summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--fs/notify/fanotify/fanotify.c16
-rw-r--r--fs/notify/fanotify/fanotify_user.c123
-rw-r--r--fs/notify/group.c1
-rw-r--r--fs/notify/mark.c4
-rw-r--r--include/linux/fanotify.h3
-rw-r--r--include/linux/fsnotify_backend.h6
-rw-r--r--include/linux/sched/user.h3
-rw-r--r--include/linux/user_namespace.h4
-rw-r--r--kernel/sysctl.c12
-rw-r--r--kernel/ucount.c4
10 files changed, 137 insertions, 39 deletions
diff --git a/fs/notify/fanotify/fanotify.c b/fs/notify/fanotify/fanotify.c
index 754e27ead874..057abd2cf887 100644
--- a/fs/notify/fanotify/fanotify.c
+++ b/fs/notify/fanotify/fanotify.c
@@ -801,12 +801,10 @@ finish:
static void fanotify_free_group_priv(struct fsnotify_group *group)
{
- struct user_struct *user;
-
kfree(group->fanotify_data.merge_hash);
- user = group->fanotify_data.user;
- atomic_dec(&user->fanotify_listeners);
- free_uid(user);
+ if (group->fanotify_data.ucounts)
+ dec_ucount(group->fanotify_data.ucounts,
+ UCOUNT_FANOTIFY_GROUPS);
}
static void fanotify_free_path_event(struct fanotify_event *event)
@@ -862,6 +860,13 @@ static void fanotify_free_event(struct fsnotify_event *fsn_event)
}
}
+static void fanotify_freeing_mark(struct fsnotify_mark *mark,
+ struct fsnotify_group *group)
+{
+ if (!FAN_GROUP_FLAG(group, FAN_UNLIMITED_MARKS))
+ dec_ucount(group->fanotify_data.ucounts, UCOUNT_FANOTIFY_MARKS);
+}
+
static void fanotify_free_mark(struct fsnotify_mark *fsn_mark)
{
kmem_cache_free(fanotify_mark_cache, fsn_mark);
@@ -871,5 +876,6 @@ const struct fsnotify_ops fanotify_fsnotify_ops = {
.handle_event = fanotify_handle_event,
.free_group_priv = fanotify_free_group_priv,
.free_event = fanotify_free_event,
+ .freeing_mark = fanotify_freeing_mark,
.free_mark = fanotify_free_mark,
};
diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c
index b89f332248bd..e81848e09646 100644
--- a/fs/notify/fanotify/fanotify_user.c
+++ b/fs/notify/fanotify/fanotify_user.c
@@ -27,8 +27,61 @@
#include "fanotify.h"
#define FANOTIFY_DEFAULT_MAX_EVENTS 16384
-#define FANOTIFY_DEFAULT_MAX_MARKS 8192
-#define FANOTIFY_DEFAULT_MAX_LISTENERS 128
+#define FANOTIFY_OLD_DEFAULT_MAX_MARKS 8192
+#define FANOTIFY_DEFAULT_MAX_GROUPS 128
+
+/*
+ * Legacy fanotify marks limits (8192) is per group and we introduced a tunable
+ * limit of marks per user, similar to inotify. Effectively, the legacy limit
+ * of fanotify marks per user is <max marks per group> * <max groups per user>.
+ * This default limit (1M) also happens to match the increased limit of inotify
+ * max_user_watches since v5.10.
+ */
+#define FANOTIFY_DEFAULT_MAX_USER_MARKS \
+ (FANOTIFY_OLD_DEFAULT_MAX_MARKS * FANOTIFY_DEFAULT_MAX_GROUPS)
+
+/*
+ * Most of the memory cost of adding an inode mark is pinning the marked inode.
+ * The size of the filesystem inode struct is not uniform across filesystems,
+ * so double the size of a VFS inode is used as a conservative approximation.
