[PATCH 2/3] Make VFS handle mount autoexpiry

From: David Howells
Date: Thu Mar 03 2011 - 11:58:31 EST

Make the VFS handle mount autoexpiry, rather than doing it in the filesystems
(such as AFS, CIFS and NFS). This simplifies the reference counting, since
do_add_mount() once again adds mounts to the expiration list, and simplifies
the filesystems since they no longer have to do anything barring set
MNT_EXPIRABLE on an expirable mount before returning it.

Additionally, provide a tuning knob to set the periodicity of the reaper in
seconds:

/proc/sys/fs/mount-expiry-period

The default is 10 minutes.

The NFS sysctl (nfs_mountpoint_timeout) is removed in its favour.

Signed-off-by: David Howells <dhowells@xxxxxxxxxx>
---

Documentation/filesystems/vfs.txt | 7 +-
fs/afs/internal.h | 1
fs/afs/mntpt.c | 60 ++-------------------
fs/afs/super.c | 1
fs/cifs/cifs_dfs_ref.c | 53 ++-----------------
fs/cifs/cifsfs.c | 3 -
fs/cifs/cifsproto.h | 1
fs/namei.c | 8 +--
fs/namespace.c | 106 ++++++++++++++++---------------------
fs/nfs/client.c | 1
fs/nfs/namespace.c | 26 ---------
fs/nfs/sysctl.c | 7 --
include/linux/mount.h | 5 +-
kernel/sysctl.c | 11 ++++
14 files changed, 76 insertions(+), 214 deletions(-)

diff --git a/Documentation/filesystems/vfs.txt b/Documentation/filesystems/vfs.txt
index c6878a0..3a40c31 100644
--- a/Documentation/filesystems/vfs.txt
+++ b/Documentation/filesystems/vfs.txt
@@ -943,10 +943,9 @@ struct dentry_operations {
ordinary directory and returned to pathwalk to continue walking.

If a vfsmount is returned, the caller will attempt to mount it on the
- mountpoint and will remove the vfsmount from its expiration list in
- the case of failure. The vfsmount should be returned with 2 refs on
- it to prevent automatic expiration - the caller will clean up the
- additional ref.
+ mountpoint and will clean it up on failure. If mnt_expiry_mark is set
+ on the vfsmount, the caller will add it to the global expiration list
+ if successfully mounted and clear the mark.

This function is only used if DCACHE_NEED_AUTOMOUNT is set on the
dentry. This is set by __d_instantiate() if S_AUTOMOUNT is set on the
diff --git a/fs/afs/internal.h b/fs/afs/internal.h
index 5a9b684..cb7d2c7 100644
--- a/fs/afs/internal.h
+++ b/fs/afs/internal.h
@@ -594,7 +594,6 @@ extern const struct file_operations afs_mntpt_file_operations;

extern struct vfsmount *afs_d_automount(struct path *);
extern int afs_mntpt_check_symlink(struct afs_vnode *, struct key *);
-extern void afs_mntpt_kill_timer(void);

/*
* proc.c
diff --git a/fs/afs/mntpt.c b/fs/afs/mntpt.c
index aa59184..74b43b9 100644
--- a/fs/afs/mntpt.c
+++ b/fs/afs/mntpt.c
@@ -24,7 +24,6 @@ static struct dentry *afs_mntpt_lookup(struct inode *dir,
struct dentry *dentry,
struct nameidata *nd);
static int afs_mntpt_open(struct inode *inode, struct file *file);
-static void afs_mntpt_expiry_timed_out(struct work_struct *work);

const struct file_operations afs_mntpt_file_operations = {
.open = afs_mntpt_open,
@@ -41,11 +40,6 @@ const struct inode_operations afs_autocell_inode_operations = {
.getattr = afs_getattr,
};

