the reason is to avoid a delay caused by the synchronize_rcu() call in kern_umount() when the mqueue mount is freed. the code: #define _GNU_SOURCE #include <sched.h> #include <error.h> #include <errno.h> #include <stdlib.h> int main() { int i; for (i = 0; i < 1000; i++) if (unshare(CLONE_NEWIPC) < 0) error(EXIT_FAILURE, errno, "unshare"); } goes from Command being timed: "./ipc-namespace" User time (seconds): 0.00 System time (seconds): 0.06 Percent of CPU this job got: 0% Elapsed (wall clock) time (h:mm:ss or m:ss): 0:08.05 to Command being timed: "./ipc-namespace" User time (seconds): 0.00 System time (seconds): 0.02 Percent of CPU this job got: 96% Elapsed (wall clock) time (h:mm:ss or m:ss): 0:00.03
Unified · Split
| @@ -68,6 +68,8 @@ struct ipc_namespace { · 67 unmodified lines | |||
| 68 | struct user_namespace *user_ns; | 68 | struct user_namespace *user_ns; |
| 69 | struct ucounts *ucounts; | 69 | struct ucounts *ucounts; |
| 70 | 70 | ||
| 71 | struct llist_node mnt_llist; | ||
| 72 | |||
| 71 | struct ns_common ns; | 73 | struct ns_common ns; |
| 72 | } __randomize_layout; | 74 | } __randomize_layout; |
| 73 | 75 | ||
| @@ -117,6 +117,10 @@ void free_ipcs(struct ipc_namespace *ns, struct ipc_ids *ids, · 116 unmodified lines | |||
| 117 | 117 | ||
| 118 | static void free_ipc_ns(struct ipc_namespace *ns) | 118 | static void free_ipc_ns(struct ipc_namespace *ns) |
| 119 | { | 119 | { |
| 120 | /* mq_put_mnt() waits for a grace period as kern_unmount() | ||
| 121 | * uses synchronize_rcu(). | ||
| 122 | */ | ||
| 123 | mq_put_mnt(ns); | ||
| 120 | sem_exit_ns(ns); | 124 | sem_exit_ns(ns); |
| 121 | msg_exit_ns(ns); | 125 | msg_exit_ns(ns); |
| 122 | shm_exit_ns(ns); | 126 | shm_exit_ns(ns); |
| @@ -127,6 +131,21 @@ static void free_ipc_ns(struct ipc_namespace *ns) · 4 unmodified lines | |||
| 127 | kfree(ns); | 131 | kfree(ns); |
| 128 | } | 132 | } |
| 129 | 133 | ||
| 134 | static LLIST_HEAD(free_ipc_list); | ||
| 135 | static void free_ipc(struct work_struct *unused) | ||
| 136 | { | ||
| 137 | struct llist_node *node = llist_del_all(&free_ipc_list); | ||
| 138 | struct ipc_namespace *n, *t; | ||
| 139 | |||
| 140 | llist_for_each_entry_safe(n, t, node, mnt_llist) | ||
| 141 | free_ipc_ns(n); | ||
| 142 | } | ||
| 143 | |||
| 144 | /* | ||
| 145 | * The work queue is used to avoid the cost of synchronize_rcu in kern_unmount. | ||
| 146 | */ | ||
| 147 | static DECLARE_WORK(free_ipc_work, free_ipc); | ||
| 148 | |||
| 130 | /* | 149 | /* |
| 131 | * put_ipc_ns - drop a reference to an ipc namespace. | 150 | * put_ipc_ns - drop a reference to an ipc namespace. |
| 132 | * @ns: the namespace to put | 151 | * @ns: the namespace to put |
| @@ -148,8 +167,9 @@ void put_ipc_ns(struct ipc_namespace *ns) · 15 unmodified lines | |||
| 148 | if (refcount_dec_and_lock(&ns->count, &mq_lock)) { | 167 | if (refcount_dec_and_lock(&ns->count, &mq_lock)) { |
| 149 | mq_clear_sbinfo(ns); | 168 | mq_clear_sbinfo(ns); |
| 150 | spin_unlock(&mq_lock); | 169 | spin_unlock(&mq_lock); |
| 151 | mq_put_mnt(ns); | 170 | |
| 152 | free_ipc_ns(ns); | 171 | if (llist_add(&ns->mnt_llist, &free_ipc_list)) |
| 172 | schedule_work(&free_ipc_work); | ||
| 153 | } | 173 | } |
| 154 | } | 174 | } |
| 155 | 175 | ||