+ */
+#define INODE_MARK_COST (2 * sizeof(struct inode))
+
+/* configurable via /proc/sys/fs/fanotify/ */
+static int fanotify_max_queued_events __read_mostly;
+
+#ifdef CONFIG_SYSCTL
+
+#include <linux/sysctl.h>
+
+struct ctl_table fanotify_table[] = {
+ {
+ .procname = "max_user_groups",
+ .data = &init_user_ns.ucount_max[UCOUNT_FANOTIFY_GROUPS],
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = SYSCTL_ZERO,
+ },
+ {
+ .procname = "max_user_marks",
+ .data = &init_user_ns.ucount_max[UCOUNT_FANOTIFY_MARKS],
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = SYSCTL_ZERO,
+ },
+ {
+ .procname = "max_queued_events",
+ .data = &fanotify_max_queued_events,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = SYSCTL_ZERO
+ },
+ { }
+};
+#endif /* CONFIG_SYSCTL */
/*
* All flags that may be specified in parameter event_f_flags of fanotify_init.
@@ -847,24 +900,38 @@ static struct fsnotify_mark *fanotify_add_new_mark(struct fsnotify_group *group,
unsigned int type,
__kernel_fsid_t *fsid)
{
+ struct ucounts *ucounts = group->fanotify_data.ucounts;
struct fsnotify_mark *mark;
int ret;
- if (atomic_read(&group->num_marks) > group->fanotify_data.max_marks)
+ /*
+ * Enforce per user marks limits per user in all containing user ns.
+ * A group with FAN_UNLIMITED_MARKS does not contribute to mark count
+ * in the limited groups account.
+ */
+ if (!FAN_GROUP_FLAG(group, FAN_UNLIMITED_MARKS) &&
+ !inc_ucount(ucounts->ns, ucounts->uid, UCOUNT_FANOTIFY_MARKS))
return ERR_PTR(-ENOSPC);
mark = kmem_cache_alloc(fanotify_mark_cache, GFP_KERNEL);
- if (!mark)
- return ERR_PTR(-ENOMEM);
+ if (!mark) {
+ ret = -ENOMEM;
+ goto out_dec_ucounts;
+ }
fsnotify_init_mark(mark, group);
ret = fsnotify_add_mark_locked(mark, connp, type, 0, fsid);
if (ret) {
fsnotify_put_mark(mark);
- return ERR_PTR(ret);
+ goto out_dec_ucounts;
}
return mark;
+
+out_dec_ucounts:
+ if (!FAN_GROUP_FLAG(group, FAN_UNLIMITED_MARKS))
+ dec_ucount(ucounts, UCOUNT_FANOTIFY_MARKS);
+ return ERR_PTR(ret);
}
@@ -963,7 +1030,6 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
{
struct fsnotify_group *group;
int f_flags, fd;
- struct user_struct *user;
unsigned int fid_mode = flags & FANOTIFY_FID_BITS;
unsigned int class = flags & FANOTIFY_CLASS_BITS;
@@ -1002,12 +1068,6 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
if ((fid_mode & FAN_REPORT_NAME) && !(fid_mode & FAN_REPORT_DIR_FID))
return -EINVAL;
- user = get_current_user();
- if (atomic_read(&user->fanotify_listeners) > FANOTIFY_DEFAULT_MAX_LISTENERS) {
- free_uid(user);
- return -EMFILE;
- }
-
f_flags = O_RDWR | FMODE_NONOTIFY;
if (flags & FAN_CLOEXEC)
f_flags |= O_CLOEXEC;
@@ -1017,13 +1077,19 @@ 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_user_group(&fanotify_fsnotify_ops);
if (IS_ERR(group)) {
- free_uid(user);
return PTR_ERR(group);
}
- group->fanotify_data.user = user;
+ /* Enforce groups limits per user in all containing user ns */
+ group->fanotify_data.ucounts = inc_ucount(current_user_ns(),
+ current_euid(),
+ UCOUNT_FANOTIFY_GROUPS);
+ if (!group->fanotify_data.ucounts) {
+ fd = -EMFILE;
+ goto out_destroy_group;
+ }
+
group->fanotify_data.flags = flags;
- atomic_inc(&user->fanotify_listeners);
group->memcg = get_mem_cgroup_from_mm(current->mm);
group->fanotify_data.merge_hash = fanotify_alloc_merge_hash();
@@ -1064,16 +1130,13 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
goto out_destroy_group;
group->max_events = UINT_MAX;
} else {
- group->max_events = FANOTIFY_DEFAULT_MAX_EVENTS;
+ group->max_events = fanotify_max_queued_events;
}
if (flags & FAN_UNLIMITED_MARKS) {
fd = -EPERM;
if (!capable(CAP_SYS_ADMIN))
goto out_destroy_group;
- group->fanotify_data.max_marks = UINT_MAX;
- } else {
- group->fanotify_data.max_marks = FANOTIFY_DEFAULT_MAX_MARKS;
}
if (flags & FAN_ENABLE_AUDIT) {
@@ -1357,6 +1420,21 @@ SYSCALL32_DEFINE6(fanotify_mark,
*/
static int __init fanotify_user_setup(void)
{
+ struct sysinfo si;
+ int max_marks;
+
+ si_meminfo(&si);
+ /*
+ * Allow up to 1% of addressable memory to be accounted for per user
+ * marks limited to the range [8192, 1048576]. mount and sb marks are
+ * a lot cheaper than inode marks, but there is no reason for a user
+ * to have many of those, so calculate by the cost of inode marks.
