/* * Copyright (C) 2008 Red Hat, Inc., Eric Paris * * 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, 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; see the file COPYING. If not, write to * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ #include #include #include #include #include #include #include #include #include "fsnotify.h" #ifndef CONFIG_TRACE_FSNOTIFY_EVENTS # define NOTRACE #endif #define CREATE_TRACE_POINTS #include "trace-fsnotify.h" /* * Clear all of the marks on an inode when it is being evicted from core */ void __fsnotify_inode_delete(struct inode *inode) { fsnotify_clear_marks_by_inode(inode); } EXPORT_SYMBOL_GPL(__fsnotify_inode_delete); void __fsnotify_vfsmount_delete(struct vfsmount *mnt) { fsnotify_clear_marks_by_mount(mnt); } /** * fsnotify_unmount_inodes - an sb is unmounting. handle any watched inodes. * @sb: superblock being unmounted. * * Called during unmount with no locks held, so needs to be safe against * concurrent modifiers. We temporarily drop sb->s_inode_list_lock and CAN block. */ void fsnotify_unmount_inodes(struct super_block *sb) { struct inode *inode, *iput_inode = NULL; spin_lock(&sb->s_inode_list_lock); list_for_each_entry(inode, &sb->s_inodes, i_sb_list) { /* * We cannot __iget() an inode in state I_FREEING, * I_WILL_FREE, or I_NEW which is fine because by that point * the inode cannot have any associated watches. */ spin_lock(&inode->i_lock); if (inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW)) { spin_unlock(&inode->i_lock); continue; } /* * If i_count is zero, the inode cannot have any watches and * doing an __iget/iput with SB_ACTIVE clear would actually * evict all inodes with zero i_count from icache which is * unnecessarily violent and may in fact be illegal to do. */ if (!atomic_read(&inode->i_count)) { spin_unlock(&inode->i_lock); continue; } __iget(inode); spin_unlock(&inode->i_lock); spin_unlock(&sb->s_inode_list_lock); if (iput_inode) iput(iput_inode); /* for each watch, send FS_UNMOUNT and then remove it */ fsnotify(inode, FS_UNMOUNT, inode, FSNOTIFY_EVENT_INODE, NULL, 0); fsnotify_inode_delete(inode); iput_inode = inode; cond_resched(); spin_lock(&sb->s_inode_list_lock); } spin_unlock(&sb->s_inode_list_lock); if (iput_inode) iput(iput_inode); /* Wait for outstanding inode references from connectors */ wait_var_event(&sb->s_fsnotify_inode_refs, !atomic_long_read(&sb->s_fsnotify_inode_refs)); } #if defined(CONFIG_FSNOTIFY_RECURSIVE) static void __fsnotify_update_child_dentry_flags_recursivly(struct dentry *dentry) { struct dentry *child; int watched_recursivly = 0; if (dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY) watched_recursivly = 1; list_for_each_entry(child, &dentry->d_subdirs, d_child) { if (!child->d_inode) continue; spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED); if (watched_recursivly) child->d_flags |= DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY; else child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY; __fsnotify_update_child_dentry_flags_recursivly(child); spin_unlock(&child->d_lock); } } #endif /* CONFIG_FSNOTIFY_RECURSIVE */ /* * Given an inode, first check if we care what happens to our children. Inotify * and dnotify both tell their parents about events. If we care about any event * on a child we run all of our children and set a dentry flag saying that the * parent cares. Thus when an event happens on a child it can quickly tell if * if there is a need to find a parent and send the event to the parent. */ void __fsnotify_update_child_dentry_flags(struct inode *inode) { struct dentry *alias; int watched; #if defined(CONFIG_FSNOTIFY_RECURSIVE) int watched_recursivly; #endif /* CONFIG_FSNOTIFY_RECURSIVE */ if (!S_ISDIR(inode->i_mode)) return; /* determine if the children should tell inode about their events */ watched = fsnotify_inode_watches_children(inode); #if defined(CONFIG_FSNOTIFY_RECURSIVE) watched_recursivly = fsnotify_inode_watches_children_recursivly(inode); #endif /* CONFIG_FSNOTIFY_RECURSIVE */ spin_lock(&inode->i_lock); /* run all of the dentries associated with this inode. Since this is a * directory, there damn well better only be one item on this list */ hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) { struct dentry *child; #if defined(CONFIG_FSNOTIFY_RECURSIVE) if (alias->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY) watched_recursivly = 1; #endif /* CONFIG_FSNOTIFY_RECURSIVE */ /* run all of the children of the original inode and fix their * d_flags to indicate parental interest (their parent is the * original inode) */ spin_lock(&alias->d_lock); list_for_each_entry(child, &alias->d_subdirs, d_child) { if (!child->d_inode) continue; spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED); if (watched) child->d_flags |= DCACHE_FSNOTIFY_PARENT_WATCHED; else child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED; #if defined(CONFIG_FSNOTIFY_RECURSIVE) if (watched_recursivly) child->d_flags |= DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY; else child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY; __fsnotify_update_child_dentry_flags_recursivly(child); #endif /* CONFIG_FSNOTIFY_RECURSIVE */ spin_unlock(&child->d_lock); } spin_unlock(&alias->d_lock); } spin_unlock(&inode->i_lock); } /* Notify this dentry's parent about a child's events. */ int __fsnotify_parent(const struct path *path, struct dentry *dentry, __u32 mask) { struct dentry *parent; struct inode *p_inode; int ret = 0; if (!dentry) dentry = path->dentry; if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED)) return 0; parent = dget_parent(dentry); p_inode = parent->d_inode; if (unlikely(!fsnotify_inode_watches_children(p_inode))) { __fsnotify_update_child_dentry_flags(p_inode); } else if (p_inode->i_fsnotify_mask & mask & ALL_FSNOTIFY_EVENTS) { struct name_snapshot name; /* we are notifying a parent so come up with the new mask which * specifies these are events which came from a child. */ mask |= FS_EVENT_ON_CHILD; take_dentry_name_snapshot(&name, dentry); if (path) ret = fsnotify(p_inode, mask, path, FSNOTIFY_EVENT_PATH, name.name, 0); else ret = fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE, name.name, 0); release_dentry_name_snapshot(&name); } dput(parent); return ret; } EXPORT_SYMBOL_GPL(__fsnotify_parent); #if defined(CONFIG_FSNOTIFY_RECURSIVE) /** * Formats a relative path beginning at an entry with the given name up to top. * * @param parent The parent of the name * @param root The parent dentry from which to start the path * @param name The name of the last component * @return A kmalloc'd string representing the path */ static char *alloc_relative_path(struct dentry *root, struct dentry *parent, const char *name) { size_t size, name_size; struct dentry *d; int seq = 0; char *buf; WARN_ON_ONCE(!rcu_read_lock_held()); name_size = strlen(name) + 1; again: size = name_size; read_seqbegin_or_lock(&rename_lock, &seq); /* * First we go upwards the tree to calculate the needed size. * __getname is supposedly faster, but wastes memory. */ for (d = parent; d != root && !IS_ROOT(d); d = READ_ONCE(d->d_parent)) size += READ_ONCE(d->d_name.len) + sizeof("/")-1; buf = kmalloc(size, GFP_ATOMIC); if (!buf) { done_seqretry(&rename_lock, seq); return ERR_PTR(-ENOMEM); } size -= name_size; memcpy(&buf[size], name, name_size); /* * Traverse the path upwards again, building the path. */ for (d = parent; d != root && !IS_ROOT(d); d = READ_ONCE(d->d_parent)) { unsigned int dlen = READ_ONCE(d->d_name.len); const char *dname = READ_ONCE(d->d_name.name); /* * Dentry names may change anytime. Here, just make sure we * don't overflow our buffer. Any inconsistencies will be * caught by the seq check below. */ if (dlen + sizeof("/")-1 > size) break; buf[--size] = '/'; size -= dlen; memcpy(&buf[size], dname, dlen); } /* * If no renames happened inbetween, we're done. Else, redo with lock. */ if (need_seqretry(&rename_lock, seq)) { kfree(buf); seq = 1; goto again; } done_seqretry(&rename_lock, seq); return buf; } /** * Find an ancestor dentry that is watching its children recursively for * selected events. * * @param dentry The dentry for which to find an ancestor * @param mask The event bitmask in question * @return An interested ancestor dentry, with a reference held, or * NULL if there are none. */ static struct dentry *__dget_watching_ancestor(struct dentry *dentry, __u32 mask) { WARN_ON_ONCE(!rcu_read_lock_held()); while (!IS_ROOT(dentry)) { struct inode *inode; dentry = READ_ONCE(dentry->d_parent); inode = READ_ONCE(dentry->d_inode); if (inode && fsnotify_inode_watches_children_recursivly(inode) && (inode->i_fsnotify_mask & mask)) return dget(dentry); if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY)) break; } return NULL; } /** * Notifies the ancestor interested. * * Walks the dentry tree upwards to find an interested ancestor and forwards the * event to it. * * DOES NOT cross fs boundaries * DOES ONLY notify the first interested ancestor dentry that is found * The immediate parent is not notified here, but by the standard * fsnotify_parent(). * * @param path path for the event * @param dentry the dentry for the event * @param parent the parent of dentry, if it is different because dentry has been moved * @param name the name of dentry, if it is different because dentry has been renamed * @param mask the event mask to pass * @param cookie the cookie to pass * @return 0 on success, !=0 otherwise */ int __fsnotify_parent_recursive(const struct path *path, struct dentry *dentry, struct dentry *parent, const unsigned char *name, __u32 mask, __u32 cookie) { struct name_snapshot d_name; struct dentry *d_to_tell; struct inode *to_tell; char *rel_path; int ret = 0; if (!dentry) dentry = path->dentry; /* * No one above us in the hierarchy is interested in our events */ if (!(dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED_RECURSIVLY)) return 0; /* * The event is not allowed to propagate upwards */ if (!(mask & FS_EVENTS_POSS_ON_CHILD_RECURSIVLY)) return 0; /* * If there is an ancestor watching recursively, we try to construct * the path from that ancestor down to the dentry that generated the * event. * * This is far from being bullet-proof. For example, since the dentry * chain can be changed anytime by concurrent renames, traversing it * need not yield reproducible results. We might fail to find a * watching ancestor, because the chain got destroyed before we came * to traverse it. In this case we lose events. * * Also, no critical section ties the fsnotify hooks to the events * that triggered them, so things can get mixed up. * * Also, maintaining the d_flags in a sub-tree is prone to races. * * But if there's not too much activity on the directory tree, * results should be good enough. Just don't count on it. ;) */ /* * Protect against dentries perishing beneath us. */ rcu_read_lock(); if (!parent) parent = READ_ONCE(dentry->d_parent); /* * Traverse upwards, looking for a watching ancestor. This skips the * parent, which is already notified non-recursively. */ d_to_tell = __dget_watching_ancestor(parent, mask); if (!d_to_tell) { rcu_read_unlock(); return 