_bytes > ci->i_rbytes ? 1 : 0;
+ buf->f_frsize = 1 << CEPH_4K_BLOCK_SHIFT;
+ }
}
spin_unlock(&ci->i_ceph_lock);
if (total) {
generated by cgit (git 2.34.1) at 2025-12-25 06:36:28 +0000