-static LIST_HEAD(afs_vfsmounts);
-static DECLARE_DELAYED_WORK(afs_mntpt_expiry_timer, afs_mntpt_expiry_timed_out);
-
-static unsigned long afs_mntpt_expiry_timeout = 10 * 60;
-
/*
* check a symbolic link to see whether it actually encodes a mountpoint
* - sets the AFS_VNODE_MOUNTPOINT flag on the vnode appropriately
@@ -136,11 +130,12 @@ static int afs_mntpt_open(struct inode *inode, struct file *file)
/*
* create a vfsmount to be automounted
*/
-static struct vfsmount *afs_mntpt_do_automount(struct dentry *mntpt)
+struct vfsmount *afs_d_automount(struct path *mountpoint)
{
struct afs_super_info *super;
struct vfsmount *mnt;
struct afs_vnode *vnode;
+ struct dentry *mntpt = mountpoint->dentry;
struct page *page;
char *devname, *options;
bool rwpath = false;
@@ -219,6 +214,9 @@ static struct vfsmount *afs_mntpt_do_automount(struct dentry *mntpt)
mnt = vfs_kern_mount(&afs_fs_type, 0, devname, options);
_debug("--- mount result %p ---", mnt);

+ if (!IS_ERR(mnt))
+ mnt->mnt_expiry_mark = 1;
+
free_page((unsigned long) devname);
free_page((unsigned long) options);
_leave(" = %p", mnt);
@@ -234,51 +232,3 @@ error_no_devname:
_leave(" = %d", ret);
return ERR_PTR(ret);
}
-
-/*
- * handle an automount point
- */
-struct vfsmount *afs_d_automount(struct path *path)
-{
- struct vfsmount *newmnt;
-
- _enter("{%s,%s}", path->mnt->mnt_devname, path->dentry->d_name.name);
-
- newmnt = afs_mntpt_do_automount(path->dentry);
- if (IS_ERR(newmnt))
- return newmnt;
-
- mntget(newmnt); /* prevent immediate expiration */
- mnt_set_expiry(newmnt, &afs_vfsmounts);
- queue_delayed_work(afs_wq, &afs_mntpt_expiry_timer,
- afs_mntpt_expiry_timeout * HZ);
- _leave(" = %p {%s}", newmnt, newmnt->mnt_devname);
- return newmnt;
-}
-
-/*
- * handle mountpoint expiry timer going off
- */
-static void afs_mntpt_expiry_timed_out(struct work_struct *work)
-{
- _enter("");
-
- if (!list_empty(&afs_vfsmounts)) {
- mark_mounts_for_expiry(&afs_vfsmounts);
- queue_delayed_work(afs_wq, &afs_mntpt_expiry_timer,
- afs_mntpt_expiry_timeout * HZ);
- }
-
- _leave("");
-}
-
-/*
- * kill the AFS mountpoint timer if it's still running
- */
-void afs_mntpt_kill_timer(void)
-{
- _enter("");
-
- ASSERT(list_empty(&afs_vfsmounts));
- cancel_delayed_work_sync(&afs_mntpt_expiry_timer);
-}
diff --git a/fs/afs/super.c b/fs/afs/super.c
index fb240e8..3217a42 100644
--- a/fs/afs/super.c
+++ b/fs/afs/super.c
@@ -115,7 +115,6 @@ void __exit afs_fs_exit(void)
{
_enter("");

- afs_mntpt_kill_timer();
unregister_filesystem(&afs_fs_type);

if (atomic_read(&afs_count_active_inodes) != 0) {
diff --git a/fs/cifs/cifs_dfs_ref.c b/fs/cifs/cifs_dfs_ref.c
index 0a265ad..923a92b 100644
--- a/fs/cifs/cifs_dfs_ref.c
+++ b/fs/cifs/cifs_dfs_ref.c
@@ -24,29 +24,6 @@
#include "dns_resolve.h"
#include "cifs_debug.h"

-static LIST_HEAD(cifs_dfs_automount_list);
-
-static void cifs_dfs_expire_automounts(struct work_struct *work);
-static DECLARE_DELAYED_WORK(cifs_dfs_automount_task,
- cifs_dfs_expire_automounts);
-static int cifs_dfs_mountpoint_expiry_timeout = 500 * HZ;
-
-static void cifs_dfs_expire_automounts(struct work_struct *work)
-{
- struct list_head *list = &cifs_dfs_automount_list;
-
- mark_mounts_for_expiry(list);
- if (!list_empty(list))
- schedule_delayed_work(&cifs_dfs_automount_task,
- cifs_dfs_mountpoint_expiry_timeout);
-}
-
-void cifs_dfs_release_automount_timer(void)
-{
- BUG_ON(!list_empty(&cifs_dfs_automount_list));
- cancel_delayed_work_sync(&cifs_dfs_automount_task);
-}
-
/**
* cifs_get_share_name - extracts share name from UNC
* @node_name: pointer to UNC string
@@ -267,8 +244,9 @@ static void dump_referral(const struct dfs_info3_param *ref)
/*
* Create a vfsmount that we can automount
*/
-static struct vfsmount *cifs_dfs_do_automount(struct dentry *mntpt)
+struct vfsmount *cifs_dfs_d_automount(struct path *mountpoint)
{
+ struct dentry *mntpt = mountpoint->dentry;
struct dfs_info3_param *referrals = NULL;
unsigned int num_referrals = 0;
struct cifs_sb_info *cifs_sb;
@@ -325,8 +303,10 @@ static struct vfsmount *cifs_dfs_do_automount(struct dentry *mntpt)
full_path, referrals + i);
cFYI(1, "%s: cifs_dfs_do_refmount:%s , mnt:%p", __func__,
referrals[i].node_name, mnt);
- if (!IS_ERR(mnt))
+ if (!IS_ERR(mnt)) {
+ mnt->mnt_expiry_mark = 1;
goto success;
+ }
}

/* no valid submounts were found; return error from get_dfs_path() by
@@ -343,28 +323,5 @@ cdda_exit:
return mnt;
}

-/*
- * Attempt to automount the referral
- */
-struct vfsmount *cifs_dfs_d_automount(struct path *path)
-{
- struct vfsmount *newmnt;
-
- cFYI(1, "in %s", __func__);
-
- newmnt = cifs_dfs_do_automount(path->dentry);
- if (IS_ERR(newmnt)) {
- cFYI(1, "leaving %s [automount failed]" , __func__);
- return newmnt;
- }
-
- mntget(newmnt); /* prevent immediate expiration */
- mnt_set_expiry(newmnt, &cifs_dfs_automount_list);
- schedule_delayed_work(&cifs_dfs_automount_task,
- cifs_dfs_mountpoint_expiry_timeout);
- cFYI(1, "leaving %s [ok]" , __func__);
- return newmnt;
-}
-
const struct inode_operations cifs_dfs_referral_inode_operations = {
};
diff --git a/fs/cifs/cifsfs.c b/fs/cifs/cifsfs.c
index f297013..c91ae83 100644
--- a/fs/cifs/cifsfs.c
+++ b/fs/cifs/cifsfs.c
@@ -1067,9 +1067,6 @@ exit_cifs(void)
cFYI(DBG2, "exit_cifs");
cifs_proc_clean();
cifs_fscache_unregister();
-#ifdef CONFIG_CIFS_DFS_UPCALL
- cifs_dfs_release_automount_timer();
-#endif
#ifdef CONFIG_CIFS_UPCALL
unregister_key_type(&cifs_spnego_key_type);
#endif
diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
index 8096f27..f3f8099 100644
--- a/fs/cifs/cifsproto.h
+++ b/fs/cifs/cifsproto.h
@@ -147,7 +147,6 @@ extern struct cifs_ntsd *get_cifs_acl(struct cifs_sb_info *, struct inode *,
extern int cifs_mount(struct super_block *, struct cifs_sb_info *, char *,
const char *);
extern int cifs_umount(struct super_block *, struct cifs_sb_info *);
-extern void cifs_dfs_release_automount_timer(void);
void cifs_proc_init(void);
void cifs_proc_clean(void);

diff --git a/fs/namei.c b/fs/namei.c
index 97c92a2..aac0375 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -923,11 +923,13 @@ static int follow_automount(struct path *path, unsigned flags,
return 0;

err = finish_automount(mnt, path);
-
switch (err) {
case -EBUSY:
/* Someone else made a mount here whilst we were busy */
- return 0;
+ err = 0;
+ default:
+ mntput(mnt);
+ return err;
case 0:
dput(path->dentry);
if (*need_mntput)
@@ -936,8 +938,6 @@ static int follow_automount(struct path *path, unsigned flags,
path->dentry = dget(mnt->mnt_root);
*need_mntput = true;
return 0;
- default:
- return err;
}

}
diff --git a/fs/namespace.c b/fs/namespace.c
index d1edf26..b9cb274 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -50,6 +50,11 @@ static struct list_head *mount_hashtable __read_mostly;
static struct kmem_cache *mnt_cache __read_mostly;
static struct rw_semaphore namespace_sem;

+unsigned long mnt_expiry_period = 15;
+static LIST_HEAD(mnt_expirable_mounts);
+static void do_periodic_mount_expiry(struct work_struct *);
+static DECLARE_DELAYED_WORK(periodic_mount_expiry, do_periodic_mount_expiry);
+
/* /sys/fs */
struct kobject *fs_kobj;
EXPORT_SYMBOL_GPL(fs_kobj);
@@ -1882,7 +1887,6 @@ static int do_new_mount(struct path *path, char *type, int flags,
int mnt_flags, char *name, void *data)
{
struct vfsmount *mnt;
- int err;

if (!type)
return -EINVAL;
@@ -1895,45 +1899,26 @@ static int do_new_mount(struct path *path, char *type, int flags,
if (IS_ERR(mnt))
return PTR_ERR(mnt);

- err = do_add_mount(mnt, path, mnt_flags);
- if (err)
- mntput(mnt);
- return err;
+ return do_add_mount(mnt, path, mnt_flags);
}

+/*
+ * Mount the given mount on the specified mountpoint.
+ * - does not drop the caller's ref from the mount.
+ */
int finish_automount(struct vfsmount *m, struct path *path)
{
- int err;
- /* The new mount record should have at least 2 refs to prevent it being
- * expired before we get a chance to add it
- */
- BUG_ON(mnt_get_count(m) < 2);
-
if (m->mnt_sb == path->mnt->mnt_sb &&
- m->mnt_root == path->dentry) {
- err = -ELOOP;
- goto fail;
- }
+ m->mnt_root == path->dentry)
+ return -ELOOP;

- err = do_add_mount(m, path, path->mnt->mnt_flags | MNT_SHRINKABLE);
- if (!err)
- return 0;
-fail:
- /* remove m from any expiration list it may be on */
- if (!list_empty(&m->mnt_expire)) {
- down_write(&namespace_sem);
- br_write_lock(vfsmount_lock);
- list_del_init(&m->mnt_expire);
- br_write_unlock(vfsmount_lock);
- up_write(&namespace_sem);
- }
- mntput(m);
- mntput(m);
- return err;
+ mntget(m);
+ return do_add_mount(m, path, path->mnt->mnt_flags | MNT_SHRINKABLE);
}

/*
- * add a mount into a namespace's mount tree
+ * Add a mount into a namespace's mount tree.
+ * - the caller's ref on the new mount is consumed unconditionally.
*/
static int do_add_mount(struct vfsmount *newmnt, struct path *path, int mnt_flags)
{
@@ -1942,62 +1927,59 @@ static int do_add_mount(struct vfsmount *newmnt, struct path *path, int mnt_flag
mnt_flags &= ~(MNT_SHARED | MNT_WRITE_HOLD | MNT_INTERNAL);

down_write(&namespace_sem);
- /* Something was mounted here while we slept */
+ /* Something may have been mounted here while we slept */
err = follow_down(path, true);
if (err < 0)
- goto unlock;
+ goto error;

err = -EINVAL;
if (!(mnt_flags & MNT_SHRINKABLE) && !check_mnt(path->mnt))
- goto unlock;
+ goto error;