0; } /* * Try to construct the path. This traverses the path again. */ if (!name) { take_dentry_name_snapshot(&d_name, dentry); name = d_name.name; } rel_path = alloc_relative_path(d_to_tell, parent, name); if (IS_ERR(rel_path)) { ret = PTR_ERR(rel_path); rcu_read_unlock(); goto err_dput; } /* * If the existing chain has been modified inbetween, the path might * not end up at the watching dentry, but at the root. This cannot be * repaired without taking the global rename_lock. Assume that we got * it right if there is still a linking chain. Otherwise, drop the * event, since we lack useful data to report. */ if (!d_ancestor(d_to_tell, parent)) { ret = 0; rcu_read_unlock(); goto err_free; } rcu_read_unlock(); to_tell = READ_ONCE(d_to_tell->d_inode); /* * We are notifying a parent so come up with the new mask which * specifies these are events which came from a child. */ mask |= FS_EVENT_ON_CHILD_RECURSIVLY; if (path) ret = fsnotify(to_tell, mask, path, FSNOTIFY_EVENT_PATH, rel_path, cookie); else ret = fsnotify(to_tell, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE, rel_path, cookie); err_free: kfree(rel_path); err_dput: dput(d_to_tell); /* * Should we repeat, looking for more watchers upwards the path? * This would require dgetting the parent, since we drop rcu * when calling fsnotify(). */ if (name == d_name.name) release_dentry_name_snapshot(&d_name); return ret; } EXPORT_SYMBOL_GPL(__fsnotify_parent_recursive); #endif /* CONFIG_FSNOTIFY_RECURSIVE */ static int send_to_group(struct inode *to_tell, __u32 mask, const void *data, int data_is, u32 cookie, const unsigned char *file_name, struct fsnotify_iter_info *iter_info) { struct fsnotify_group *group = NULL; __u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS); __u32 marks_mask = 0; __u32 marks_ignored_mask = 0; struct fsnotify_mark *mark; int type; if (WARN_ON(!iter_info->report_mask)) return 0; /* clear ignored on inode modification */ if (mask & FS_MODIFY) { fsnotify_foreach_obj_type(type) { if (!fsnotify_iter_should_report_type(iter_info, type)) continue; mark = iter_info->marks[type]; if (mark && !(mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY)) mark->ignored_mask = 0; } } fsnotify_foreach_obj_type(type) { if (!fsnotify_iter_should_report_type(iter_info, type)) continue; mark = iter_info->marks[type]; /* does the object mark tell us to do something? */ if (mark) { group = mark->group; marks_mask |= mark->mask; marks_ignored_mask |= mark->ignored_mask; } } pr_debug("%s: group=%p to_tell=%p mask=%x marks_mask=%x marks_ignored_mask=%x" " data=%p data_is=%d cookie=%d\n", __func__, group, to_tell, mask, marks_mask, marks_ignored_mask, data, data_is, cookie); if (!(test_mask & marks_mask & ~marks_ignored_mask)) return 0; return group->ops->handle_event(group, to_tell, mask, data, data_is, file_name, cookie, iter_info); } static struct fsnotify_mark *fsnotify_first_mark(struct fsnotify_mark_connector **connp) { struct fsnotify_mark_connector *conn; struct hlist_node *node = NULL; conn = srcu_dereference(*connp, &fsnotify_mark_srcu); if (conn) node = srcu_dereference(conn->list.first, &fsnotify_mark_srcu); return hlist_entry_safe(node, struct fsnotify_mark, obj_list); } static struct fsnotify_mark *fsnotify_next_mark(struct fsnotify_mark *mark) { struct hlist_node *node = NULL; if (mark) node = srcu_dereference(mark->obj_list.next, &fsnotify_mark_srcu); return hlist_entry_safe(node, struct fsnotify_mark, obj_list); } /* * iter_info is a multi head priority queue of marks. * Pick a subset of marks from queue heads, all with the * same group and set the report_mask for selected subset. * Returns the report_mask of the selected subset. */ static unsigned int