fanotify: Pre-allocate pool of error events

Pre-allocate slots for file system errors to have greater chances of
succeeding, since error events can happen in GFP_NOFS context.  This
patch introduces a group-wide mempool of error events, shared by all
FAN_FS_ERROR marks in this group.

Link: https://lore.kernel.org/r/20211025192746.66445-20-krisman@collabora.com
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.com>
Signed-off-by: Jan Kara <jack@suse.cz>

fs/notify/fanotify/fanotify.c

@@ -819,6 +819,9 @@ static void fanotify_free_group_priv(struct fsnotify_group *group)
 	if (group->fanotify_data.ucounts)
 		dec_ucount(group->fanotify_data.ucounts,
 			   UCOUNT_FANOTIFY_GROUPS);
+
+	if (mempool_initialized(&group->fanotify_data.error_events_pool))
+		mempool_exit(&group->fanotify_data.error_events_pool);
 }
 
 static void fanotify_free_path_event(struct fanotify_event *event)

fs/notify/fanotify/fanotify.h

@@ -141,6 +141,7 @@ enum fanotify_event_type {
 	FANOTIFY_EVENT_TYPE_PATH,
 	FANOTIFY_EVENT_TYPE_PATH_PERM,
 	FANOTIFY_EVENT_TYPE_OVERFLOW, /* struct fanotify_event */
+	FANOTIFY_EVENT_TYPE_FS_ERROR, /* struct fanotify_error_event */
 	__FANOTIFY_EVENT_TYPE_NUM
 };
 
@@ -196,6 +197,16 @@ FANOTIFY_NE(struct fanotify_event *event)
 	return container_of(event, struct fanotify_name_event, fae);
 }
 
+struct fanotify_error_event {
+	struct fanotify_event fae;
+};
+
+static inline struct fanotify_error_event *
+FANOTIFY_EE(struct fanotify_event *event)
+{
+	return container_of(event, struct fanotify_error_event, fae);
+}
+
 static inline __kernel_fsid_t *fanotify_event_fsid(struct fanotify_event *event)
 {
 	if (event->type == FANOTIFY_EVENT_TYPE_FID)

fs/notify/fanotify/fanotify_user.c

@@ -30,6 +30,7 @@
 #define FANOTIFY_DEFAULT_MAX_EVENTS	16384
 #define FANOTIFY_OLD_DEFAULT_MAX_MARKS	8192
 #define FANOTIFY_DEFAULT_MAX_GROUPS	128
+#define FANOTIFY_DEFAULT_FEE_POOL_SIZE	32
 
 /*
  * Legacy fanotify marks limits (8192) is per group and we introduced a tunable
@@ -1054,6 +1055,15 @@ out_dec_ucounts:
 	return ERR_PTR(ret);
 }
 
+static int fanotify_group_init_error_pool(struct fsnotify_group *group)
+{
+	if (mempool_initialized(&group->fanotify_data.error_events_pool))
+		return 0;
+
+	return mempool_init_kmalloc_pool(&group->fanotify_data.error_events_pool,
+					 FANOTIFY_DEFAULT_FEE_POOL_SIZE,
+					 sizeof(struct fanotify_error_event));
+}
 
 static int fanotify_add_mark(struct fsnotify_group *group,
 			     fsnotify_connp_t *connp, unsigned int type,
@@ -1062,6 +1072,7 @@ static int fanotify_add_mark(struct fsnotify_group *group,
 {
 	struct fsnotify_mark *fsn_mark;
 	__u32 added;
+	int ret = 0;
 
 	mutex_lock(&group->mark_mutex);
 	fsn_mark = fsnotify_find_mark(connp, group);
@@ -1072,13 +1083,26 @@ static int fanotify_add_mark(struct fsnotify_group *group,
 			return PTR_ERR(fsn_mark);
 		}
 	}
+
+	/*
+	 * 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)) {
+		ret = fanotify_group_init_error_pool(group);
+		if (ret)
+			goto out;
+	}
+
 	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
 	if (added & ~fsnotify_conn_mask(fsn_mark->connector))
 		fsnotify_recalc_mask(fsn_mark->connector);
+
+out:
 	mutex_unlock(&group->mark_mutex);
 
 	fsnotify_put_mark(fsn_mark);
-	return 0;
+	return ret;
 }
 
 static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,

include/linux/fsnotify_backend.h

@@ -19,6 +19,7 @@
 #include <linux/atomic.h>
 #include <linux/user_namespace.h>
 #include <linux/refcount.h>
+#include <linux/mempool.h>
 
 /*
  * IN_* from inotfy.h lines up EXACTLY with FS_*, this is so we can easily
@@ -246,6 +247,7 @@ struct fsnotify_group {
 			int flags;           /* flags from fanotify_init() */
 			int f_flags; /* event_f_flags from fanotify_init() */
 			struct ucounts *ucounts;
+			mempool_t error_events_pool;
 		} fanotify_data;
 #endif /* CONFIG_FANOTIFY */
 	};