+ */
+ max_marks = (((si.totalram - si.totalhigh) / 100) << PAGE_SHIFT) /
+ INODE_MARK_COST;
+ max_marks = clamp(max_marks, FANOTIFY_OLD_DEFAULT_MAX_MARKS,
+ FANOTIFY_DEFAULT_MAX_USER_MARKS);
+
BUILD_BUG_ON(HWEIGHT32(FANOTIFY_INIT_FLAGS) != 10);
BUILD_BUG_ON(HWEIGHT32(FANOTIFY_MARK_FLAGS) != 9);
@@ -1371,6 +1449,11 @@ static int __init fanotify_user_setup(void)
KMEM_CACHE(fanotify_perm_event, SLAB_PANIC);
}
+ fanotify_max_queued_events = FANOTIFY_DEFAULT_MAX_EVENTS;
+ init_user_ns.ucount_max[UCOUNT_FANOTIFY_GROUPS] =
+ FANOTIFY_DEFAULT_MAX_GROUPS;
+ init_user_ns.ucount_max[UCOUNT_FANOTIFY_MARKS] = max_marks;
+
return 0;
}
device_initcall(fanotify_user_setup);
diff --git a/fs/notify/group.c b/fs/notify/group.c
index ffd723ffe46d..fb89c351295d 100644
--- a/fs/notify/group.c
+++ b/fs/notify/group.c
@@ -122,7 +122,6 @@ static struct fsnotify_group *__fsnotify_alloc_group(
/* set to 0 when there a no external references to this group */
refcount_set(&group->refcnt, 1);
- atomic_set(&group->num_marks, 0);
atomic_set(&group->user_waits, 0);
spin_lock_init(&group->notification_lock);
diff --git a/fs/notify/mark.c b/fs/notify/mark.c
index 8387937b9d01..d32ab349db74 100644
--- a/fs/notify/mark.c
+++ b/fs/notify/mark.c
@@ -391,8 +391,6 @@ void fsnotify_detach_mark(struct fsnotify_mark *mark)
list_del_init(&mark->g_list);
spin_unlock(&mark->lock);
- atomic_dec(&group->num_marks);
-
/* Drop mark reference acquired in fsnotify_add_mark_locked() */
fsnotify_put_mark(mark);
}
@@ -656,7 +654,6 @@ int fsnotify_add_mark_locked(struct fsnotify_mark *mark,
mark->flags |= FSNOTIFY_MARK_FLAG_ALIVE | FSNOTIFY_MARK_FLAG_ATTACHED;
list_add(&mark->g_list, &group->marks_list);
- atomic_inc(&group->num_marks);
fsnotify_get_mark(mark); /* for g_list */
spin_unlock(&mark->lock);
@@ -674,7 +671,6 @@ err:
FSNOTIFY_MARK_FLAG_ATTACHED);
list_del_init(&mark->g_list);
spin_unlock(&mark->lock);
- atomic_dec(&group->num_marks);
fsnotify_put_mark(mark);
return ret;
diff --git a/include/linux/fanotify.h b/include/linux/fanotify.h
index 3e9c56ee651f..031a97d8369a 100644
--- a/include/linux/fanotify.h
+++ b/include/linux/fanotify.h
@@ -2,8 +2,11 @@
#ifndef _LINUX_FANOTIFY_H
#define _LINUX_FANOTIFY_H
+#include <linux/sysctl.h>
#include <uapi/linux/fanotify.h>
+extern struct ctl_table fanotify_table[]; /* for sysctl */
+
#define FAN_GROUP_FLAG(group, flag) \
((group)->fanotify_data.flags & (flag))
diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h
index 63fb766f0f3e..1ce66748a2d2 100644
--- a/include/linux/fsnotify_backend.h
+++ b/include/linux/fsnotify_backend.h
@@ -206,9 +206,6 @@ struct fsnotify_group {
/* stores all fastpath marks assoc with this group so they can be cleaned on unregister */