fsnotify_iter_select_report_types( struct fsnotify_iter_info *iter_info) { struct fsnotify_group *max_prio_group = NULL; struct fsnotify_mark *mark; int type; /* Choose max prio group among groups of all queue heads */ fsnotify_foreach_obj_type(type) { mark = iter_info->marks[type]; if (mark && fsnotify_compare_groups(max_prio_group, mark->group) > 0) max_prio_group = mark->group; } if (!max_prio_group) return 0; /* Set the report mask for marks from same group as max prio group */ iter_info->report_mask = 0; fsnotify_foreach_obj_type(type) { mark = iter_info->marks[type]; if (mark && fsnotify_compare_groups(max_prio_group, mark->group) == 0) fsnotify_iter_set_report_type(iter_info, type); } return iter_info->report_mask; } /* * Pop from iter_info multi head queue, the marks that were iterated in the * current iteration step. */ static void fsnotify_iter_next(struct fsnotify_iter_info *iter_info) { int type; fsnotify_foreach_obj_type(type) { if (fsnotify_iter_should_report_type(iter_info, type)) iter_info->marks[type] = fsnotify_next_mark(iter_info->marks[type]); } } /* * This is the main call to fsnotify. The VFS calls into hook specific functions * in linux/fsnotify.h. Those functions then in turn call here. Here will call * out to all of the registered fsnotify_group. Those groups can then use the * notification event in whatever means they feel necessary. */ int fsnotify(struct inode *to_tell, __u32 mask, const void *data, int data_is, const unsigned char *file_name, u32 cookie) { struct fsnotify_iter_info iter_info = {}; struct mount *mnt; int ret = 0; __u32 test_mask = (mask & ALL_FSNOTIFY_EVENTS); if (data_is == FSNOTIFY_EVENT_PATH) mnt = real_mount(((const struct path *)data)->mnt); else mnt = NULL; /* An event "on child" is not intended for a mount mark */ if (mask & FS_EVENT_ON_CHILD) mnt = NULL; /* * Optimization: srcu_read_lock() has a memory barrier which can * be expensive. It protects walking the *_fsnotify_marks lists. * However, if we do not walk the lists, we do not have to do * SRCU because we have no references to any objects and do not * need SRCU to keep them "alive". */ if (!to_tell->i_fsnotify_marks && (!mnt || !mnt->mnt_fsnotify_marks)) return 0; /* * if this is a modify event we may need to clear the ignored masks * otherwise return if neither the inode nor the vfsmount care about * this type of event. */ if (!(mask & FS_MODIFY) && !(test_mask & to_tell->i_fsnotify_mask) && !(mnt && test_mask & mnt->mnt_fsnotify_mask)) return 0; iter_info.srcu_idx = srcu_read_lock(&fsnotify_mark_srcu); trace_fsnotify_event(to_tell, mask, file_name); iter_info.marks[FSNOTIFY_OBJ_TYPE_INODE] = fsnotify_first_mark(&to_tell->i_fsnotify_marks); if (mnt) { iter_info.marks[FSNOTIFY_OBJ_TYPE_VFSMOUNT] = fsnotify_first_mark(&mnt->mnt_fsnotify_marks); } /* * We need to merge inode & vfsmount mark lists so that inode mark * ignore masks are properly reflected for mount mark notifications. * That's why this traversal is so complicated... */ while (fsnotify_iter_select_report_types(&iter_info)) { ret = send_to_group(to_tell, mask, data, data_is, cookie, file_name, &iter_info); if (ret && (mask & ALL_FSNOTIFY_PERM_EVENTS)) goto out; fsnotify_iter_next(&iter_info); } ret = 0; out: srcu_read_unlock(&fsnotify_mark_srcu, iter_info.srcu_idx); return ret; } EXPORT_SYMBOL_GPL(fsnotify); extern struct kmem_cache *fsnotify_mark_connector_cachep; static __init int fsnotify_init(void) { int ret; BUG_ON(hweight32(ALL_FSNOTIFY_BITS) != 23); ret = init_srcu_struct(&fsnotify_mark_srcu); if (ret) panic("initializing fsnotify_mark_srcu"); fsnotify_mark_connector_cachep = KMEM_CACHE(fsnotify_mark_connector, SLAB_PANIC); return 0; } core_initcall(fsnotify_init);