From a32e697cda27679a0327ae2cafdad8c7170f548f Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:12 +0300 Subject: [PATCH 01/19] inotify: show inotify mask flags in proc fdinfo The inotify mask flags IN_ONESHOT and IN_EXCL_UNLINK are not "internal to kernel" and should be exposed in procfs fdinfo so CRIU can restore them. Fixes: 6933599697c9 ("inotify: hide internal kernel bits from fdinfo") Link: https://lore.kernel.org/r/20220422120327.3459282-2-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fdinfo.c | 11 ++--------- fs/notify/inotify/inotify.h | 12 ++++++++++++ fs/notify/inotify/inotify_user.c | 2 +- 3 files changed, 15 insertions(+), 10 deletions(-) diff --git a/fs/notify/fdinfo.c b/fs/notify/fdinfo.c index 57f0d5d9f934..3451708fd035 100644 --- a/fs/notify/fdinfo.c +++ b/fs/notify/fdinfo.c @@ -83,16 +83,9 @@ static void inotify_fdinfo(struct seq_file *m, struct fsnotify_mark *mark) inode_mark = container_of(mark, struct inotify_inode_mark, fsn_mark); inode = igrab(fsnotify_conn_inode(mark->connector)); if (inode) { - /* - * IN_ALL_EVENTS represents all of the mask bits - * that we expose to userspace. There is at - * least one bit (FS_EVENT_ON_CHILD) which is - * used only internally to the kernel. - */ - u32 mask = mark->mask & IN_ALL_EVENTS; - seq_printf(m, "inotify wd:%x ino:%lx sdev:%x mask:%x ignored_mask:%x ", + seq_printf(m, "inotify wd:%x ino:%lx sdev:%x mask:%x ignored_mask:0 ", inode_mark->wd, inode->i_ino, inode->i_sb->s_dev, - mask, mark->ignored_mask); + inotify_mark_user_mask(mark)); show_mark_fhandle(m, inode); seq_putc(m, '\n'); iput(inode); diff --git a/fs/notify/inotify/inotify.h b/fs/notify/inotify/inotify.h index 2007e3711916..8f00151eb731 100644 --- a/fs/notify/inotify/inotify.h +++ b/fs/notify/inotify/inotify.h @@ -22,6 +22,18 @@ static inline struct inotify_event_info *INOTIFY_E(struct fsnotify_event *fse) return container_of(fse, struct inotify_event_info, fse); } +/* + * INOTIFY_USER_FLAGS represents all of the mask bits that we expose to + * userspace. There is at least one bit (FS_EVENT_ON_CHILD) which is + * used only internally to the kernel. + */ +#define INOTIFY_USER_MASK (IN_ALL_EVENTS | IN_ONESHOT | IN_EXCL_UNLINK) + +static inline __u32 inotify_mark_user_mask(struct fsnotify_mark *fsn_mark) +{ + return fsn_mark->mask & INOTIFY_USER_MASK; +} + extern void inotify_ignored_and_remove_idr(struct fsnotify_mark *fsn_mark, struct fsnotify_group *group); extern int inotify_handle_inode_event(struct fsnotify_mark *inode_mark, diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c index 54583f62dc44..3ef57db0ec9d 100644 --- a/fs/notify/inotify/inotify_user.c +++ b/fs/notify/inotify/inotify_user.c @@ -110,7 +110,7 @@ static inline __u32 inotify_arg_to_mask(struct inode *inode, u32 arg) mask |= FS_EVENT_ON_CHILD; /* mask off the flags used to open the fd */ - mask |= (arg & (IN_ALL_EVENTS | IN_ONESHOT | IN_EXCL_UNLINK)); + mask |= (arg & INOTIFY_USER_MASK); return mask; } From 38035c04f5865c4ef9597d6beed6a7178f90f64a Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:13 +0300 Subject: [PATCH 02/19] inotify: move control flags from mask to mark flags The inotify control flags in the mark mask (e.g. FS_IN_ONE_SHOT) are not relevant to object interest mask, so move them to the mark flags. This frees up some bits in the object interest mask. Link: https://lore.kernel.org/r/20220422120327.3459282-3-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fsnotify.c | 4 +-- fs/notify/inotify/inotify.h | 11 ++++++-- fs/notify/inotify/inotify_fsnotify.c | 2 +- fs/notify/inotify/inotify_user.c | 38 ++++++++++++++++++---------- include/linux/fsnotify_backend.h | 16 +++++++----- 5 files changed, 45 insertions(+), 26 deletions(-) diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c index 70a8516b78bc..6eee19d15e8c 100644 --- a/fs/notify/fsnotify.c +++ b/fs/notify/fsnotify.c @@ -253,7 +253,7 @@ static int fsnotify_handle_inode_event(struct fsnotify_group *group, if (WARN_ON_ONCE(!inode && !dir)) return 0; - if ((inode_mark->mask & FS_EXCL_UNLINK) && + if ((inode_mark->flags & FSNOTIFY_MARK_FLAG_EXCL_UNLINK) && path && d_unlinked(path->dentry)) return 0; @@ -581,7 +581,7 @@ static __init int fsnotify_init(void) { int ret; - BUILD_BUG_ON(HWEIGHT32(ALL_FSNOTIFY_BITS) != 25); + BUILD_BUG_ON(HWEIGHT32(ALL_FSNOTIFY_BITS) != 23); ret = init_srcu_struct(&fsnotify_mark_srcu); if (ret) diff --git a/fs/notify/inotify/inotify.h b/fs/notify/inotify/inotify.h index 8f00151eb731..7d5df7a21539 100644 --- a/fs/notify/inotify/inotify.h +++ b/fs/notify/inotify/inotify.h @@ -27,11 +27,18 @@ static inline struct inotify_event_info *INOTIFY_E(struct fsnotify_event *fse) * userspace. There is at least one bit (FS_EVENT_ON_CHILD) which is * used only internally to the kernel. */ -#define INOTIFY_USER_MASK (IN_ALL_EVENTS | IN_ONESHOT | IN_EXCL_UNLINK) +#define INOTIFY_USER_MASK (IN_ALL_EVENTS) static inline __u32 inotify_mark_user_mask(struct fsnotify_mark *fsn_mark) { - return fsn_mark->mask & INOTIFY_USER_MASK; + __u32 mask = fsn_mark->mask & INOTIFY_USER_MASK; + + if (fsn_mark->flags & FSNOTIFY_MARK_FLAG_EXCL_UNLINK) + mask |= IN_EXCL_UNLINK; + if (fsn_mark->flags & FSNOTIFY_MARK_FLAG_IN_ONESHOT) + mask |= IN_ONESHOT; + + return mask; } extern void inotify_ignored_and_remove_idr(struct fsnotify_mark *fsn_mark, diff --git a/fs/notify/inotify/inotify_fsnotify.c b/fs/notify/inotify/inotify_fsnotify.c index d92d7b0adc9a..49cfe2ae6d23 100644 --- a/fs/notify/inotify/inotify_fsnotify.c +++ b/fs/notify/inotify/inotify_fsnotify.c @@ -122,7 +122,7 @@ int inotify_handle_inode_event(struct fsnotify_mark *inode_mark, u32 mask, fsnotify_destroy_event(group, fsn_event); } - if (inode_mark->mask & IN_ONESHOT) + if (inode_mark->flags & FSNOTIFY_MARK_FLAG_IN_ONESHOT) fsnotify_destroy_mark(inode_mark, group); return 0; diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c index 3ef57db0ec9d..d8907d32a05b 100644 --- a/fs/notify/inotify/inotify_user.c +++ b/fs/notify/inotify/inotify_user.c @@ -115,6 +115,21 @@ static inline __u32 inotify_arg_to_mask(struct inode *inode, u32 arg) return mask; } +#define INOTIFY_MARK_FLAGS \ + (FSNOTIFY_MARK_FLAG_EXCL_UNLINK | FSNOTIFY_MARK_FLAG_IN_ONESHOT) + +static inline unsigned int inotify_arg_to_flags(u32 arg) +{ + unsigned int flags = 0; + + if (arg & IN_EXCL_UNLINK) + flags |= FSNOTIFY_MARK_FLAG_EXCL_UNLINK; + if (arg & IN_ONESHOT) + flags |= FSNOTIFY_MARK_FLAG_IN_ONESHOT; + + return flags; +} + static inline u32 inotify_mask_to_arg(__u32 mask) { return mask & (IN_ALL_EVENTS | IN_ISDIR | IN_UNMOUNT | IN_IGNORED | @@ -526,13 +541,10 @@ static int inotify_update_existing_watch(struct fsnotify_group *group, struct fsnotify_mark *fsn_mark; struct inotify_inode_mark *i_mark; __u32 old_mask, new_mask; - __u32 mask; - int add = (arg & IN_MASK_ADD); + int replace = !(arg & IN_MASK_ADD); int create = (arg & IN_MASK_CREATE); int ret; - mask = inotify_arg_to_mask(inode, arg); - fsn_mark = fsnotify_find_mark(&inode->i_fsnotify_marks, group); if (!fsn_mark) return -ENOENT; @@ -545,10 +557,12 @@ static int inotify_update_existing_watch(struct fsnotify_group *group, spin_lock(&fsn_mark->lock); old_mask = fsn_mark->mask; - if (add) - fsn_mark->mask |= mask; - else - fsn_mark->mask = mask; + if (replace) { + fsn_mark->mask = 0; + fsn_mark->flags &= ~INOTIFY_MARK_FLAGS; + } + fsn_mark->mask |= inotify_arg_to_mask(inode, arg); + fsn_mark->flags |= inotify_arg_to_flags(arg); new_mask = fsn_mark->mask; spin_unlock(&fsn_mark->lock); @@ -579,19 +593,17 @@ static int inotify_new_watch(struct fsnotify_group *group, u32 arg) { struct inotify_inode_mark *tmp_i_mark; - __u32 mask; int ret; struct idr *idr = &group->inotify_data.idr; spinlock_t *idr_lock = &group->inotify_data.idr_lock; - mask = inotify_arg_to_mask(inode, arg); - tmp_i_mark = kmem_cache_alloc(inotify_inode_mark_cachep, GFP_KERNEL); if (unlikely(!tmp_i_mark)) return -ENOMEM; fsnotify_init_mark(&tmp_i_mark->fsn_mark, group); - tmp_i_mark->fsn_mark.mask = mask; + tmp_i_mark->fsn_mark.mask = inotify_arg_to_mask(inode, arg); + tmp_i_mark->fsn_mark.flags = inotify_arg_to_flags(arg); tmp_i_mark->wd = -1; ret = inotify_add_to_idr(idr, idr_lock, tmp_i_mark); @@ -845,9 +857,7 @@ static int __init inotify_user_setup(void) BUILD_BUG_ON(IN_UNMOUNT != FS_UNMOUNT); BUILD_BUG_ON(IN_Q_OVERFLOW != FS_Q_OVERFLOW); BUILD_BUG_ON(IN_IGNORED != FS_IN_IGNORED); - BUILD_BUG_ON(IN_EXCL_UNLINK != FS_EXCL_UNLINK); BUILD_BUG_ON(IN_ISDIR != FS_ISDIR); - BUILD_BUG_ON(IN_ONESHOT != FS_IN_ONESHOT); BUILD_BUG_ON(HWEIGHT32(ALL_INOTIFY_BITS) != 22); diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h index 0805b74cae44..b1c72edd9784 100644 --- a/include/linux/fsnotify_backend.h +++ b/include/linux/fsnotify_backend.h @@ -55,7 +55,6 @@ #define FS_ACCESS_PERM 0x00020000 /* access event in a permissions hook */ #define FS_OPEN_EXEC_PERM 0x00040000 /* open/exec event in a permission hook */ -#define FS_EXCL_UNLINK 0x04000000 /* do not send events if object is unlinked */ /* * Set on inode mark that cares about things that happen to its children. * Always set for dnotify and inotify. @@ -66,7 +65,6 @@ #define FS_RENAME 0x10000000 /* File was renamed */ #define FS_DN_MULTISHOT 0x20000000 /* dnotify multishot */ #define FS_ISDIR 0x40000000 /* event occurred against dir */ -#define FS_IN_ONESHOT 0x80000000 /* only send event once */ #define FS_MOVE (FS_MOVED_FROM | FS_MOVED_TO) @@ -106,8 +104,7 @@ FS_ERROR) /* Extra flags that may be reported with event or control handling of events */ -#define ALL_FSNOTIFY_FLAGS (FS_EXCL_UNLINK | FS_ISDIR | FS_IN_ONESHOT | \ - FS_DN_MULTISHOT | FS_EVENT_ON_CHILD) +#define ALL_FSNOTIFY_FLAGS (FS_ISDIR | FS_EVENT_ON_CHILD | FS_DN_MULTISHOT) #define ALL_FSNOTIFY_BITS (ALL_FSNOTIFY_EVENTS | ALL_FSNOTIFY_FLAGS) @@ -473,9 +470,14 @@ struct fsnotify_mark { struct fsnotify_mark_connector *connector; /* Events types to ignore [mark->lock, group->mark_mutex] */ __u32 ignored_mask; -#define FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY 0x01 -#define FSNOTIFY_MARK_FLAG_ALIVE 0x02 -#define FSNOTIFY_MARK_FLAG_ATTACHED 0x04 + /* General fsnotify mark flags */ +#define FSNOTIFY_MARK_FLAG_ALIVE 0x0001 +#define FSNOTIFY_MARK_FLAG_ATTACHED 0x0002 + /* inotify mark flags */ +#define FSNOTIFY_MARK_FLAG_EXCL_UNLINK 0x0010 +#define FSNOTIFY_MARK_FLAG_IN_ONESHOT 0x0020 + /* fanotify mark flags */ +#define FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY 0x0100 unsigned int flags; /* flags [mark->lock] */ }; From 623af4f538b5df9b416e1b82f720af7371b4c771 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:14 +0300 Subject: [PATCH 03/19] fsnotify: fix wrong lockdep annotations Commit 6960b0d909cd ("fsnotify: change locking order") changed some of the mark_mutex locks in direct reclaim path to use: mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); This change is explained: "...It uses nested locking to avoid deadlock in case we do the final iput() on an inode which still holds marks and thus would take the mutex again when calling fsnotify_inode_delete() in destroy_inode()." The problem is that the mutex_lock_nested() is not a nested lock at all. In fact, it has the opposite effect of preventing lockdep from warning about a very possible deadlock. Due to these wrong annotations, a deadlock that was introduced with nfsd filecache in kernel v5.4 went unnoticed in v5.4.y for over two years until it was reported recently by Khazhismel Kumykov, only to find out that the deadlock was already fixed in kernel v5.5. Fix the wrong lockdep annotations. Cc: Khazhismel Kumykov Fixes: 6960b0d909cd ("fsnotify: change locking order") Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Link: https://lore.kernel.org/r/20220422120327.3459282-4-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/mark.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/notify/mark.c b/fs/notify/mark.c index 4853184f7dde..c86982be2d50 100644 --- a/fs/notify/mark.c +++ b/fs/notify/mark.c @@ -452,7 +452,7 @@ void fsnotify_free_mark(struct fsnotify_mark *mark) void fsnotify_destroy_mark(struct fsnotify_mark *mark, struct fsnotify_group *group) { - mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); + mutex_lock(&group->mark_mutex); fsnotify_detach_mark(mark); mutex_unlock(&group->mark_mutex); fsnotify_free_mark(mark); @@ -770,7 +770,7 @@ void fsnotify_clear_marks_by_group(struct fsnotify_group *group, * move marks to free to to_free list in one go and then free marks in * to_free list one by one. */ - mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); + mutex_lock(&group->mark_mutex); list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) { if (mark->connector->type == obj_type) list_move(&mark->g_list, &to_free); @@ -779,7 +779,7 @@ void fsnotify_clear_marks_by_group(struct fsnotify_group *group, clear: while (1) { - mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING); + mutex_lock(&group->mark_mutex); if (list_empty(head)) { mutex_unlock(&group->mark_mutex); break; From 867a448d587e7fa845bceaf4ee1c632448f2a9fa Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:15 +0300 Subject: [PATCH 04/19] fsnotify: pass flags argument to fsnotify_alloc_group() Add flags argument to fsnotify_alloc_group(), define and use the flag FSNOTIFY_GROUP_USER in inotify and fanotify instead of the helper fsnotify_alloc_user_group() to indicate user allocation. Although the flag FSNOTIFY_GROUP_USER is currently not used after group allocation, we store the flags argument in the group struct for future use of other group flags. Link: https://lore.kernel.org/r/20220422120327.3459282-5-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/nfsd/filecache.c | 3 ++- fs/notify/dnotify/dnotify.c | 2 +- fs/notify/fanotify/fanotify_user.c | 3 ++- fs/notify/group.c | 21 +++++++++------------ fs/notify/inotify/inotify_user.c | 3 ++- include/linux/fsnotify_backend.h | 8 ++++++-- kernel/audit_fsnotify.c | 3 ++- kernel/audit_tree.c | 2 +- kernel/audit_watch.c | 2 +- 9 files changed, 26 insertions(+), 21 deletions(-) diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c index 2c1b027774d4..74ddb828fd75 100644 --- a/fs/nfsd/filecache.c +++ b/fs/nfsd/filecache.c @@ -678,7 +678,8 @@ nfsd_file_cache_init(void) goto out_shrinker; } - nfsd_file_fsnotify_group = fsnotify_alloc_group(&nfsd_file_fsnotify_ops); + nfsd_file_fsnotify_group = fsnotify_alloc_group(&nfsd_file_fsnotify_ops, + 0); if (IS_ERR(nfsd_file_fsnotify_group)) { pr_err("nfsd: unable to create fsnotify group: %ld\n", PTR_ERR(nfsd_file_fsnotify_group)); diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c index 829dd4a61b66..e4779926edf4 100644 --- a/fs/notify/dnotify/dnotify.c +++ b/fs/notify/dnotify/dnotify.c @@ -401,7 +401,7 @@ static int __init dnotify_init(void) SLAB_PANIC|SLAB_ACCOUNT); dnotify_mark_cache = KMEM_CACHE(dnotify_mark, SLAB_PANIC|SLAB_ACCOUNT); - dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops); + dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops, 0); if (IS_ERR(dnotify_group)) panic("unable to allocate fsnotify group for dnotify\n"); dnotify_sysctl_init(); diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c index 9b32b76a9c30..3649c99b3e45 100644 --- a/fs/notify/fanotify/fanotify_user.c +++ b/fs/notify/fanotify/fanotify_user.c @@ -1355,7 +1355,8 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags) f_flags |= O_NONBLOCK; /* fsnotify_alloc_group takes a ref. Dropped in fanotify_release */ - group = fsnotify_alloc_user_group(&fanotify_fsnotify_ops); + group = fsnotify_alloc_group(&fanotify_fsnotify_ops, + FSNOTIFY_GROUP_USER); if (IS_ERR(group)) { return PTR_ERR(group); } diff --git a/fs/notify/group.c b/fs/notify/group.c index b7d4d64f87c2..18446b7b0d49 100644 --- a/fs/notify/group.c +++ b/fs/notify/group.c @@ -112,7 +112,8 @@ void fsnotify_put_group(struct fsnotify_group *group) EXPORT_SYMBOL_GPL(fsnotify_put_group); static struct fsnotify_group *__fsnotify_alloc_group( - const struct fsnotify_ops *ops, gfp_t gfp) + const struct fsnotify_ops *ops, + int flags, gfp_t gfp) { struct fsnotify_group *group; @@ -133,6 +134,7 @@ static struct fsnotify_group *__fsnotify_alloc_group( INIT_LIST_HEAD(&group->marks_list); group->ops = ops; + group->flags = flags; return group; } @@ -140,21 +142,16 @@ static struct fsnotify_group *__fsnotify_alloc_group( /* * Create a new fsnotify_group and hold a reference for the group returned. */ -struct fsnotify_group *fsnotify_alloc_group(const struct fsnotify_ops *ops) +struct fsnotify_group *fsnotify_alloc_group(const struct fsnotify_ops *ops, + int flags) { - return __fsnotify_alloc_group(ops, GFP_KERNEL); + gfp_t gfp = (flags & FSNOTIFY_GROUP_USER) ? GFP_KERNEL_ACCOUNT : + GFP_KERNEL; + + return __fsnotify_alloc_group(ops, flags, gfp); } EXPORT_SYMBOL_GPL(fsnotify_alloc_group); -/* - * Create a new fsnotify_group and hold a reference for the group returned. - */ -struct fsnotify_group *fsnotify_alloc_user_group(const struct fsnotify_ops *ops) -{ - return __fsnotify_alloc_group(ops, GFP_KERNEL_ACCOUNT); -} -EXPORT_SYMBOL_GPL(fsnotify_alloc_user_group); - int fsnotify_fasync(int fd, struct file *file, int on) { struct fsnotify_group *group = file->private_data; diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c index d8907d32a05b..146890ecd93a 100644 --- a/fs/notify/inotify/inotify_user.c +++ b/fs/notify/inotify/inotify_user.c @@ -656,7 +656,8 @@ static struct fsnotify_group *inotify_new_group(unsigned int max_events) struct fsnotify_group *group; struct inotify_event_info *oevent; - group = fsnotify_alloc_user_group(&inotify_fsnotify_ops); + group = fsnotify_alloc_group(&inotify_fsnotify_ops, + FSNOTIFY_GROUP_USER); if (IS_ERR(group)) return group; diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h index b1c72edd9784..f0bf557af009 100644 --- a/include/linux/fsnotify_backend.h +++ b/include/linux/fsnotify_backend.h @@ -210,6 +210,9 @@ struct fsnotify_group { unsigned int priority; bool shutdown; /* group is being shut down, don't queue more events */ +#define FSNOTIFY_GROUP_USER 0x01 /* user allocated group */ + int flags; + /* stores all fastpath marks assoc with this group so they can be cleaned on unregister */ struct mutex mark_mutex; /* protect marks_list */ atomic_t user_waits; /* Number of tasks waiting for user @@ -543,8 +546,9 @@ static inline void fsnotify_update_flags(struct dentry *dentry) /* called from fsnotify listeners, such as fanotify or dnotify */ /* create a new group */ -extern struct fsnotify_group *fsnotify_alloc_group(const struct fsnotify_ops *ops); -extern struct fsnotify_group *fsnotify_alloc_user_group(const struct fsnotify_ops *ops); +extern struct fsnotify_group *fsnotify_alloc_group( + const struct fsnotify_ops *ops, + int flags); /* get reference to a group */ extern void fsnotify_get_group(struct fsnotify_group *group); /* drop reference on a group from fsnotify_alloc_group */ diff --git a/kernel/audit_fsnotify.c b/kernel/audit_fsnotify.c index 02348b48447c..35fe149586c8 100644 --- a/kernel/audit_fsnotify.c +++ b/kernel/audit_fsnotify.c @@ -181,7 +181,8 @@ static const struct fsnotify_ops audit_mark_fsnotify_ops = { static int __init audit_fsnotify_init(void) { - audit_fsnotify_group = fsnotify_alloc_group(&audit_mark_fsnotify_ops); + audit_fsnotify_group = fsnotify_alloc_group(&audit_mark_fsnotify_ops, + 0); if (IS_ERR(audit_fsnotify_group)) { audit_fsnotify_group = NULL; audit_panic("cannot create audit fsnotify group"); diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index e7315d487163..b5c02f8573fe 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -1074,7 +1074,7 @@ static int __init audit_tree_init(void) audit_tree_mark_cachep = KMEM_CACHE(audit_tree_mark, SLAB_PANIC); - audit_tree_group = fsnotify_alloc_group(&audit_tree_ops); + audit_tree_group = fsnotify_alloc_group(&audit_tree_ops, 0); if (IS_ERR(audit_tree_group)) audit_panic("cannot initialize fsnotify group for rectree watches"); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 713b256be944..4b0957aa2cd4 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -493,7 +493,7 @@ static const struct fsnotify_ops audit_watch_fsnotify_ops = { static int __init audit_watch_init(void) { - audit_watch_group = fsnotify_alloc_group(&audit_watch_fsnotify_ops); + audit_watch_group = fsnotify_alloc_group(&audit_watch_fsnotify_ops, 0); if (IS_ERR(audit_watch_group)) { audit_watch_group = NULL; audit_panic("cannot create audit fsnotify group"); From f3010343d9e119da35ee864b3a28993bb5c78ed7 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:16 +0300 Subject: [PATCH 05/19] fsnotify: make allow_dups a property of the group Instead of passing the allow_dups argument to fsnotify_add_mark() as an argument, define the group flag FSNOTIFY_GROUP_DUPS to express the allow_dups behavior and set this behavior at group creation time for all calls of fsnotify_add_mark(). Rename the allow_dups argument to generic add_flags argument for future use. Link: https://lore.kernel.org/r/20220422120327.3459282-6-amir73il@gmail.com Suggested-by: Jan Kara Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/mark.c | 12 ++++++------ include/linux/fsnotify_backend.h | 13 +++++++------ kernel/audit_fsnotify.c | 4 ++-- 3 files changed, 15 insertions(+), 14 deletions(-) diff --git a/fs/notify/mark.c b/fs/notify/mark.c index c86982be2d50..1fb246ea6175 100644 --- a/fs/notify/mark.c +++ b/fs/notify/mark.c @@ -574,7 +574,7 @@ out: static int fsnotify_add_mark_list(struct fsnotify_mark *mark, fsnotify_connp_t *connp, unsigned int obj_type, - int allow_dups, __kernel_fsid_t *fsid) + int add_flags, __kernel_fsid_t *fsid) { struct fsnotify_mark *lmark, *last = NULL; struct fsnotify_mark_connector *conn; @@ -633,7 +633,7 @@ restart: if ((lmark->group == mark->group) && (lmark->flags & FSNOTIFY_MARK_FLAG_ATTACHED) && - !allow_dups) { + !(mark->group->flags & FSNOTIFY_GROUP_DUPS)) { err = -EEXIST; goto out_err; } @@ -668,7 +668,7 @@ out_err: */ int fsnotify_add_mark_locked(struct fsnotify_mark *mark, fsnotify_connp_t *connp, unsigned int obj_type, - int allow_dups, __kernel_fsid_t *fsid) + int add_flags, __kernel_fsid_t *fsid) { struct fsnotify_group *group = mark->group; int ret = 0; @@ -688,7 +688,7 @@ int fsnotify_add_mark_locked(struct fsnotify_mark *mark, fsnotify_get_mark(mark); /* for g_list */ spin_unlock(&mark->lock); - ret = fsnotify_add_mark_list(mark, connp, obj_type, allow_dups, fsid); + ret = fsnotify_add_mark_list(mark, connp, obj_type, add_flags, fsid); if (ret) goto err; @@ -708,14 +708,14 @@ err: } int fsnotify_add_mark(struct fsnotify_mark *mark, fsnotify_connp_t *connp, - unsigned int obj_type, int allow_dups, + unsigned int obj_type, int add_flags, __kernel_fsid_t *fsid) { int ret; struct fsnotify_group *group = mark->group; mutex_lock(&group->mark_mutex); - ret = fsnotify_add_mark_locked(mark, connp, obj_type, allow_dups, fsid); + ret = fsnotify_add_mark_locked(mark, connp, obj_type, add_flags, fsid); mutex_unlock(&group->mark_mutex); return ret; } diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h index f0bf557af009..dd440e6ff528 100644 --- a/include/linux/fsnotify_backend.h +++ b/include/linux/fsnotify_backend.h @@ -211,6 +211,7 @@ struct fsnotify_group { bool shutdown; /* group is being shut down, don't queue more events */ #define FSNOTIFY_GROUP_USER 0x01 /* user allocated group */ +#define FSNOTIFY_GROUP_DUPS 0x02 /* allow multiple marks per object */ int flags; /* stores all fastpath marks assoc with this group so they can be cleaned on unregister */ @@ -641,26 +642,26 @@ extern int fsnotify_get_conn_fsid(const struct fsnotify_mark_connector *conn, /* attach the mark to the object */ extern int fsnotify_add_mark(struct fsnotify_mark *mark, fsnotify_connp_t *connp, unsigned int obj_type, - int allow_dups, __kernel_fsid_t *fsid); + int add_flags, __kernel_fsid_t *fsid); extern int fsnotify_add_mark_locked(struct fsnotify_mark *mark, fsnotify_connp_t *connp, - unsigned int obj_type, int allow_dups, + unsigned int obj_type, int add_flags, __kernel_fsid_t *fsid); /* attach the mark to the inode */ static inline int fsnotify_add_inode_mark(struct fsnotify_mark *mark, struct inode *inode, - int allow_dups) + int add_flags) { return fsnotify_add_mark(mark, &inode->i_fsnotify_marks, - FSNOTIFY_OBJ_TYPE_INODE, allow_dups, NULL); + FSNOTIFY_OBJ_TYPE_INODE, add_flags, NULL); } static inline int fsnotify_add_inode_mark_locked(struct fsnotify_mark *mark, struct inode *inode, - int allow_dups) + int add_flags) { return fsnotify_add_mark_locked(mark, &inode->i_fsnotify_marks, - FSNOTIFY_OBJ_TYPE_INODE, allow_dups, + FSNOTIFY_OBJ_TYPE_INODE, add_flags, NULL); } diff --git a/kernel/audit_fsnotify.c b/kernel/audit_fsnotify.c index 35fe149586c8..6432a37ac1c9 100644 --- a/kernel/audit_fsnotify.c +++ b/kernel/audit_fsnotify.c @@ -100,7 +100,7 @@ struct audit_fsnotify_mark *audit_alloc_mark(struct audit_krule *krule, char *pa audit_update_mark(audit_mark, dentry->d_inode); audit_mark->rule = krule; - ret = fsnotify_add_inode_mark(&audit_mark->mark, inode, true); + ret = fsnotify_add_inode_mark(&audit_mark->mark, inode, 0); if (ret < 0) { fsnotify_put_mark(&audit_mark->mark); audit_mark = ERR_PTR(ret); @@ -182,7 +182,7 @@ static const struct fsnotify_ops audit_mark_fsnotify_ops = { static int __init audit_fsnotify_init(void) { audit_fsnotify_group = fsnotify_alloc_group(&audit_mark_fsnotify_ops, - 0); + FSNOTIFY_GROUP_DUPS); if (IS_ERR(audit_fsnotify_group)) { audit_fsnotify_group = NULL; audit_panic("cannot create audit fsnotify group"); From 43b245a788e2d8f1bb742668a9bdace02fcb3e96 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:17 +0300 Subject: [PATCH 06/19] fsnotify: create helpers for group mark_mutex lock Create helpers to take and release the group mark_mutex lock. Define a flag FSNOTIFY_GROUP_NOFS in fsnotify_group that determines if the mark_mutex lock is fs reclaim safe or not. If not safe, the lock helpers take the lock and disable direct fs reclaim. In that case we annotate the mutex with a different lockdep class to express to lockdep that an allocation of mark of an fs reclaim safe group may take the group lock of another "NOFS" group to evict inodes. For now, converted only the callers in common code and no backend defines the NOFS flag. It is intended to be set by fanotify for evictable marks support. Link: https://lore.kernel.org/r/20220422120327.3459282-7-amir73il@gmail.com Suggested-by: Jan Kara Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fdinfo.c | 4 ++-- fs/notify/group.c | 11 +++++++++++ fs/notify/mark.c | 24 +++++++++++------------- include/linux/fsnotify_backend.h | 28 ++++++++++++++++++++++++++++ 4 files changed, 52 insertions(+), 15 deletions(-) diff --git a/fs/notify/fdinfo.c b/fs/notify/fdinfo.c index 3451708fd035..1f34c5c29fdb 100644 --- a/fs/notify/fdinfo.c +++ b/fs/notify/fdinfo.c @@ -28,13 +28,13 @@ static void show_fdinfo(struct seq_file *m, struct file *f, struct fsnotify_group *group = f->private_data; struct fsnotify_mark *mark; - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); list_for_each_entry(mark, &group->marks_list, g_list) { show(m, mark); if (seq_has_overflowed(m)) break; } - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); } #if defined(CONFIG_EXPORTFS) diff --git a/fs/notify/group.c b/fs/notify/group.c index 18446b7b0d49..1de6631a3925 100644 --- a/fs/notify/group.c +++ b/fs/notify/group.c @@ -115,6 +115,7 @@ static struct fsnotify_group *__fsnotify_alloc_group( const struct fsnotify_ops *ops, int flags, gfp_t gfp) { + static struct lock_class_key nofs_marks_lock; struct fsnotify_group *group; group = kzalloc(sizeof(struct fsnotify_group), gfp); @@ -135,6 +136,16 @@ static struct fsnotify_group *__fsnotify_alloc_group( group->ops = ops; group->flags = flags; + /* + * For most backends, eviction of inode with a mark is not expected, + * because marks hold a refcount on the inode against eviction. + * + * Use a different lockdep class for groups that support evictable + * inode marks, because with evictable marks, mark_mutex is NOT + * fs-reclaim safe - the mutex is taken when evicting inodes. + */ + if (flags & FSNOTIFY_GROUP_NOFS) + lockdep_set_class(&group->mark_mutex, &nofs_marks_lock); return group; } diff --git a/fs/notify/mark.c b/fs/notify/mark.c index 1fb246ea6175..982ca2f20ff5 100644 --- a/fs/notify/mark.c +++ b/fs/notify/mark.c @@ -398,9 +398,7 @@ void fsnotify_finish_user_wait(struct fsnotify_iter_info *iter_info) */ void fsnotify_detach_mark(struct fsnotify_mark *mark) { - struct fsnotify_group *group = mark->group; - - WARN_ON_ONCE(!mutex_is_locked(&group->mark_mutex)); + fsnotify_group_assert_locked(mark->group); WARN_ON_ONCE(!srcu_read_lock_held(&fsnotify_mark_srcu) && refcount_read(&mark->refcnt) < 1 + !!(mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED)); @@ -452,9 +450,9 @@ void fsnotify_free_mark(struct fsnotify_mark *mark) void fsnotify_destroy_mark(struct fsnotify_mark *mark, struct fsnotify_group *group) { - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); fsnotify_detach_mark(mark); - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); fsnotify_free_mark(mark); } EXPORT_SYMBOL_GPL(fsnotify_destroy_mark); @@ -673,7 +671,7 @@ int fsnotify_add_mark_locked(struct fsnotify_mark *mark, struct fsnotify_group *group = mark->group; int ret = 0; - BUG_ON(!mutex_is_locked(&group->mark_mutex)); + fsnotify_group_assert_locked(group); /* * LOCKING ORDER!!!! @@ -714,9 +712,9 @@ int fsnotify_add_mark(struct fsnotify_mark *mark, fsnotify_connp_t *connp, int ret; struct fsnotify_group *group = mark->group; - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); ret = fsnotify_add_mark_locked(mark, connp, obj_type, add_flags, fsid); - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); return ret; } EXPORT_SYMBOL_GPL(fsnotify_add_mark); @@ -770,24 +768,24 @@ void fsnotify_clear_marks_by_group(struct fsnotify_group *group, * move marks to free to to_free list in one go and then free marks in * to_free list one by one. */ - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) { if (mark->connector->type == obj_type) list_move(&mark->g_list, &to_free); } - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); clear: while (1) { - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); if (list_empty(head)) { - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); break; } mark = list_first_entry(head, struct fsnotify_mark, g_list); fsnotify_get_mark(mark); fsnotify_detach_mark(mark); - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); fsnotify_free_mark(mark); fsnotify_put_mark(mark); } diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h index dd440e6ff528..d62111e83244 100644 --- a/include/linux/fsnotify_backend.h +++ b/include/linux/fsnotify_backend.h @@ -20,6 +20,7 @@ #include #include #include +#include /* * IN_* from inotfy.h lines up EXACTLY with FS_*, this is so we can easily @@ -212,7 +213,9 @@ struct fsnotify_group { #define FSNOTIFY_GROUP_USER 0x01 /* user allocated group */ #define FSNOTIFY_GROUP_DUPS 0x02 /* allow multiple marks per object */ +#define FSNOTIFY_GROUP_NOFS 0x04 /* group lock is not direct reclaim safe */ int flags; + unsigned int owner_flags; /* stored flags of mark_mutex owner */ /* stores all fastpath marks assoc with this group so they can be cleaned on unregister */ struct mutex mark_mutex; /* protect marks_list */ @@ -254,6 +257,31 @@ struct fsnotify_group { }; }; +/* + * These helpers are used to prevent deadlock when reclaiming inodes with + * evictable marks of the same group that is allocating a new mark. + */ +static inline void fsnotify_group_lock(struct fsnotify_group *group) +{ + mutex_lock(&group->mark_mutex); + if (group->flags & FSNOTIFY_GROUP_NOFS) + group->owner_flags = memalloc_nofs_save(); +} + +static inline void fsnotify_group_unlock(struct fsnotify_group *group) +{ + if (group->flags & FSNOTIFY_GROUP_NOFS) + memalloc_nofs_restore(group->owner_flags); + mutex_unlock(&group->mark_mutex); +} + +static inline void fsnotify_group_assert_locked(struct fsnotify_group *group) +{ + WARN_ON_ONCE(!mutex_is_locked(&group->mark_mutex)); + if (group->flags & FSNOTIFY_GROUP_NOFS) + WARN_ON_ONCE(!(current->flags & PF_MEMALLOC_NOFS)); +} + /* When calling fsnotify tell it if the data is a path or inode */ enum fsnotify_data_type { FSNOTIFY_EVENT_NONE, From 642054b87058019be36033f73c3e48ffff1915aa Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:18 +0300 Subject: [PATCH 07/19] inotify: use fsnotify group lock helpers inotify inode marks pin the inode so there is no need to set the FSNOTIFY_GROUP_NOFS flag. Link: https://lore.kernel.org/r/20220422120327.3459282-8-amir73il@gmail.com Suggested-by: Jan Kara Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/inotify/inotify_user.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c index 146890ecd93a..ed42a189faa2 100644 --- a/fs/notify/inotify/inotify_user.c +++ b/fs/notify/inotify/inotify_user.c @@ -640,13 +640,13 @@ static int inotify_update_watch(struct fsnotify_group *group, struct inode *inod { int ret = 0; - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); /* try to update and existing watch with the new arg */ ret = inotify_update_existing_watch(group, inode, arg); /* no mark present, try to add a new one */ if (ret == -ENOENT) ret = inotify_new_watch(group, inode, arg); - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); return ret; } From 960bdff24ce802e38df918ebfcbfa62744d6ae22 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:19 +0300 Subject: [PATCH 08/19] audit: use fsnotify group lock helpers audit inode marks pin the inode so there is no need to set the FSNOTIFY_GROUP_NOFS flag. Link: https://lore.kernel.org/r/20220422120327.3459282-9-amir73il@gmail.com Suggested-by: Jan Kara Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- kernel/audit_tree.c | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-) diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index b5c02f8573fe..e867c17d3f84 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -351,7 +351,7 @@ static void untag_chunk(struct audit_chunk *chunk, struct fsnotify_mark *mark) struct audit_chunk *new; int size; - mutex_lock(&audit_tree_group->mark_mutex); + fsnotify_group_lock(audit_tree_group); /* * mark_mutex stabilizes chunk attached to the mark so we can check * whether it didn't change while we've dropped hash_lock. @@ -368,7 +368,7 @@ static void untag_chunk(struct audit_chunk *chunk, struct fsnotify_mark *mark) replace_mark_chunk(mark, NULL); spin_unlock(&hash_lock); fsnotify_detach_mark(mark); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); audit_mark_put_chunk(chunk); fsnotify_free_mark(mark); return; @@ -385,12 +385,12 @@ static void untag_chunk(struct audit_chunk *chunk, struct fsnotify_mark *mark) */ replace_chunk(new, chunk); spin_unlock(&hash_lock); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); audit_mark_put_chunk(chunk); return; out_mutex: - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); } /* Call with group->mark_mutex held, releases it */ @@ -400,19 +400,19 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) struct audit_chunk *chunk = alloc_chunk(1); if (!chunk) { - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); return -ENOMEM; } mark = alloc_mark(); if (!mark) { - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); kfree(chunk); return -ENOMEM; } if (fsnotify_add_inode_mark_locked(mark, inode, 0)) { - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); fsnotify_put_mark(mark); kfree(chunk); return -ENOSPC; @@ -422,7 +422,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) if (tree->goner) { spin_unlock(&hash_lock); fsnotify_detach_mark(mark); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); fsnotify_free_mark(mark); fsnotify_put_mark(mark); kfree(chunk); @@ -444,7 +444,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) */ insert_hash(chunk); spin_unlock(&hash_lock); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); /* * Drop our initial reference. When mark we point to is getting freed, * we get notification through ->freeing_mark callback and cleanup @@ -462,7 +462,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) struct audit_node *p; int n; - mutex_lock(&audit_tree_group->mark_mutex); + fsnotify_group_lock(audit_tree_group); mark = fsnotify_find_mark(&inode->i_fsnotify_marks, audit_tree_group); if (!mark) return create_chunk(inode, tree); @@ -478,7 +478,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) for (n = 0; n < old->count; n++) { if (old->owners[n].owner == tree) { spin_unlock(&hash_lock); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); fsnotify_put_mark(mark); return 0; } @@ -487,7 +487,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) chunk = alloc_chunk(old->count + 1); if (!chunk) { - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); fsnotify_put_mark(mark); return -ENOMEM; } @@ -495,7 +495,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) spin_lock(&hash_lock); if (tree->goner) { spin_unlock(&hash_lock); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); fsnotify_put_mark(mark); kfree(chunk); return 0; @@ -515,7 +515,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) */ replace_chunk(chunk, old); spin_unlock(&hash_lock); - mutex_unlock(&audit_tree_group->mark_mutex); + fsnotify_group_unlock(audit_tree_group); fsnotify_put_mark(mark); /* pair to fsnotify_find_mark */ audit_mark_put_chunk(old); @@ -1044,12 +1044,12 @@ static void audit_tree_freeing_mark(struct fsnotify_mark *mark, { struct audit_chunk *chunk; - mutex_lock(&mark->group->mark_mutex); + fsnotify_group_lock(mark->group); spin_lock(&hash_lock); chunk = mark_chunk(mark); replace_mark_chunk(mark, NULL); spin_unlock(&hash_lock); - mutex_unlock(&mark->group->mark_mutex); + fsnotify_group_unlock(mark->group); if (chunk) { evict_chunk(chunk); audit_mark_put_chunk(chunk); From b8962a9d8cc2d8c93362e2f684091c79f702f6f3 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:20 +0300 Subject: [PATCH 09/19] nfsd: use fsnotify group lock helpers Before commit 9542e6a643fc6 ("nfsd: Containerise filecache laundrette") nfsd would close open files in direct reclaim context and that could cause a deadlock when fsnotify mark allocation went into direct reclaim and nfsd shrinker tried to free existing fsnotify marks. To avoid issues like this in future code, set the FSNOTIFY_GROUP_NOFS flag on nfsd fsnotify group to prevent going into direct reclaim from fsnotify_add_inode_mark(). Link: https://lore.kernel.org/r/20220422120327.3459282-10-amir73il@gmail.com Suggested-by: Jan Kara Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/nfsd/filecache.