Commit dd206bec authored by Alexey Dobriyan's avatar Alexey Dobriyan Committed by Eric W. Biederman

pidns: simpler allocation of pid_* caches

Those pid_* caches are created on demand when a process advances to the new
level of pid namespace. Which means pointers are stable, write only and
thus can be packed into an array instead of spreading them over and using
lists(!) to find them.

Both first and subsequent clone/unshare(CLONE_NEWPID) become faster.
Signed-off-by: default avatarAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: default avatarEric W. Biederman <ebiederm@xmission.com>
parent 91ab883e
...@@ -23,55 +23,39 @@ ...@@ -23,55 +23,39 @@
#include <linux/sched/signal.h> #include <linux/sched/signal.h>
#include <linux/idr.h> #include <linux/idr.h>
struct pid_cache {
int nr_ids;
char name[16];
struct kmem_cache *cachep;
struct list_head list;
};
static LIST_HEAD(pid_caches_lh);
static DEFINE_MUTEX(pid_caches_mutex); static DEFINE_MUTEX(pid_caches_mutex);
static struct kmem_cache *pid_ns_cachep; static struct kmem_cache *pid_ns_cachep;
/* MAX_PID_NS_LEVEL is needed for limiting size of 'struct pid' */
#define MAX_PID_NS_LEVEL 32
/* Write once array, filled from the beginning. */
static struct kmem_cache *pid_cache[MAX_PID_NS_LEVEL];
/* /*
* creates the kmem cache to allocate pids from. * creates the kmem cache to allocate pids from.
* @nr_ids: the number of numerical ids this pid will have to carry * @level: pid namespace level
*/ */
static struct kmem_cache *create_pid_cachep(int nr_ids) static struct kmem_cache *create_pid_cachep(unsigned int level)
{ {
struct pid_cache *pcache; /* Level 0 is init_pid_ns.pid_cachep */
struct kmem_cache *cachep; struct kmem_cache **pkc = &pid_cache[level - 1];
struct kmem_cache *kc;
char name[4 + 10 + 1];
unsigned int len;
kc = READ_ONCE(*pkc);
if (kc)
return kc;
snprintf(name, sizeof(name), "pid_%u", level + 1);
len = sizeof(struct pid) + level * sizeof(struct upid);
mutex_lock(&pid_caches_mutex); mutex_lock(&pid_caches_mutex);
list_for_each_entry(pcache, &pid_caches_lh, list) /* Name collision forces to do allocation under mutex. */
if (pcache->nr_ids == nr_ids) if (!*pkc)
goto out; *pkc = kmem_cache_create(name, len, 0, SLAB_HWCACHE_ALIGN, 0);
pcache = kmalloc(sizeof(struct pid_cache), GFP_KERNEL);
if (pcache == NULL)
goto err_alloc;
snprintf(pcache->name, sizeof(pcache->name), "pid_%d", nr_ids);
cachep = kmem_cache_create(pcache->name,
sizeof(struct pid) + (nr_ids - 1) * sizeof(struct upid),
0, SLAB_HWCACHE_ALIGN, NULL);
if (cachep == NULL)
goto err_cachep;
pcache->nr_ids = nr_ids;
pcache->cachep = cachep;
list_add(&pcache->list, &pid_caches_lh);
out:
mutex_unlock(&pid_caches_mutex);
return pcache->cachep;
err_cachep:
kfree(pcache);
err_alloc:
mutex_unlock(&pid_caches_mutex); mutex_unlock(&pid_caches_mutex);
return NULL; /* current can fail, but someone else can succeed. */
return READ_ONCE(*pkc);
} }
static void proc_cleanup_work(struct work_struct *work) static void proc_cleanup_work(struct work_struct *work)
...@@ -80,9 +64,6 @@ static void proc_cleanup_work(struct work_struct *work) ...@@ -80,9 +64,6 @@ static void proc_cleanup_work(struct work_struct *work)
pid_ns_release_proc(ns); pid_ns_release_proc(ns);
} }
/* MAX_PID_NS_LEVEL is needed for limiting size of 'struct pid' */
#define MAX_PID_NS_LEVEL 32
static struct ucounts *inc_pid_namespaces(struct user_namespace *ns) static struct ucounts *inc_pid_namespaces(struct user_namespace *ns)
{ {
return inc_ucount(ns, current_euid(), UCOUNT_PID_NAMESPACES); return inc_ucount(ns, current_euid(), UCOUNT_PID_NAMESPACES);
...@@ -119,7 +100,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns ...@@ -119,7 +100,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns
idr_init(&ns->idr); idr_init(&ns->idr);
ns->pid_cachep = create_pid_cachep(level + 1); ns->pid_cachep = create_pid_cachep(level);
if (ns->pid_cachep == NULL) if (ns->pid_cachep == NULL)
goto out_free_idr; goto out_free_idr;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment