From 5f57cbcc02cf18f6b22ef4066bb10afeb8f930ff Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Fri, 7 Jan 2011 17:49:54 +1100 Subject: fs: dcache remove d_mounted Rather than keep a d_mounted count in the dentry, set a dentry flag instead. The flag can be cleared by checking the hash table to see if there are any mounts left, which is not time critical because it is performed at detach time. The mounted state of a dentry is only used to speculatively take a look in the mount hash table if it is set -- before following the mount, vfsmount lock is taken and mount re-checked without races. This saves 4 bytes on 32-bit, nothing on 64-bit but it does provide a hole I might use later (and some configs have larger than 32-bit spinlocks which might make use of the hole). Autofs4 conversion and changelog by Ian Kent : In autofs4, when expring direct (or offset) mounts we need to ensure that we block user path walks into the autofs mount, which is covered by another mount. To do this we clear the mounted status so that follows stop before walking into the mount and are essentially blocked until the expire is completed. The automount daemon still finds the correct dentry for the umount due to the follow mount logic in fs/autofs4/root.c:autofs4_follow_link(), which is set as an inode operation for direct and offset mounts only and is called following the lookup that stopped at the covered mount. At the end of the expire the covering mount probably has gone away so the mounted status need not be restored. But we need to check this and only restore the mounted status if the expire failed. XXX: autofs may not work right if we have other mounts go over the top of it? Signed-off-by: Nick Piggin --- fs/namespace.c | 29 ++++++++++++++++++++++++++--- 1 file changed, 26 insertions(+), 3 deletions(-) (limited to 'fs/namespace.c') diff --git a/fs/namespace.c b/fs/namespace.c index 3dbfc072ec70..39a7d507fba0 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -491,6 +491,27 @@ static void __touch_mnt_namespace(struct mnt_namespace *ns) } } +/* + * Clear dentry's mounted state if it has no remaining mounts. + * vfsmount_lock must be held for write. + */ +static void dentry_reset_mounted(struct vfsmount *mnt, struct dentry *dentry) +{ + unsigned u; + + for (u = 0; u < HASH_SIZE; u++) { + struct vfsmount *p; + + list_for_each_entry(p, &mount_hashtable[u], mnt_hash) { + if (p->mnt_mountpoint == dentry) + return; + } + } + spin_lock(&dentry->d_lock); + dentry->d_flags &= ~DCACHE_MOUNTED; + spin_unlock(&dentry->d_lock); +} + /* * vfsmount lock must be held for write */ @@ -502,7 +523,7 @@ static void detach_mnt(struct vfsmount *mnt, struct path *old_path) mnt->mnt_mountpoint = mnt->mnt_root; list_del_init(&mnt->mnt_child); list_del_init(&mnt->mnt_hash); - old_path->dentry->d_mounted--; + dentry_reset_mounted(old_path->mnt, old_path->dentry); } /* @@ -513,7 +534,9 @@ void mnt_set_mountpoint(struct vfsmount *mnt, struct dentry *dentry, { child_mnt->mnt_parent = mntget(mnt); child_mnt->mnt_mountpoint = dget(dentry); - dentry->d_mounted++; + spin_lock(&dentry->d_lock); + dentry->d_flags |= DCACHE_MOUNTED; + spin_unlock(&dentry->d_lock); } /* @@ -1073,7 +1096,7 @@ void umount_tree(struct vfsmount *mnt, int propagate, struct list_head *kill) list_del_init(&p->mnt_child); if (p->mnt_parent != p) { p->mnt_parent->mnt_ghosts++; - p->mnt_mountpoint->d_mounted--; + dentry_reset_mounted(p->mnt_parent, p->mnt_mountpoint); } change_mnt_propagation(p, MS_PRIVATE); } -- cgit v1.2.2 From c6653a838b1b2738561aff0b8c0f62a9b714bdd9 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Fri, 7 Jan 2011 17:50:10 +1100 Subject: fs: rename vfsmount counter helpers Suggested by Andreas, mnt_ prefix is clearer namespace, follows kernel conventions better, and is easier for tab complete. I introduced these names so I'll admit they were not good choices. Signed-off-by: Nick Piggin --- fs/namespace.c | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) (limited to 'fs/namespace.c') diff --git a/fs/namespace.c b/fs/namespace.c index 39a7d507fba0..03b82350f020 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -216,7 +216,7 @@ int __mnt_is_readonly(struct vfsmount *mnt) } EXPORT_SYMBOL_GPL(__mnt_is_readonly); -static inline void inc_mnt_writers(struct vfsmount *mnt) +static inline void mnt_inc_writers(struct vfsmount *mnt) { #ifdef CONFIG_SMP (*per_cpu_ptr(mnt->mnt_writers, smp_processor_id()))++; @@ -225,7 +225,7 @@ static inline void inc_mnt_writers(struct vfsmount *mnt) #endif } -static inline void dec_mnt_writers(struct vfsmount *mnt) +static inline void mnt_dec_writers(struct vfsmount *mnt) { #ifdef CONFIG_SMP (*per_cpu_ptr(mnt->mnt_writers, smp_processor_id()))--; @@ -234,7 +234,7 @@ static inline void dec_mnt_writers(struct vfsmount *mnt) #endif } -static unsigned int count_mnt_writers(struct vfsmount *mnt) +static unsigned int mnt_get_writers(struct vfsmount *mnt) { #ifdef CONFIG_SMP unsigned int count = 0; @@ -273,9 +273,9 @@ int mnt_want_write(struct vfsmount *mnt) int ret = 0; preempt_disable(); - inc_mnt_writers(mnt); + mnt_inc_writers(mnt); /* - * The store to inc_mnt_writers must be visible before we pass + * The store to mnt_inc_writers must be visible before we pass * MNT_WRITE_HOLD loop below, so that the slowpath can see our * incremented count after it has set MNT_WRITE_HOLD. */ @@ -289,7 +289,7 @@ int mnt_want_write(struct vfsmount *mnt) */ smp_rmb(); if (__mnt_is_readonly(mnt)) { - dec_mnt_writers(mnt); + mnt_dec_writers(mnt); ret = -EROFS; goto out; } @@ -317,7 +317,7 @@ int mnt_clone_write(struct vfsmount *mnt) if (__mnt_is_readonly(mnt)) return -EROFS; preempt_disable(); - inc_mnt_writers(mnt); + mnt_inc_writers(mnt); preempt_enable(); return 0; } @@ -351,7 +351,7 @@ EXPORT_SYMBOL_GPL(mnt_want_write_file); void mnt_drop_write(struct vfsmount *mnt) { preempt_disable(); - dec_mnt_writers(mnt); + mnt_dec_writers(mnt); preempt_enable(); } EXPORT_SYMBOL_GPL(mnt_drop_write); @@ -384,7 +384,7 @@ static int mnt_make_readonly(struct vfsmount *mnt) * MNT_WRITE_HOLD, so it can't be decremented by another CPU while * we're counting up here. */ - if (count_mnt_writers(mnt) > 0) + if (mnt_get_writers(mnt) > 0) ret = -EBUSY; else mnt->mnt_flags |= MNT_READONLY; @@ -663,9 +663,9 @@ static inline void __mntput(struct vfsmount *mnt) */ /* * atomic_dec_and_lock() used to deal with ->mnt_count decrements - * provides barriers, so count_mnt_writers() below is safe. AV + * provides barriers, so mnt_get_writers() below is safe. AV */ - WARN_ON(count_mnt_writers(mnt)); + WARN_ON(mnt_get_writers(mnt)); fsnotify_vfsmount_delete(mnt); dput(mnt->mnt_root); free_vfsmnt(mnt); -- cgit v1.2.2 From b3e19d924b6eaf2ca7d22cba99a517c5171007b6 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Fri, 7 Jan 2011 17:50:11 +1100 Subject: fs: scale mntget/mntput The problem that this patch aims to fix is vfsmount refcounting scalability. We need to take a reference on the vfsmount for every successful path lookup, which often go to the same mount point. The fundamental difficulty is that a "simple" reference count can never be made scalable, because any time a reference is dropped, we must check whether that was the last reference. To do that requires communication with all other CPUs that may have taken a reference count. We can make refcounts more scalable in a couple of ways, involving keeping distributed counters, and checking for the global-zero condition less frequently. - check the global sum