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c index 74ddb828fd75..489c9c1d8f31 100644 --- a/fs/nfsd/filecache.c +++ b/fs/nfsd/filecache.c @@ -119,14 +119,14 @@ nfsd_file_mark_find_or_create(struct nfsd_file *nf) struct inode *inode = nf->nf_inode; do { - mutex_lock(&nfsd_file_fsnotify_group->mark_mutex); + fsnotify_group_lock(nfsd_file_fsnotify_group); mark = fsnotify_find_mark(&inode->i_fsnotify_marks, - nfsd_file_fsnotify_group); + nfsd_file_fsnotify_group); if (mark) { nfm = nfsd_file_mark_get(container_of(mark, struct nfsd_file_mark, nfm_mark)); - mutex_unlock(&nfsd_file_fsnotify_group->mark_mutex); + fsnotify_group_unlock(nfsd_file_fsnotify_group); if (nfm) { fsnotify_put_mark(mark); break; @@ -134,8 +134,9 @@ nfsd_file_mark_find_or_create(struct nfsd_file *nf) /* Avoid soft lockup race with nfsd_file_mark_put() */ fsnotify_destroy_mark(mark, nfsd_file_fsnotify_group); fsnotify_put_mark(mark); - } else - mutex_unlock(&nfsd_file_fsnotify_group->mark_mutex); + } else { + fsnotify_group_unlock(nfsd_file_fsnotify_group); + } /* allocate a new nfm */ new = kmem_cache_alloc(nfsd_file_mark_slab, GFP_KERNEL); @@ -679,7 +680,7 @@ nfsd_file_cache_init(void) } nfsd_file_fsnotify_group = fsnotify_alloc_group(&nfsd_file_fsnotify_ops, - 0); + FSNOTIFY_GROUP_NOFS); if (IS_ERR(nfsd_file_fsnotify_group)) { pr_err("nfsd: unable to create fsnotify group: %ld\n", PTR_ERR(nfsd_file_fsnotify_group)); From aabb45fdcb31f00f1e7cae2bce83e83474a87c03 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:21 +0300 Subject: [PATCH 10/19] dnotify: use fsnotify group lock helpers Before commit 9542e6a643fc6 ("nfsd: Containerise filecache laundrette") nfsd would close open files in direct reclaim context. There is no guarantee that others memory shrinkers don't do the same and no guarantee that future shrinkers won't do that. For example, if overlayfs implements inode cache of fscache would keep open files to cached objects, inode shrinkers could end up closing open files to underlying fs. Direct reclaim from dnotify mark allocation context may try to close open files that have dnotify marks of the same group and hit a deadlock on mark_mutex. Set the FSNOTIFY_GROUP_NOFS flag to prevent going into direct reclaim from allocations under dnotify group lock and use the safe group lock helpers. Link: https://lore.kernel.org/r/20220422120327.3459282-11-amir73il@gmail.com Suggested-by: Jan Kara Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/dnotify/dnotify.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c index e4779926edf4..190aa717fa32 100644 --- a/fs/notify/dnotify/dnotify.c +++ b/fs/notify/dnotify/dnotify.c @@ -168,7 +168,7 @@ void dnotify_flush(struct file *filp, fl_owner_t id) return; dn_mark = container_of(fsn_mark, struct dnotify_mark, fsn_mark); - mutex_lock(&dnotify_group->mark_mutex); + fsnotify_group_lock(dnotify_group); spin_lock(&fsn_mark->lock); prev = &dn_mark->dn; @@ -191,7 +191,7 @@ void dnotify_flush(struct file *filp, fl_owner_t id) free = true; } - mutex_unlock(&dnotify_group->mark_mutex); + fsnotify_group_unlock(dnotify_group); if (free) fsnotify_free_mark(fsn_mark); @@ -324,7 +324,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg) new_dn_mark->dn = NULL; /* this is needed to prevent the fcntl/close race described below */ - mutex_lock(&dnotify_group->mark_mutex); + fsnotify_group_lock(dnotify_group); /* add the new_fsn_mark or find an old one. */ fsn_mark = fsnotify_find_mark(&inode->i_fsnotify_marks, dnotify_group); @@ -334,7 +334,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg) } else { error = fsnotify_add_inode_mark_locked(new_fsn_mark, inode, 0); if (error) { - mutex_unlock(&dnotify_group->mark_mutex); + fsnotify_group_unlock(dnotify_group); goto out_err; } spin_lock(&new_fsn_mark->lock); @@ -383,7 +383,7 @@ out: if (destroy) fsnotify_detach_mark(fsn_mark); - mutex_unlock(&dnotify_group->mark_mutex); + fsnotify_group_unlock(dnotify_group); if (destroy) fsnotify_free_mark(fsn_mark); fsnotify_put_mark(fsn_mark); @@ -401,7 +401,8 @@ static int __init dnotify_init(void) SLAB_PANIC|SLAB_ACCOUNT); dnotify_mark_cache = KMEM_CACHE(dnotify_mark, SLAB_PANIC|SLAB_ACCOUNT); - dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops, 0); + dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops, + FSNOTIFY_GROUP_NOFS); if (IS_ERR(dnotify_group)) panic("unable to allocate fsnotify group for dnotify\n"); dnotify_sysctl_init(); From c3638b5b13740fa31762d414bbce8b7a694e582a Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:22 +0300 Subject: [PATCH 11/19] fsnotify: allow adding an inode mark without pinning inode fsnotify_add_mark() and variants implicitly take a reference on inode when attaching a mark to an inode. Make that behavior opt-out with the mark flag FSNOTIFY_MARK_FLAG_NO_IREF. Instead of taking the inode reference when attaching connector to inode and dropping the inode reference when detaching connector from inode, take the inode reference on attach of the first mark that wants to hold an inode reference and drop the inode reference on detach of the last mark that wants to hold an inode reference. Backends can "upgrade" an existing mark to take an inode reference, but cannot "downgrade" a mark with inode reference to release the refernce. This leaves the choice to the backend whether or not to pin the inode when adding an inode mark. This is intended to be used when adding a mark with ignored mask that is used for optimization in cases where group can afford getting unneeded events and reinstate the mark with ignored mask when inode is accessed again after being evicted. Link: https://lore.kernel.org/r/20220422120327.3459282-12-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/mark.c | 76 +++++++++++++++++++++++--------- include/linux/fsnotify_backend.h | 2 + 2 files changed, 58 insertions(+), 20 deletions(-) diff --git a/fs/notify/mark.c b/fs/notify/mark.c index 982ca2f20ff5..c74ef947447d 100644 --- a/fs/notify/mark.c +++ b/fs/notify/mark.c @@ -116,20 +116,64 @@ __u32 fsnotify_conn_mask(struct fsnotify_mark_connector *conn) return *fsnotify_conn_mask_p(conn); } -static void __fsnotify_recalc_mask(struct fsnotify_mark_connector *conn) +static void fsnotify_get_inode_ref(struct inode *inode) +{ + ihold(inode); + atomic_long_inc(&inode->i_sb->s_fsnotify_connectors); +} + +/* + * Grab or drop inode reference for the connector if needed. + * + * When it's time to drop the reference, we only clear the HAS_IREF flag and + * return the inode object. fsnotify_drop_object() will be resonsible for doing + * iput() outside of spinlocks. This happens when last mark that wanted iref is + * detached. + */ +static struct inode *fsnotify_update_iref(struct fsnotify_mark_connector *conn, + bool want_iref) +{ + bool has_iref = conn->flags & FSNOTIFY_CONN_FLAG_HAS_IREF; + struct inode *inode = NULL; + + if (conn->type != FSNOTIFY_OBJ_TYPE_INODE || + want_iref == has_iref) + return NULL; + + if (want_iref) { + /* Pin inode if any mark wants inode refcount held */ + fsnotify_get_inode_ref(fsnotify_conn_inode(conn)); + conn->flags |= FSNOTIFY_CONN_FLAG_HAS_IREF; + } else { + /* Unpin inode after detach of last mark that wanted iref */ + inode = fsnotify_conn_inode(conn); + conn->flags &= ~FSNOTIFY_CONN_FLAG_HAS_IREF; + } + + return inode; +} + +static void *__fsnotify_recalc_mask(struct fsnotify_mark_connector *conn) { u32 new_mask = 0; + bool want_iref = false; struct fsnotify_mark *mark; assert_spin_locked(&conn->lock); /* We can get detached connector here when inode is getting unlinked. */ if (!fsnotify_valid_obj_type(conn->type)) - return; + return NULL; hlist_for_each_entry(mark, &conn->list, obj_list) { - if (mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED) - new_mask |= fsnotify_calc_mask(mark); + if (!(mark->flags & FSNOTIFY_MARK_FLAG_ATTACHED)) + continue; + new_mask |= fsnotify_calc_mask(mark); + if (conn->type == FSNOTIFY_OBJ_TYPE_INODE && + !(mark->flags & FSNOTIFY_MARK_FLAG_NO_IREF)) + want_iref = true; } *fsnotify_conn_mask_p(conn) = new_mask; + + return fsnotify_update_iref(conn, want_iref); } /* @@ -169,12 +213,6 @@ static void fsnotify_connector_destroy_workfn(struct work_struct *work) } } -static void fsnotify_get_inode_ref(struct inode *inode) -{ - ihold(inode); - atomic_long_inc(&inode->i_sb->s_fsnotify_connectors); -} - static void fsnotify_put_inode_ref(struct inode *inode) { struct super_block *sb = inode->i_sb; @@ -213,6 +251,10 @@ static void *fsnotify_detach_connector_from_object( if (conn->type == FSNOTIFY_OBJ_TYPE_INODE) { inode = fsnotify_conn_inode(conn); inode->i_fsnotify_mask = 0; + + /* Unpin inode when detaching from connector */ + if (!(conn->flags & FSNOTIFY_CONN_FLAG_HAS_IREF)) + inode = NULL; } else if (conn->type == FSNOTIFY_OBJ_TYPE_VFSMOUNT) { fsnotify_conn_mount(conn)->mnt_fsnotify_mask = 0; } else if (conn->type == FSNOTIFY_OBJ_TYPE_SB) { @@ -274,7 +316,8 @@ void fsnotify_put_mark(struct fsnotify_mark *mark) objp = fsnotify_detach_connector_from_object(conn, &type); free_conn = true; } else { - __fsnotify_recalc_mask(conn); + objp = __fsnotify_recalc_mask(conn); + type = conn->type; } WRITE_ONCE(mark->connector, NULL); spin_unlock(&conn->lock); @@ -497,7 +540,6 @@ static int fsnotify_attach_connector_to_object(fsnotify_connp_t *connp, unsigned int obj_type, __kernel_fsid_t *fsid) { - struct inode *inode = NULL; struct fsnotify_mark_connector *conn; conn = kmem_cache_alloc(fsnotify_mark_connector_cachep, GFP_KERNEL); @@ -505,6 +547,7 @@ static int fsnotify_attach_connector_to_object(fsnotify_connp_t *connp, return -ENOMEM; spin_lock_init(&conn->lock); INIT_HLIST_HEAD(&conn->list); + conn->flags = 0; conn->type = obj_type; conn->obj = connp; /* Cache fsid of filesystem containing the object */ @@ -515,10 +558,6 @@ static int fsnotify_attach_connector_to_object(fsnotify_connp_t *connp, conn->fsid.val[0] = conn->fsid.val[1] = 0; conn->flags = 0; } - if (conn->type == FSNOTIFY_OBJ_TYPE_INODE) { - inode = fsnotify_conn_inode(conn); - fsnotify_get_inode_ref(inode); - } fsnotify_get_sb_connectors(conn); /* @@ -527,8 +566,6 @@ static int fsnotify_attach_connector_to_object(fsnotify_connp_t *connp, */ if (cmpxchg(connp, NULL, conn)) { /* Someone else created list structure for us */ - if (inode) - fsnotify_put_inode_ref(inode); fsnotify_put_sb_connectors(conn); kmem_cache_free(fsnotify_mark_connector_cachep, conn); } @@ -690,8 +727,7 @@ int fsnotify_add_mark_locked(struct fsnotify_mark *mark, if (ret) goto err; - if (mark->mask || mark->ignored_mask) - fsnotify_recalc_mask(mark->connector); + fsnotify_recalc_mask(mark->connector); return ret; err: diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h index d62111e83244..9a1a9e78f69f 100644 --- a/include/linux/fsnotify_backend.h +++ b/include/linux/fsnotify_backend.h @@ -456,6 +456,7 @@ struct fsnotify_mark_connector { spinlock_t lock; unsigned short type; /* Type of object [lock] */ #define FSNOTIFY_CONN_FLAG_HAS_FSID 0x01 +#define FSNOTIFY_CONN_FLAG_HAS_IREF 0x02 unsigned short flags; /* flags [lock] */ __kernel_fsid_t fsid; /* fsid of filesystem containing object */ union { @@ -510,6 +511,7 @@ struct fsnotify_mark { #define FSNOTIFY_MARK_FLAG_IN_ONESHOT 0x0020 /* fanotify mark flags */ #define FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY 0x0100 +#define FSNOTIFY_MARK_FLAG_NO_IREF 0x0200 unsigned int flags; /* flags [mark->lock] */ }; From 4adce25ccfff215939ee465b8c0aa70526d5c352 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:23 +0300 Subject: [PATCH 12/19] fanotify: create helper fanotify_mark_user_flags() To translate from fsnotify mark flags to user visible flags. Link: https://lore.kernel.org/r/20220422120327.3459282-13-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fanotify/fanotify.h | 10 ++++++++++ fs/notify/fdinfo.c | 6 ++---- 2 files changed, 12 insertions(+), 4 deletions(-) diff --git a/fs/notify/fanotify/fanotify.h b/fs/notify/fanotify/fanotify.h index a3d5b751cac5..87142bc0131a 100644 --- a/fs/notify/fanotify/fanotify.h +++ b/fs/notify/fanotify/fanotify.h @@ -490,3 +490,13 @@ static inline unsigned int fanotify_event_hash_bucket( { return event->hash & FANOTIFY_HTABLE_MASK; } + +static inline unsigned int fanotify_mark_user_flags(struct fsnotify_mark *mark) +{ + unsigned int mflags = 0; + + if (mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY) + mflags |= FAN_MARK_IGNORED_SURV_MODIFY; + + return mflags; +} diff --git a/fs/notify/fdinfo.c b/fs/notify/fdinfo.c index 1f34c5c29fdb..59fb40abe33d 100644 --- a/fs/notify/fdinfo.c +++ b/fs/notify/fdinfo.c @@ -14,6 +14,7 @@ #include #include "inotify/inotify.h" +#include "fanotify/fanotify.h" #include "fdinfo.h" #include "fsnotify.h" @@ -103,12 +104,9 @@ void inotify_show_fdinfo(struct seq_file *m, struct file *f) static void fanotify_fdinfo(struct seq_file *m, struct fsnotify_mark *mark) { - unsigned int mflags = 0; + unsigned int mflags = fanotify_mark_user_flags(mark); struct inode *inode; - if (mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY) - mflags |= FAN_MARK_IGNORED_SURV_MODIFY; - if (mark->connector->type == FSNOTIFY_OBJ_TYPE_INODE) { inode = igrab(fsnotify_conn_inode(mark->connector)); if (!inode) From 8998d110835e3781ccd3f1ae061a590b4aaba911 Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:24 +0300 Subject: [PATCH 13/19] fanotify: factor out helper fanotify_mark_update_flags() Handle FAN_MARK_IGNORED_SURV_MODIFY flag change in a helper that is called after updating the mark mask. Replace the added and removed return values and help variables with bool recalc return values and help variable, which makes the code a bit easier to follow. Rename flags argument to fan_flags to emphasize the difference from mark->flags. Link: https://lore.kernel.org/r/20220422120327.3459282-14-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fanotify/fanotify_user.c | 47 ++++++++++++++++-------------- 1 file changed, 25 insertions(+), 22 deletions(-) diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c index 3649c99b3e45..4005ee8e6e2c 100644 --- a/fs/notify/fanotify/fanotify_user.c +++ b/fs/notify/fanotify/fanotify_user.c @@ -1081,42 +1081,45 @@ static int fanotify_remove_inode_mark(struct fsnotify_group *group, flags, umask); } -static void fanotify_mark_add_ignored_mask(struct fsnotify_mark *fsn_mark, - __u32 mask, unsigned int flags, - __u32 *removed) +static bool fanotify_mark_update_flags(struct fsnotify_mark *fsn_mark, + unsigned int fan_flags) { - fsn_mark->ignored_mask |= mask; + bool recalc = false; /* * Setting FAN_MARK_IGNORED_SURV_MODIFY for the first time may lead to * the removal of the FS_MODIFY bit in calculated mask if it was set * because of an ignored mask that is now going to survive FS_MODIFY. */ - if ((flags & FAN_MARK_IGNORED_SURV_MODIFY) && + if ((fan_flags & FAN_MARK_IGNORED_MASK) && + (fan_flags & FAN_MARK_IGNORED_SURV_MODIFY) && !(fsn_mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY)) { fsn_mark->flags |= FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY; if (!(fsn_mark->mask & FS_MODIFY)) - *removed = FS_MODIFY; + recalc = true; } + + return recalc; } -static __u32 fanotify_mark_add_to_mask(struct fsnotify_mark *fsn_mark, - __u32 mask, unsigned int flags, - __u32 *removed) +static bool fanotify_mark_add_to_mask(struct fsnotify_mark *fsn_mark, + __u32 mask, unsigned int fan_flags) { - __u32 oldmask, newmask; + bool recalc; spin_lock(&fsn_mark->lock); - oldmask = fsnotify_calc_mask(fsn_mark); - if (!(flags & FAN_MARK_IGNORED_MASK)) { + if (!(fan_flags & FAN_MARK_IGNORED_MASK)) fsn_mark->mask |= mask; - } else { - fanotify_mark_add_ignored_mask(fsn_mark, mask, flags, removed); - } - newmask = fsnotify_calc_mask(fsn_mark); + else + fsn_mark->ignored_mask |= mask; + + recalc = fsnotify_calc_mask(fsn_mark) & + ~fsnotify_conn_mask(fsn_mark->connector); + + recalc |= fanotify_mark_update_flags(fsn_mark, fan_flags); spin_unlock(&fsn_mark->lock); - return newmask & ~oldmask; + return recalc; } static struct fsnotify_mark *fanotify_add_new_mark(struct fsnotify_group *group, @@ -1170,11 +1173,11 @@ static int fanotify_group_init_error_pool(struct fsnotify_group *group) static int fanotify_add_mark(struct fsnotify_group *group, fsnotify_connp_t *connp, unsigned int obj_type, - __u32 mask, unsigned int flags, + __u32 mask, unsigned int fan_flags, __kernel_fsid_t *fsid) { struct fsnotify_mark *fsn_mark; - __u32 added, removed = 0; + bool recalc; int ret = 0; mutex_lock(&group->mark_mutex); @@ -1191,14 +1194,14 @@ static int fanotify_add_mark(struct fsnotify_group *group, * Error events are pre-allocated per group, only if strictly * needed (i.e. FAN_FS_ERROR was requested). */ - if (!(flags & FAN_MARK_IGNORED_MASK) && (mask & FAN_FS_ERROR)) { + if (!(fan_flags & FAN_MARK_IGNORED_MASK) && (mask & FAN_FS_ERROR)) { ret = fanotify_group_init_error_pool(group); if (ret) goto out; } - added = fanotify_mark_add_to_mask(fsn_mark, mask, flags, &removed); - if (removed || (added & ~fsnotify_conn_mask(fsn_mark->connector))) + recalc = fanotify_mark_add_to_mask(fsn_mark, mask, fan_flags); + if (recalc) fsnotify_recalc_mask(fsn_mark->connector); out: From 7d5e005d982527e4029b0139823d179986e34cdc Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:25 +0300 Subject: [PATCH 14/19] fanotify: implement "evictable" inode marks When an inode mark is created with flag FAN_MARK_EVICTABLE, it will not pin the marked inode to inode cache, so when inode is evicted from cache due to memory pressure, the mark will be lost. When an inode mark with flag FAN_MARK_EVICATBLE is updated without using this flag, the marked inode is pinned to inode cache. When an inode mark is updated with flag FAN_MARK_EVICTABLE but an existing mark already has the inode pinned, the mark update fails with error EEXIST. Evictable inode marks can be used to setup inode marks with ignored mask to suppress events from uninteresting files or directories in a lazy manner, upon receiving the first event, without having to iterate all the uninteresting files or directories before hand. The evictbale inode mark feature allows performing this lazy marks setup without exhausting the system memory with pinned inodes. This change does not enable the feature yet. Link: https://lore.kernel.org/linux-fsdevel/CAOQ4uxiRDpuS=2uA6+ZUM7yG9vVU-u212tkunBmSnP_u=mkv=Q@mail.gmail.com/ Link: https://lore.kernel.org/r/20220422120327.3459282-15-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fanotify/fanotify.h | 2 ++ fs/notify/fanotify/fanotify_user.c | 38 ++++++++++++++++++++++++++++-- include/uapi/linux/fanotify.h | 1 + 3 files changed, 39 insertions(+), 2 deletions(-) diff --git a/fs/notify/fanotify/fanotify.h b/fs/notify/fanotify/fanotify.h index 87142bc0131a..80e0ec95b113 100644 --- a/fs/notify/fanotify/fanotify.h +++ b/fs/notify/fanotify/fanotify.h @@ -497,6 +497,8 @@ static inline unsigned int fanotify_mark_user_flags(struct fsnotify_mark *mark) if (mark->flags & FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY) mflags |= FAN_MARK_IGNORED_SURV_MODIFY; + if (mark->flags & FSNOTIFY_MARK_FLAG_NO_IREF) + mflags |= FAN_MARK_EVICTABLE; return mflags; } diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c index 4005ee8e6e2c..ae36138afead 100644 --- a/fs/notify/fanotify/fanotify_user.c +++ b/fs/notify/fanotify/fanotify_user.c @@ -1084,6 +1084,7 @@ static int fanotify_remove_inode_mark(struct fsnotify_group *group, static bool fanotify_mark_update_flags(struct fsnotify_mark *fsn_mark, unsigned int fan_flags) { + bool want_iref = !(fan_flags & FAN_MARK_EVICTABLE); bool recalc = false; /* @@ -1099,7 +1100,18 @@ static bool fanotify_mark_update_flags(struct fsnotify_mark *fsn_mark, recalc = true; } - return recalc; + if (fsn_mark->connector->type != FSNOTIFY_OBJ_TYPE_INODE || + want_iref == !