/* Refuse the same filesystem on the same mount point */
err = -EBUSY;
if (path->mnt->mnt_sb == newmnt->mnt_sb &&
path->mnt->mnt_root == path->dentry)
- goto unlock;
+ goto error;

err = -EINVAL;
if (S_ISLNK(newmnt->mnt_root->d_inode->i_mode))
- goto unlock;
+ goto error;

newmnt->mnt_flags = mnt_flags;
err = graft_tree(newmnt, path);
+ if (err < 0)
+ goto error;

-unlock:
+ if (newmnt->mnt_expiry_mark) {
+ newmnt->mnt_expiry_mark = 0;
+ br_write_lock(vfsmount_lock);
+ list_add_tail(&newmnt->mnt_expire, &mnt_expirable_mounts);
+ br_write_unlock(vfsmount_lock);
+ schedule_delayed_work(&periodic_mount_expiry,
+ mnt_expiry_period * HZ);
+ }
up_write(&namespace_sem);
- return err;
-}
-
-/**
- * mnt_set_expiry - Put a mount on an expiration list
- * @mnt: The mount to list.
- * @expiry_list: The list to add the mount to.
- */
-void mnt_set_expiry(struct vfsmount *mnt, struct list_head *expiry_list)
-{
- down_write(&namespace_sem);
- br_write_lock(vfsmount_lock);
-
- list_add_tail(&mnt->mnt_expire, expiry_list);
+ return 0;

- br_write_unlock(vfsmount_lock);
+error:
up_write(&namespace_sem);
+ mntput(newmnt);
+ return err;
}
-EXPORT_SYMBOL(mnt_set_expiry);

/*
- * process a list of expirable mountpoints with the intent of discarding any
- * mountpoints that aren't in use and haven't been touched since last we came
- * here
+ * Periodically process the list of expirable mountpoints with the intent of
+ * discarding any mountpoints that aren't in use and haven't been touched since
+ * last we came here.
*/
-void mark_mounts_for_expiry(struct list_head *mounts)
+static void do_periodic_mount_expiry(struct work_struct *work)
{
struct vfsmount *mnt, *next;
LIST_HEAD(graveyard);
LIST_HEAD(umounts);

- if (list_empty(mounts))
+ if (list_empty(&mnt_expirable_mounts))
return;

down_write(&namespace_sem);
@@ -2009,7 +1991,7 @@ void mark_mounts_for_expiry(struct list_head *mounts)
* - still marked for expiry (marked on the last call here; marks are
* cleared by mntput())
*/
- list_for_each_entry_safe(mnt, next, mounts, mnt_expire) {
+ list_for_each_entry_safe(mnt, next, &mnt_expirable_mounts, mnt_expire) {
if (!xchg(&mnt->mnt_expiry_mark, 1) ||
propagate_mount_busy(mnt, 1))
continue;
@@ -2024,9 +2006,11 @@ void mark_mounts_for_expiry(struct list_head *mounts)
up_write(&namespace_sem);

release_mounts(&umounts);
-}

-EXPORT_SYMBOL_GPL(mark_mounts_for_expiry);
+ if (!list_empty(&mnt_expirable_mounts))
+ schedule_delayed_work(&periodic_mount_expiry,
+ mnt_expiry_period * HZ);
+}

/*
* Ripoff of 'select_parent()'
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index bd3ca32..5552ff2 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -1083,7 +1083,6 @@ void nfs_free_server(struct nfs_server *server)
nfs_free_iostats(server->io_stats);
bdi_destroy(&server->backing_dev_info);
kfree(server);
- nfs_release_automount_timer();
dprintk("<-- nfs_free_server()\n");
}

diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
index f32b860..3a4a2fa 100644
--- a/fs/nfs/namespace.c
+++ b/fs/nfs/namespace.c
@@ -19,12 +19,6 @@