once every interval (this will delay zero detection for some interval, so it's probably a showstopper for vfsmounts). - keep a local count and only taking the global sum when local reaches 0 (this is difficult for vfsmounts, because we can't hold preempt off for the life of a reference, so a counter would need to be per-thread or tied strongly to a particular CPU which requires more locking). - keep a local difference of increments and decrements, which allows us to sum the total difference and hence find the refcount when summing all CPUs. Then, keep a single integer "long" refcount for slow and long lasting references, and only take the global sum of local counters when the long refcount is 0. This last scheme is what I implemented here. Attached mounts and process root and working directory references are "long" references, and everything else is a short reference. This allows scalable vfsmount references during path walking over mounted subtrees and unattached (lazy umounted) mounts with processes still running in them. This results in one fewer atomic op in the fastpath: mntget is now just a per-CPU inc, rather than an atomic inc; and mntput just requires a spinlock and non-atomic decrement in the common case. However code is otherwise bigger and heavier, so single threaded performance is basically a wash. Signed-off-by: Nick Piggin --- fs/namespace.c | 242 +++++++++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 199 insertions(+), 43 deletions(-) (limited to 'fs/namespace.c') diff --git a/fs/namespace.c b/fs/namespace.c index 03b82350f020..3ddfd9046c44 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -138,6 +138,64 @@ void mnt_release_group_id(struct vfsmount *mnt) mnt->mnt_group_id = 0; } +/* + * vfsmount lock must be held for read + */ +static inline void mnt_add_count(struct vfsmount *mnt, int n) +{ +#ifdef CONFIG_SMP + this_cpu_add(mnt->mnt_pcp->mnt_count, n); +#else + preempt_disable(); + mnt->mnt_count += n; + preempt_enable(); +#endif +} + +static inline void mnt_set_count(struct vfsmount *mnt, int n) +{ +#ifdef CONFIG_SMP + this_cpu_write(mnt->mnt_pcp->mnt_count, n); +#else + mnt->mnt_count = n; +#endif +} + +/* + * vfsmount lock must be held for read + */ +static inline void mnt_inc_count(struct vfsmount *mnt) +{ + mnt_add_count(mnt, 1); +} + +/* + * vfsmount lock must be held for read + */ +static inline void mnt_dec_count(struct vfsmount *mnt) +{ + mnt_add_count(mnt, -1); +} + +/* + * vfsmount lock must be held for write + */ +unsigned int mnt_get_count(struct vfsmount *mnt) +{ +#ifdef CONFIG_SMP + unsigned int count = atomic_read(&mnt->mnt_longrefs); + int cpu; + + for_each_possible_cpu(cpu) { + count += per_cpu_ptr(mnt->mnt_pcp, cpu)->mnt_count; + } + + return count; +#else + return mnt->mnt_count; +#endif +} + struct vfsmount *alloc_vfsmnt(const char *name) { struct vfsmount *mnt = kmem_cache_zalloc(mnt_cache, GFP_KERNEL); @@ -154,7 +212,17 @@ struct vfsmount *alloc_vfsmnt(const char *name) goto out_free_id; } - atomic_set(&mnt->mnt_count, 1); +#ifdef CONFIG_SMP + mnt->mnt_pcp = alloc_percpu(struct mnt_pcp); + if (!mnt->mnt_pcp) + goto out_free_devname; + + atomic_set(&mnt->mnt_longrefs, 1); +#else + mnt->mnt_count = 1; + mnt->mnt_writers = 0; +#endif + INIT_LIST_HEAD(&mnt->mnt_hash); INIT_LIST_HEAD(&mnt->mnt_child); INIT_LIST_HEAD(&mnt->mnt_mounts); @@ -165,13 +233,6 @@ struct vfsmount *alloc_vfsmnt(const char *name) INIT_LIST_HEAD(&mnt->mnt_slave); #ifdef CONFIG_FSNOTIFY INIT_HLIST_HEAD(&mnt->mnt_fsnotify_marks); -#endif -#ifdef CONFIG_SMP - mnt->mnt_writers = alloc_percpu(int); - if (!mnt->mnt_writers) - goto out_free_devname; -#else - mnt->mnt_writers = 0; #endif } return mnt; @@ -219,7 +280,7 @@ EXPORT_SYMBOL_GPL(__mnt_is_readonly); static