(fsn_mark->flags & FSNOTIFY_MARK_FLAG_NO_IREF)) + return recalc; + + /* + * NO_IREF may be removed from a mark, but not added. + * When removed, fsnotify_recalc_mask() will take the inode ref. + */ + WARN_ON_ONCE(!want_iref); + fsn_mark->flags &= ~FSNOTIFY_MARK_FLAG_NO_IREF; + + return true; } static bool fanotify_mark_add_to_mask(struct fsnotify_mark *fsn_mark, @@ -1125,6 +1137,7 @@ static bool fanotify_mark_add_to_mask(struct fsnotify_mark *fsn_mark, static struct fsnotify_mark *fanotify_add_new_mark(struct fsnotify_group *group, fsnotify_connp_t *connp, unsigned int obj_type, + unsigned int fan_flags, __kernel_fsid_t *fsid) { struct ucounts *ucounts = group->fanotify_data.ucounts; @@ -1147,6 +1160,9 @@ static struct fsnotify_mark *fanotify_add_new_mark(struct fsnotify_group *group, } fsnotify_init_mark(mark, group); + if (fan_flags & FAN_MARK_EVICTABLE) + mark->flags |= FSNOTIFY_MARK_FLAG_NO_IREF; + ret = fsnotify_add_mark_locked(mark, connp, obj_type, 0, fsid); if (ret) { fsnotify_put_mark(mark); @@ -1183,13 +1199,23 @@ static int fanotify_add_mark(struct fsnotify_group *group, mutex_lock(&group->mark_mutex); fsn_mark = fsnotify_find_mark(connp, group); if (!fsn_mark) { - fsn_mark = fanotify_add_new_mark(group, connp, obj_type, fsid); + fsn_mark = fanotify_add_new_mark(group, connp, obj_type, + fan_flags, fsid); if (IS_ERR(fsn_mark)) { mutex_unlock(&group->mark_mutex); return PTR_ERR(fsn_mark); } } + /* + * Non evictable mark cannot be downgraded to evictable mark. + */ + if (fan_flags & FAN_MARK_EVICTABLE && + !(fsn_mark->flags & FSNOTIFY_MARK_FLAG_NO_IREF)) { + ret = -EEXIST; + goto out; + } + /* * Error events are pre-allocated per group, only if strictly * needed (i.e. FAN_FS_ERROR was requested). @@ -1601,6 +1627,14 @@ static int do_fanotify_mark(int fanotify_fd, unsigned int flags, __u64 mask, mark_type != FAN_MARK_FILESYSTEM) goto fput_and_out; + /* + * Evictable is only relevant for inode marks, because only inode object + * can be evicted on memory pressure. + */ + if (flags & FAN_MARK_EVICTABLE && + mark_type != FAN_MARK_INODE) + goto fput_and_out; + /* * Events that do not carry enough information to report * event->fd require a group that supports reporting fid. Those diff --git a/include/uapi/linux/fanotify.h b/include/uapi/linux/fanotify.h index e8ac38cc2fd6..f1f89132d60e 100644 --- a/include/uapi/linux/fanotify.h +++ b/include/uapi/linux/fanotify.h @@ -82,6 +82,7 @@ #define FAN_MARK_IGNORED_SURV_MODIFY 0x00000040 #define FAN_MARK_FLUSH 0x00000080 /* FAN_MARK_FILESYSTEM is 0x00000100 */ +#define FAN_MARK_EVICTABLE 0x00000200 /* These are NOT bitwise flags. Both bits can be used togther. */ #define FAN_MARK_INODE 0x00000000 From e79719a2ca5c61912c0493bc1367db52759cf6fd Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:26 +0300 Subject: [PATCH 15/19] fanotify: use fsnotify group lock helpers Direct reclaim from fanotify mark allocation context may try to evict inodes with evictable marks of the same group and hit this deadlock: [<0>] fsnotify_destroy_mark+0x1f/0x3a [<0>] fsnotify_destroy_marks+0x71/0xd9 [<0>] __destroy_inode+0x24/0x7e [<0>] destroy_inode+0x2c/0x67 [<0>] dispose_list+0x49/0x68 [<0>] prune_icache_sb+0x5b/0x79 [<0>] super_cache_scan+0x11c/0x16f [<0>] shrink_slab.constprop.0+0x23e/0x40f [<0>] shrink_node+0x218/0x3e7 [<0>] do_try_to_free_pages+0x12a/0x2d2 [<0>] try_to_free_pages+0x166/0x242 [<0>] __alloc_pages_slowpath.constprop.0+0x30c/0x903 [<0>] __alloc_pages+0xeb/0x1c7 [<0>] cache_grow_begin+0x6f/0x31e [<0>] fallback_alloc+0xe0/0x12d [<0>] ____cache_alloc_node+0x15a/0x17e [<0>] kmem_cache_alloc_trace+0xa1/0x143 [<0>] fanotify_add_mark+0xd5/0x2b2 [<0>] do_fanotify_mark+0x566/0x5eb [<0>] __x64_sys_fanotify_mark+0x21/0x24 [<0>] do_syscall_64+0x6d/0x80 [<0>] entry_SYSCALL_64_after_hwframe+0x44/0xae Set the FSNOTIFY_GROUP_NOFS flag to prevent going into direct reclaim from allocations under fanotify group lock and use the safe group lock helpers. Link: https://lore.kernel.org/r/20220422120327.3459282-16-amir73il@gmail.com Suggested-by: Jan Kara Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fanotify/fanotify_user.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c index ae36138afead..228cf25e9230 100644 --- a/fs/notify/fanotify/fanotify_user.c +++ b/fs/notify/fanotify/fanotify_user.c @@ -1035,10 +1035,10 @@ static int fanotify_remove_mark(struct fsnotify_group *group, __u32 removed; int destroy_mark; - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); fsn_mark = fsnotify_find_mark(connp, group); if (!fsn_mark) { - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); return -ENOENT; } @@ -1048,7 +1048,7 @@ static int fanotify_remove_mark(struct fsnotify_group *group, fsnotify_recalc_mask(fsn_mark->connector); if (destroy_mark) fsnotify_detach_mark(fsn_mark); - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); if (destroy_mark) fsnotify_free_mark(fsn_mark); @@ -1196,13 +1196,13 @@ static int fanotify_add_mark(struct fsnotify_group *group, bool recalc; int ret = 0; - mutex_lock(&group->mark_mutex); + fsnotify_group_lock(group); fsn_mark = fsnotify_find_mark(connp, group); if (!fsn_mark) { fsn_mark = fanotify_add_new_mark(group, connp, obj_type, fan_flags, fsid); if (IS_ERR(fsn_mark)) { - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); return PTR_ERR(fsn_mark); } } @@ -1231,7 +1231,7 @@ static int fanotify_add_mark(struct fsnotify_group *group, fsnotify_recalc_mask(fsn_mark->connector); out: - mutex_unlock(&group->mark_mutex); + fsnotify_group_unlock(group); fsnotify_put_mark(fsn_mark); return ret; @@ -1385,7 +1385,7 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags) /* fsnotify_alloc_group takes a ref. Dropped in fanotify_release */ group = fsnotify_alloc_group(&fanotify_fsnotify_ops, - FSNOTIFY_GROUP_USER); + FSNOTIFY_GROUP_USER | FSNOTIFY_GROUP_NOFS); if (IS_ERR(group)) { return PTR_ERR(group); } From 5f9d3bd520261fd7a850818c71809fd580e0f30c Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Fri, 22 Apr 2022 15:03:27 +0300 Subject: [PATCH 16/19] fanotify: enable "evictable" inode marks Now that the direct reclaim path is handled we can enable evictable inode marks. Link: https://lore.kernel.org/r/20220422120327.3459282-17-amir73il@gmail.com Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara --- fs/notify/fanotify/fanotify_user.c | 2 +- include/linux/fanotify.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c index 228cf25e9230..edad67d674dc 100644 --- a/fs/notify/fanotify/fanotify_user.c +++ b/fs/notify/fanotify/fanotify_user.c @@ -1787,7 +1787,7 @@ static int __init fanotify_user_setup(void) BUILD_BUG_ON(FANOTIFY_INIT_FLAGS & FANOTIFY_INTERNAL_GROUP_FLAGS); BUILD_BUG_ON(HWEIGHT32(FANOTIFY_INIT_FLAGS) != 12); - BUILD_BUG_ON(HWEIGHT32(FANOTIFY_MARK_FLAGS) != 9); + BUILD_BUG_ON(HWEIGHT32(FANOTIFY_MARK_FLAGS) != 10); fanotify_mark_cache = KMEM_CACHE(fsnotify_mark, SLAB_PANIC|SLAB_ACCOUNT); diff --git a/include/linux/fanotify.h b/include/linux/fanotify.h index 419cadcd7ff5..edc28555814c 100644 --- a/include/linux/fanotify.h +++ b/include/linux/fanotify.h @@ -66,6 +66,7 @@ FAN_MARK_ONLYDIR | \ FAN_MARK_IGNORED_MASK | \ FAN_MARK_IGNORED_SURV_MODIFY | \ + FAN_MARK_EVICTABLE | \ FAN_MARK_FLUSH) /* From 14362a2541797cf9df0e86fb12dcd7950baf566e Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Wed, 11 May 2022 22:02:12 +0300 Subject: [PATCH 17/19] fsnotify: introduce mark type iterator fsnotify_foreach_iter_mark_type() is used to reduce boilerplate code of iterating all marks of a specific group interested in an event by consulting the iterator report_mask. Use an open coded version of that iterator in fsnotify_iter_next() that collects all marks of the current iteration group without consulting the iterator report_mask. At the moment, the two iterator variants are the same, but this decoupling will allow us to exclude some of the group's marks from reporting the event, for example for event on child and inode marks on parent did not request to watch events on children. Fixes: 2f02fd3fa13e ("fanotify: fix ignore mask logic for events on child and on dir") Reported-by: Jan Kara Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara Link: https://lore.kernel.org/r/20220511190213.831646-2-amir73il@gmail.com --- fs/notify/fanotify/fanotify.c | 14 +++------ fs/notify/fsnotify.c | 53 ++++++++++++++++---------------- include/linux/fsnotify_backend.h | 31 ++++++++++++++----- 3 files changed, 54 insertions(+), 44 deletions(-) diff --git a/fs/notify/fanotify/fanotify.c b/fs/notify/fanotify/fanotify.c index 985e995d2a39..263d303d8f8f 100644 --- a/fs/notify/fanotify/fanotify.c +++ b/fs/notify/fanotify/fanotify.c @@ -319,11 +319,7 @@ static u32 fanotify_group_event_mask(struct fsnotify_group *group, return 0; } - fsnotify_foreach_iter_type(type) { - if (!fsnotify_iter_should_report_type(iter_info, type)) - continue; - mark = iter_info->marks[type]; - + fsnotify_foreach_iter_mark_type(iter_info, mark, type) { /* Apply ignore mask regardless of ISDIR and ON_CHILD flags */ marks_ignored_mask |= mark->ignored_mask; @@ -849,16 +845,14 @@ out: */ static __kernel_fsid_t fanotify_get_fsid(struct fsnotify_iter_info *iter_info) { + struct fsnotify_mark *mark; int type; __kernel_fsid_t fsid = {}; - fsnotify_foreach_iter_type(type) { + fsnotify_foreach_iter_mark_type(iter_info, mark, type) { struct fsnotify_mark_connector *conn; - if (!fsnotify_iter_should_report_type(iter_info, type)) - continue; - - conn = READ_ONCE(iter_info->marks[type]->connector); + conn = READ_ONCE(mark->connector); /* Mark is just getting destroyed or created? */ if (!conn) continue; diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c index 6eee19d15e8c..35740a64ee45 100644 --- a/fs/notify/fsnotify.c +++ b/fs/notify/fsnotify.c @@ -335,31 +335,23 @@ static int send_to_group(__u32 mask, const void *data, int data_type, struct fsnotify_mark *mark; int type; - if (WARN_ON(!iter_info->report_mask)) + if (!iter_info->report_mask) return 0; /* clear ignored on inode modification */ if (mask & FS_MODIFY) { - fsnotify_foreach_iter_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)) + fsnotify_foreach_iter_mark_type(iter_info, mark, type) { + if (!