struct mutex mark_mutex; /* protect marks_list */
- atomic_t num_marks; /* 1 for each mark and 1 for not being
- * past the point of no return when freeing
- * a group */
atomic_t user_waits; /* Number of tasks waiting for user
* response */
struct list_head marks_list; /* all inode marks for this group */
@@ -240,8 +237,7 @@ struct fsnotify_group {
wait_queue_head_t access_waitq;
int flags; /* flags from fanotify_init() */
int f_flags; /* event_f_flags from fanotify_init() */
- unsigned int max_marks;
- struct user_struct *user;
+ struct ucounts *ucounts;
} fanotify_data;
#endif /* CONFIG_FANOTIFY */
};
diff --git a/include/linux/sched/user.h b/include/linux/sched/user.h
index a8ec3b6093fc..3632c5d6ec55 100644
--- a/include/linux/sched/user.h
+++ b/include/linux/sched/user.h
@@ -14,9 +14,6 @@ struct user_struct {
refcount_t __count; /* reference count */
atomic_t processes; /* How many processes does this user have? */
atomic_t sigpending; /* How many pending signals does this user have? */
-#ifdef CONFIG_FANOTIFY
- atomic_t fanotify_listeners;
-#endif
#ifdef CONFIG_EPOLL
atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
#endif
diff --git a/include/linux/user_namespace.h b/include/linux/user_namespace.h
index 64cf8ebdc4ec..f0d961a15fba 100644
--- a/include/linux/user_namespace.h
+++ b/include/linux/user_namespace.h
@@ -50,6 +50,10 @@ enum ucount_type {
UCOUNT_INOTIFY_INSTANCES,
UCOUNT_INOTIFY_WATCHES,
#endif
+#ifdef CONFIG_FANOTIFY
+ UCOUNT_FANOTIFY_GROUPS,
+ UCOUNT_FANOTIFY_MARKS,
+#endif
UCOUNT_COUNTS,
};
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 62fbd09b5dc1..4b6b9de89da8 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -148,6 +148,9 @@ static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
#ifdef CONFIG_INOTIFY_USER
#include <linux/inotify.h>
#endif
+#ifdef CONFIG_FANOTIFY
+#include <linux/fanotify.h>
+#endif
#ifdef CONFIG_PROC_SYSCTL
@@ -3258,7 +3261,14 @@ static struct ctl_table fs_table[] = {
.mode = 0555,
.child = inotify_table,
},
-#endif
+#endif
+#ifdef CONFIG_FANOTIFY
+ {
+ .procname = "fanotify",
+ .mode = 0555,
+ .child = fanotify_table,
+ },
+#endif
#ifdef CONFIG_EPOLL
{
.procname = "epoll",
diff --git a/kernel/ucount.c b/kernel/ucount.c
index 11b1596e2542..8d8874f1c35e 100644
--- a/kernel/ucount.c
+++ b/kernel/ucount.c
@@ -74,6 +74,10 @@ static struct ctl_table user_table[] = {
UCOUNT_ENTRY("max_inotify_instances"),
UCOUNT_ENTRY("max_inotify_watches"),
#endif
+#ifdef CONFIG_FANOTIFY
+ UCOUNT_ENTRY("max_fanotify_groups"),
+ UCOUNT_ENTRY("max_fanotify_marks"),
+#endif
{ }
};
#endif /* CONFIG_SYSCTL */
generated by cgit v1.2.3 (git 2.39.0) at 2024-11-25 16:27:45 +0300