#define NFSDBG_FACILITY NFSDBG_VFS

-static void nfs_expire_automounts(struct work_struct *work);
-
-static LIST_HEAD(nfs_automount_list);
-static DECLARE_DELAYED_WORK(nfs_automount_task, nfs_expire_automounts);
-int nfs_mountpoint_expiry_timeout = 500 * HZ;
-
static struct vfsmount *nfs_do_submount(const struct vfsmount *mnt_parent,
const struct dentry *dentry,
struct nfs_fh *fh,
@@ -149,11 +143,8 @@ struct vfsmount *nfs_d_automount(struct path *path)
if (IS_ERR(mnt))
goto out;

+ mnt->mnt_expiry_mark = 1;
dprintk("%s: done, success\n", __func__);
- mntget(mnt); /* prevent immediate expiration */
- mnt_set_expiry(mnt, &nfs_automount_list);
- schedule_delayed_work(&nfs_automount_task, nfs_mountpoint_expiry_timeout);
-
out:
nfs_free_fattr(fattr);
nfs_free_fhandle(fh);
@@ -169,21 +160,6 @@ const struct inode_operations nfs_mountpoint_inode_operations = {
const struct inode_operations nfs_referral_inode_operations = {
};

-static void nfs_expire_automounts(struct work_struct *work)
-{
- struct list_head *list = &nfs_automount_list;
-
- mark_mounts_for_expiry(list);
- if (!list_empty(list))
- schedule_delayed_work(&nfs_automount_task, nfs_mountpoint_expiry_timeout);
-}
-
-void nfs_release_automount_timer(void)
-{
- if (list_empty(&nfs_automount_list))
- cancel_delayed_work(&nfs_automount_task);
-}
-
/*
* Clone a mountpoint of the appropriate type
*/
diff --git a/fs/nfs/sysctl.c b/fs/nfs/sysctl.c
index 978aaeb..aa35724 100644
--- a/fs/nfs/sysctl.c
+++ b/fs/nfs/sysctl.c
@@ -43,13 +43,6 @@ static ctl_table nfs_cb_sysctls[] = {
#endif /* CONFIG_NFS_USE_NEW_IDMAPPER */
#endif
{
- .procname = "nfs_mountpoint_timeout",
- .data = &nfs_mountpoint_expiry_timeout,
- .maxlen = sizeof(nfs_mountpoint_expiry_timeout),
- .mode = 0644,
- .proc_handler = proc_dointvec_jiffies,
- },
- {
.procname = "nfs_congestion_kb",
.data = &nfs_congestion_kb,
.maxlen = sizeof(nfs_congestion_kb),
diff --git a/include/linux/mount.h b/include/linux/mount.h
index 604f122..0e46c54 100644
--- a/include/linux/mount.h
+++ b/include/linux/mount.h
@@ -108,9 +108,8 @@ extern struct vfsmount *vfs_kern_mount(struct file_system_type *type,
int flags, const char *name,
void *data);

-extern void mnt_set_expiry(struct vfsmount *mnt, struct list_head *expiry_list);
-extern void mark_mounts_for_expiry(struct list_head *mounts);
-
extern dev_t name_to_dev_t(char *name);

+extern unsigned long mnt_expiry_period;
+
#endif /* _LINUX_MOUNT_H */
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 0f1bd83..f132dbd 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -56,6 +56,7 @@
#include <linux/kprobes.h>
#include <linux/pipe_fs_i.h>
#include <linux/oom.h>
+#include <linux/mount.h>

#include <asm/uaccess.h>
#include <asm/processor.h>
@@ -122,6 +123,7 @@ static int one_hundred = 100;
#ifdef CONFIG_PRINTK
static int ten_thousand = 10000;
#endif
+static unsigned long max_timeout_ul = ULONG_MAX / HZ;

/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
@@ -1484,6 +1486,15 @@ static struct ctl_table fs_table[] = {
.proc_handler = &pipe_proc_fn,
.extra1 = &pipe_min_size,
},
+ {
+ .procname = "mount-expiry-period",
+ .data = &mnt_expiry_period,
+ .maxlen = sizeof(mnt_expiry_period),
+ .mode = 0644,
+ .proc_handler = proc_doulongvec_minmax,
+ .extra1 = &one_ul,
+ .extra2 = &max_timeout_ul,
+ },
{ }
};


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/