inline void mnt_inc_writers(struct vfsmount *mnt) { #ifdef CONFIG_SMP - (*per_cpu_ptr(mnt->mnt_writers, smp_processor_id()))++; + this_cpu_inc(mnt->mnt_pcp->mnt_writers); #else mnt->mnt_writers++; #endif @@ -228,7 +289,7 @@ static inline void mnt_inc_writers(struct vfsmount *mnt) static inline void mnt_dec_writers(struct vfsmount *mnt) { #ifdef CONFIG_SMP - (*per_cpu_ptr(mnt->mnt_writers, smp_processor_id()))--; + this_cpu_dec(mnt->mnt_pcp->mnt_writers); #else mnt->mnt_writers--; #endif @@ -241,7 +302,7 @@ static unsigned int mnt_get_writers(struct vfsmount *mnt) int cpu; for_each_possible_cpu(cpu) { - count += *per_cpu_ptr(mnt->mnt_writers, cpu); + count += per_cpu_ptr(mnt->mnt_pcp, cpu)->mnt_writers; } return count; @@ -418,7 +479,7 @@ void free_vfsmnt(struct vfsmount *mnt) kfree(mnt->mnt_devname); mnt_free_id(mnt); #ifdef CONFIG_SMP - free_percpu(mnt->mnt_writers); + free_percpu(mnt->mnt_pcp); #endif kmem_cache_free(mnt_cache, mnt); } @@ -652,9 +713,10 @@ static struct vfsmount *clone_mnt(struct vfsmount *old, struct dentry *root, return NULL; } -static inline void __mntput(struct vfsmount *mnt) +static inline void mntfree(struct vfsmount *mnt) { struct super_block *sb = mnt->mnt_sb; + /* * This probably indicates that somebody messed * up a mnt_want/drop_write() pair. If this @@ -662,8 +724,8 @@ static inline void __mntput(struct vfsmount *mnt) * to make r/w->r/o transitions. */ /* - * atomic_dec_and_lock() used to deal with ->mnt_count decrements - * provides barriers, so mnt_get_writers() below is safe. AV + * The locking used to deal with mnt_count decrement provides barriers, + * so mnt_get_writers() below is safe. */ WARN_ON(mnt_get_writers(mnt)); fsnotify_vfsmount_delete(mnt); @@ -672,28 +734,113 @@ static inline void __mntput(struct vfsmount *mnt) deactivate_super(sb); } -void mntput_no_expire(struct vfsmount *mnt) -{ -repeat: - if (atomic_add_unless(&mnt->mnt_count, -1, 1)) - return; +#ifdef CONFIG_SMP +static inline void __mntput(struct vfsmount *mnt, int longrefs) +{ + if (!longrefs) { +put_again: + br_read_lock(vfsmount_lock); + if (likely(atomic_read(&mnt->mnt_longrefs))) { + mnt_dec_count(mnt); + br_read_unlock(vfsmount_lock); + return; + } + br_read_unlock(vfsmount_lock); + } else { + BUG_ON(!atomic_read(&mnt->mnt_longrefs)); + if (atomic_add_unless(&mnt->mnt_longrefs, -1, 1)) + return; + } + br_write_lock(vfsmount_lock); - if (!atomic_dec_and_test(&mnt->mnt_count)) { + if (!longrefs) + mnt_dec_count(mnt); + else + atomic_dec(&mnt->mnt_longrefs); + if (mnt_get_count(mnt)) { br_write_unlock(vfsmount_lock); return; } - if (likely(!mnt->mnt_pinned)) { + if (unlikely(mnt->mnt_pinned)) { + mnt_add_count(mnt, mnt->mnt_pinned + 1); + mnt->mnt_pinned = 0; br_write_unlock(vfsmount_lock); - __mntput(mnt); + acct_auto_close_mnt(mnt); + goto put_again; + } + br_write_unlock(vfsmount_lock); + mntfree(mnt); +} +#else +static inline void __mntput(struct vfsmount *mnt, int longrefs) +{ +put_again: + mnt_dec_count(mnt); + if (likely(mnt_get_count(mnt))) return; + br_write_lock(vfsmount_lock); + if (unlikely(mnt->mnt_pinned)) { + mnt_add_count(mnt, mnt->mnt_pinned + 1); + mnt->mnt_pinned = 0; + br_write_unlock(vfsmount_lock); + acct_auto_close_mnt(mnt); + goto put_again; } - atomic_add(mnt->mnt_pinned + 1, &mnt->mnt_count); - mnt->mnt_pinned = 0; br_write_unlock(vfsmount_lock); - acct_auto_close_mnt(mnt); - goto repeat; + mntfree(mnt); +} +#endif + +static void mntput_no_expire(struct vfsmount *mnt) +{ + __mntput(mnt, 0); +} + +void mntput(struct vfsmount *mnt) +{ + if (mnt) { + /* avoid cacheline pingpong, hope gcc doesn't get "smart" */ + if (unlikely(mnt->mnt_expiry_mark)) + mnt->mnt_expiry_mark = 0; + __mntput(mnt, 0); + } +} +EXPORT_SYMBOL(mntput); + +struct