(mark->flags & + FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY)) mark->ignored_mask = 0; } } - fsnotify_foreach_iter_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; - } + /* Are any of the group marks interested in this event? */ + fsnotify_foreach_iter_mark_type(iter_info, mark, type) { + group = mark->group; + marks_mask |= mark->mask; + marks_ignored_mask |= mark->ignored_mask; } pr_debug("%s: group=%p mask=%x marks_mask=%x marks_ignored_mask=%x data=%p data_type=%d dir=%p cookie=%d\n", @@ -403,11 +395,11 @@ static struct fsnotify_mark *fsnotify_next_mark(struct fsnotify_mark *mark) /* * 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. + * Pick a subset of marks from queue heads, all with the same group + * and set the report_mask to a subset of the selected marks. + * Returns false if there are no more groups to iterate. */ -static unsigned int fsnotify_iter_select_report_types( +static bool fsnotify_iter_select_report_types( struct fsnotify_iter_info *iter_info) { struct fsnotify_group *max_prio_group = NULL; @@ -423,30 +415,37 @@ static unsigned int fsnotify_iter_select_report_types( } if (!max_prio_group) - return 0; + return false; /* Set the report mask for marks from same group as max prio group */ + iter_info->current_group = max_prio_group; iter_info->report_mask = 0; fsnotify_foreach_iter_type(type) { mark = iter_info->marks[type]; - if (mark && - fsnotify_compare_groups(max_prio_group, mark->group) == 0) + if (mark && mark->group == iter_info->current_group) fsnotify_iter_set_report_type(iter_info, type); } - return iter_info->report_mask; + return true; } /* - * Pop from iter_info multi head queue, the marks that were iterated in the + * Pop from iter_info multi head queue, the marks that belong to the group of * current iteration step. */ static void fsnotify_iter_next(struct fsnotify_iter_info *iter_info) { + struct fsnotify_mark *mark; int type; + /* + * We cannot use fsnotify_foreach_iter_mark_type() here because we + * may need to advance a mark of type X that belongs to current_group + * but was not selected for reporting. + */ fsnotify_foreach_iter_type(type) { - if (fsnotify_iter_should_report_type(iter_info, type)) + mark = iter_info->marks[type]; + if (mark && mark->group == iter_info->current_group) iter_info->marks[type] = fsnotify_next_mark(iter_info->marks[type]); } diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h index 9a1a9e78f69f..9560734759fa 100644 --- a/include/linux/fsnotify_backend.h +++ b/include/linux/fsnotify_backend.h @@ -399,6 +399,7 @@ static inline bool fsnotify_valid_obj_type(unsigned int obj_type) struct fsnotify_iter_info { struct fsnotify_mark *marks[FSNOTIFY_ITER_TYPE_COUNT]; + struct fsnotify_group *current_group; unsigned int report_mask; int srcu_idx; }; @@ -415,20 +416,31 @@ static inline void fsnotify_iter_set_report_type( iter_info->report_mask |= (1U << iter_type); } -static inline void fsnotify_iter_set_report_type_mark( - struct fsnotify_iter_info *iter_info, int iter_type, - struct fsnotify_mark *mark) +static inline struct fsnotify_mark *fsnotify_iter_mark( + struct fsnotify_iter_info *iter_info, int iter_type) { - iter_info->marks[iter_type] = mark; - iter_info->report_mask |= (1U << iter_type); + if (fsnotify_iter_should_report_type(iter_info, iter_type)) + return iter_info->marks[iter_type]; + return NULL; +} + +static inline int fsnotify_iter_step(struct fsnotify_iter_info *iter, int type, + struct fsnotify_mark **markp) +{ + while (type < FSNOTIFY_ITER_TYPE_COUNT) { + *markp = fsnotify_iter_mark(iter, type); + if (*markp) + break; + type++; + } + return type; } #define FSNOTIFY_ITER_FUNCS(name, NAME) \ static inline struct fsnotify_mark *fsnotify_iter_##name##_mark( \ struct fsnotify_iter_info *iter_info) \ { \ - return (iter_info->report_mask & (1U << FSNOTIFY_ITER_TYPE_##NAME)) ? \ - iter_info->marks[FSNOTIFY_ITER_TYPE_##NAME] : NULL; \ + return fsnotify_iter_mark(iter_info, FSNOTIFY_ITER_TYPE_##NAME); \ } FSNOTIFY_ITER_FUNCS(inode, INODE) @@ -438,6 +450,11 @@ FSNOTIFY_ITER_FUNCS(sb, SB) #define fsnotify_foreach_iter_type(type) \ for (type = 0; type < FSNOTIFY_ITER_TYPE_COUNT; type++) +#define fsnotify_foreach_iter_mark_type(iter, mark, type) \ + for (type = 0; \ + type = fsnotify_iter_step(iter, type, &mark), \ + type < FSNOTIFY_ITER_TYPE_COUNT; \ + type++) /* * fsnotify_connp_t is what we embed in objects which connector can be attached From e730558adffb88a52e562db089e969ee9510184a Mon Sep 17 00:00:00 2001 From: Amir Goldstein Date: Wed, 11 May 2022 22:02:13 +0300 Subject: [PATCH 18/19] fsnotify: consistent behavior for parent not watching children The logic for handling events on child in groups that have a mark on the parent inode, but without FS_EVENT_ON_CHILD flag in the mask is duplicated in several places and inconsistent. Move the logic into the preparation of mark type iterator, so that the parent mark type will be excluded from all mark type iterations in that case. This results in several subtle changes of behavior, hopefully all desired changes of behavior, for example: - Group A has a mount mark with FS_MODIFY in mask - Group A has a mark with ignore mask that does not survive FS_MODIFY and does not watch children on directory D. - Group B has a mark with FS_MODIFY in mask that does watch children on directory D. - FS_MODIFY event on file D/foo should not clear the ignore mask of group A, but before this change it does And if group A ignore mask was set to survive FS_MODIFY: - FS_MODIFY event on file D/foo should be reported to group A on account of the mount mark, but before this change it is wrongly ignored Fixes: 2f02fd3fa13e ("fanotify: fix ignore mask logic for events on child and on dir") Reported-by: Jan Kara Link: https://lore.kernel.org/linux-fsdevel/20220314113337.j7slrb5srxukztje@quack3.lan/ Signed-off-by: Amir Goldstein Signed-off-by: Jan Kara Link: https://lore.kernel.org/r/20220511190213.831646-3-amir73il@gmail.com --- fs/notify/fanotify/fanotify.c | 10 +--------- fs/notify/fsnotify.c | 34 +++++++++++++++++++--------------- 2 files changed, 20 insertions(+), 24 deletions(-) diff --git a/fs/notify/fanotify/fanotify.c b/fs/notify/fanotify/fanotify.c index 263d303d8f8f..4f897e109547 100644 --- a/fs/notify/fanotify/fanotify.c +++ b/fs/notify/fanotify/fanotify.c @@ -320,7 +320,7 @@ static u32 fanotify_group_event_mask(struct fsnotify_group *group, } fsnotify_foreach_iter_mark_type(iter_info, mark, type) { - /* Apply ignore mask regardless of ISDIR and ON_CHILD flags */ + /* Apply ignore mask regardless of mark's ISDIR flag */ marks_ignored_mask |= mark->ignored_mask; /* @@ -330,14 +330,6 @@ static u32 fanotify_group_event_mask(struct fsnotify_group *group, if (event_mask & FS_ISDIR && !(mark->mask & FS_ISDIR)) continue; - /* - * If the event is on a child and this mark is on a parent not - * watching children, don't send it! - */ - if (type == FSNOTIFY_ITER_TYPE_PARENT && - !(mark->mask & FS_EVENT_ON_CHILD)) - continue; - marks_mask |= mark->mask; /* Record the mark types of this group that matched the event */ diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c index 35740a64ee45..0b3e74935cb4 100644 --- a/fs/notify/fsnotify.c +++ b/fs/notify/fsnotify.c @@ -290,22 +290,15 @@ static int fsnotify_handle_event(struct fsnotify_group *group, __u32 mask, } if (parent_mark) { - /* - * parent_mark indicates that the parent inode is watching - * children and interested in this event, which is an event - * possible on child. But is *this mark* watching children and - * interested in this event? - */ - if (parent_mark->mask & FS_EVENT_ON_CHILD) { - ret = fsnotify_handle_inode_event(group, parent_mark, mask, - data, data_type, dir, name, 0); - if (ret) - return ret; - } - if (!inode_mark) - return 0; + ret = fsnotify_handle_inode_event(group, parent_mark, mask, + data, data_type, dir, name, 0); + if (ret) + return ret; } + if (!inode_mark) + return 0; + if (mask & FS_EVENT_ON_CHILD) { /* * Some events can be sent on both parent dir and child marks @@ -422,8 +415,19 @@ static bool fsnotify_iter_select_report_types( iter_info->report_mask = 0; fsnotify_foreach_iter_type(type) { mark = iter_info->marks[type]; - if (mark && mark->group == iter_info->current_group) + if (mark && mark->group == iter_info->current_group) { + /* + * FSNOTIFY_ITER_TYPE_PARENT indicates that this inode + * is watching children and interested in this event, + * which is an event possible on child. + * But is *this mark* watching children? + */ + if (type == FSNOTIFY_ITER_TYPE_PARENT && + !(mark->mask & FS_EVENT_ON_CHILD)) + continue; + fsnotify_iter_set_report_type(iter_info, type); + } } return true; From dccd855771b37820b6d976a99729c88259549f85 Mon Sep 17 00:00:00 2001 From: Vasily Averin Date: Sun, 22 May 2022 15:08:02 +0300 Subject: [PATCH 19/19] fanotify: fix incorrect fmode_t casts Fixes sparce warnings: fs/notify/fanotify/fanotify_user.c:267:63: sparse: warning: restricted fmode_t degrades to integer fs/notify/fanotify/fanotify_user.c:1351:28: sparse: warning: restricted fmode_t degrades to integer FMODE_NONTIFY have bitwise fmode_t type and requires __force attribute for any casts. Signed-off-by: Vasily Averin Reviewed-by: Christian Brauner (Microsoft) Reviewed-by: Christoph Hellwig Signed-off-by: Jan Kara Link: https://lore.kernel.org/r/9adfd6ac-1b89-791e-796b-49ada3293985@openvz.org --- fs/notify/fanotify/fanotify_user.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c index edad67d674dc..b81064cdcde1 100644 --- a/fs/notify/fanotify/fanotify_user.c +++ b/fs/notify/fanotify/fanotify_user.c @@ -264,7 +264,7 @@ static int create_fd(struct fsnotify_group *group, struct path *path, * originally opened O_WRONLY. */ new_file = dentry_open(path, - group->fanotify_data.f_flags | FMODE_NONOTIFY, + group->fanotify_data.f_flags | __FMODE_NONOTIFY, current_cred()); if (IS_ERR(new_file)) { /* @@ -1377,7 +1377,7 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags) (!(fid_mode & FAN_REPORT_NAME) || !(fid_mode & FAN_REPORT_FID))) return -EINVAL; - f_flags = O_RDWR | FMODE_NONOTIFY; + f_flags = O_RDWR | __FMODE_NONOTIFY; if (flags & FAN_CLOEXEC) f_flags |= O_CLOEXEC; if (flags & FAN_NONBLOCK)