vfsmount *mntget(struct vfsmount *mnt) +{ + if (mnt) + mnt_inc_count(mnt); + return mnt; +} +EXPORT_SYMBOL(mntget); + +void mntput_long(struct vfsmount *mnt) +{ +#ifdef CONFIG_SMP + if (mnt) { + /* avoid cacheline pingpong, hope gcc doesn't get "smart" */ + if (unlikely(mnt->mnt_expiry_mark)) + mnt->mnt_expiry_mark = 0; + __mntput(mnt, 1); + } +#else + mntput(mnt); +#endif } -EXPORT_SYMBOL(mntput_no_expire); +EXPORT_SYMBOL(mntput_long); + +struct vfsmount *mntget_long(struct vfsmount *mnt) +{ +#ifdef CONFIG_SMP + if (mnt) + atomic_inc(&mnt->mnt_longrefs); + return mnt; +#else + return mntget(mnt); +#endif +} +EXPORT_SYMBOL(mntget_long); void mnt_pin(struct vfsmount *mnt) { @@ -701,19 +848,17 @@ void mnt_pin(struct vfsmount *mnt) mnt->mnt_pinned++; br_write_unlock(vfsmount_lock); } - EXPORT_SYMBOL(mnt_pin); void mnt_unpin(struct vfsmount *mnt) { br_write_lock(vfsmount_lock); if (mnt->mnt_pinned) { - atomic_inc(&mnt->mnt_count); + mnt_inc_count(mnt); mnt->mnt_pinned--; } br_write_unlock(vfsmount_lock); } - EXPORT_SYMBOL(mnt_unpin); static inline void mangle(struct seq_file *m, const char *s) @@ -1008,12 +1153,13 @@ int may_umount_tree(struct vfsmount *mnt) int minimum_refs = 0; struct vfsmount *p; - br_read_lock(vfsmount_lock); + /* write lock needed for mnt_get_count */ + br_write_lock(vfsmount_lock); for (p = mnt; p; p = next_mnt(p, mnt)) { - actual_refs += atomic_read(&p->mnt_count); + actual_refs += mnt_get_count(p); minimum_refs += 2; } - br_read_unlock(vfsmount_lock); + br_write_unlock(vfsmount_lock); if (actual_refs > minimum_refs) return 0; @@ -1040,10 +1186,10 @@ int may_umount(struct vfsmount *mnt) { int ret = 1; down_read(&namespace_sem); - br_read_lock(vfsmount_lock); + br_write_lock(vfsmount_lock); if (propagate_mount_busy(mnt, 2)) ret = 0; - br_read_unlock(vfsmount_lock); + br_write_unlock(vfsmount_lock); up_read(&namespace_sem); return ret; } @@ -1070,7 +1216,7 @@ void release_mounts(struct list_head *head) dput(dentry); mntput(m); } - mntput(mnt); + mntput_long(mnt); } } @@ -1125,8 +1271,16 @@ static int do_umount(struct vfsmount *mnt, int flags) flags & (MNT_FORCE | MNT_DETACH)) return -EINVAL; - if (atomic_read(&mnt->mnt_count) != 2) + /* + * probably don't strictly need the lock here if we examined + * all race cases, but it's a slowpath. + */ + br_write_lock(vfsmount_lock); + if (mnt_get_count(mnt) != 2) { + br_write_lock(vfsmount_lock); return -EBUSY; + } + br_write_unlock(vfsmount_lock); if (!xchg(&mnt->mnt_expiry_mark, 1)) return -EAGAIN; @@ -1815,7 +1969,7 @@ int do_add_mount(struct vfsmount *newmnt, struct path *path, unlock: up_write(&namespace_sem); - mntput(newmnt); + mntput_long(newmnt); return err; } @@ -2148,11 +2302,11 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns, if (fs) { if (p == fs->root.mnt) { rootmnt = p; - fs->root.mnt = mntget(q); + fs->root.mnt = mntget_long(q); } if (p == fs->pwd.mnt) { pwdmnt = p; - fs->pwd.mnt = mntget(q); + fs->pwd.mnt = mntget_long(q); } } p = next_mnt(p, mnt_ns->root); @@ -2161,9 +2315,9 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns, up_write(&namespace_sem); if (rootmnt) - mntput(rootmnt); + mntput_long(rootmnt); if (pwdmnt) - mntput(pwdmnt); + mntput_long(pwdmnt); return new_ns; } @@ -2350,6 +2504,7 @@ SYSCALL_DEFINE2(pivot_root, const char __user *, new_root, touch_mnt_namespace(current->nsproxy->mnt_ns); br_write_unlock(vfsmount_lock); chroot_fs_refs(&root, &new); + error = 0; path_put(&root_parent); path_put(&parent_path); @@ -2376,6 +2531,7 @@ static void __init init_mount_tree(void) mnt = do_kern_mount("rootfs", 0, "rootfs", NULL); if (IS_ERR(mnt)) panic("Can't create rootfs"); + ns = create_mnt_ns(mnt); if (IS_ERR(ns)) panic("Can't allocate initial namespace"); -- cgit v1.2.2