linux/kernel/cgroup/cgroup-v1.c
Prateek Sood 116d2f7496 cgroup: Fix deadlock in cpu hotplug path
Deadlock during cgroup migration from cpu hotplug path when a task T is
being moved from source to destination cgroup.

kworker/0:0
cpuset_hotplug_workfn()
   cpuset_hotplug_update_tasks()
      hotplug_update_tasks_legacy()
        remove_tasks_in_empty_cpuset()
          cgroup_transfer_tasks() // stuck in iterator loop
            cgroup_migrate()
              cgroup_migrate_add_task()

In cgroup_migrate_add_task() it checks for PF_EXITING flag of task T.
Task T will not migrate to destination cgroup. css_task_iter_start()
will keep pointing to task T in loop waiting for task T cg_list node
to be removed.

Task T
do_exit()
  exit_signals() // sets PF_EXITING
  exit_task_namespaces()
    switch_task_namespaces()
      free_nsproxy()
        put_mnt_ns()
          drop_collected_mounts()
            namespace_unlock()
              synchronize_rcu()
                _synchronize_rcu_expedited()
                  schedule_work() // on cpu0 low priority worker pool
                  wait_event() // waiting for work item to execute

Task T inserted a work item in the worklist of cpu0 low priority
worker pool. It is waiting for expedited grace period work item
to execute. This work item will only be executed once kworker/0:0
complete execution of cpuset_hotplug_workfn().

kworker/0:0 ==> Task T ==>kworker/0:0

In case of PF_EXITING task being migrated from source to destination
cgroup, migrate next available task in source cgroup.

Signed-off-by: Prateek Sood <prsood@codeaurora.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
2017-12-19 05:38:47 -08:00

1321 lines
34 KiB
C

#include "cgroup-internal.h"
#include <linux/ctype.h>
#include <linux/kmod.h>
#include <linux/sort.h>
#include <linux/delay.h>
#include <linux/mm.h>
#include <linux/sched/signal.h>
#include <linux/sched/task.h>
#include <linux/magic.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/delayacct.h>
#include <linux/pid_namespace.h>
#include <linux/cgroupstats.h>
#include <trace/events/cgroup.h>
/*
* pidlists linger the following amount before being destroyed. The goal
* is avoiding frequent destruction in the middle of consecutive read calls
* Expiring in the middle is a performance problem not a correctness one.
* 1 sec should be enough.
*/
#define CGROUP_PIDLIST_DESTROY_DELAY HZ
/* Controllers blocked by the commandline in v1 */
static u16 cgroup_no_v1_mask;
/*
* pidlist destructions need to be flushed on cgroup destruction. Use a
* separate workqueue as flush domain.
*/
static struct workqueue_struct *cgroup_pidlist_destroy_wq;
/*
* Protects cgroup_subsys->release_agent_path. Modifying it also requires
* cgroup_mutex. Reading requires either cgroup_mutex or this spinlock.
*/
static DEFINE_SPINLOCK(release_agent_path_lock);
bool cgroup1_ssid_disabled(int ssid)
{
return cgroup_no_v1_mask & (1 << ssid);
}
/**
* cgroup_attach_task_all - attach task 'tsk' to all cgroups of task 'from'
* @from: attach to all cgroups of a given task
* @tsk: the task to be attached
*/
int cgroup_attach_task_all(struct task_struct *from, struct task_struct *tsk)
{
struct cgroup_root *root;
int retval = 0;
mutex_lock(&cgroup_mutex);
percpu_down_write(&cgroup_threadgroup_rwsem);
for_each_root(root) {
struct cgroup *from_cgrp;
if (root == &cgrp_dfl_root)
continue;
spin_lock_irq(&css_set_lock);
from_cgrp = task_cgroup_from_root(from, root);
spin_unlock_irq(&css_set_lock);
retval = cgroup_attach_task(from_cgrp, tsk, false);
if (retval)
break;
}
percpu_up_write(&cgroup_threadgroup_rwsem);
mutex_unlock(&cgroup_mutex);
return retval;
}
EXPORT_SYMBOL_GPL(cgroup_attach_task_all);
/**
* cgroup_trasnsfer_tasks - move tasks from one cgroup to another
* @to: cgroup to which the tasks will be moved
* @from: cgroup in which the tasks currently reside
*
* Locking rules between cgroup_post_fork() and the migration path
* guarantee that, if a task is forking while being migrated, the new child
* is guaranteed to be either visible in the source cgroup after the
* parent's migration is complete or put into the target cgroup. No task
* can slip out of migration through forking.
*/
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from)
{
DEFINE_CGROUP_MGCTX(mgctx);
struct cgrp_cset_link *link;
struct css_task_iter it;
struct task_struct *task;
int ret;
if (cgroup_on_dfl(to))
return -EINVAL;
ret = cgroup_migrate_vet_dst(to);
if (ret)
return ret;
mutex_lock(&cgroup_mutex);
percpu_down_write(&cgroup_threadgroup_rwsem);
/* all tasks in @from are being moved, all csets are source */
spin_lock_irq(&css_set_lock);
list_for_each_entry(link, &from->cset_links, cset_link)
cgroup_migrate_add_src(link->cset, to, &mgctx);
spin_unlock_irq(&css_set_lock);
ret = cgroup_migrate_prepare_dst(&mgctx);
if (ret)
goto out_err;
/*
* Migrate tasks one-by-one until @from is empty. This fails iff
* ->can_attach() fails.
*/
do {
css_task_iter_start(&from->self, 0, &it);
do {
task = css_task_iter_next(&it);
} while (task && (task->flags & PF_EXITING));
if (task)
get_task_struct(task);
css_task_iter_end(&it);
if (task) {
ret = cgroup_migrate(task, false, &mgctx);
if (!ret)
trace_cgroup_transfer_tasks(to, task, false);
put_task_struct(task);
}
} while (task && !ret);
out_err:
cgroup_migrate_finish(&mgctx);
percpu_up_write(&cgroup_threadgroup_rwsem);
mutex_unlock(&cgroup_mutex);
return ret;
}
/*
* Stuff for reading the 'tasks'/'procs' files.
*
* Reading this file can return large amounts of data if a cgroup has
* *lots* of attached tasks. So it may need several calls to read(),
* but we cannot guarantee that the information we produce is correct
* unless we produce it entirely atomically.
*
*/
/* which pidlist file are we talking about? */
enum cgroup_filetype {
CGROUP_FILE_PROCS,
CGROUP_FILE_TASKS,
};
/*
* A pidlist is a list of pids that virtually represents the contents of one
* of the cgroup files ("procs" or "tasks"). We keep a list of such pidlists,
* a pair (one each for procs, tasks) for each pid namespace that's relevant
* to the cgroup.
*/
struct cgroup_pidlist {
/*
* used to find which pidlist is wanted. doesn't change as long as
* this particular list stays in the list.
*/
struct { enum cgroup_filetype type; struct pid_namespace *ns; } key;
/* array of xids */
pid_t *list;
/* how many elements the above list has */
int length;
/* each of these stored in a list by its cgroup */
struct list_head links;
/* pointer to the cgroup we belong to, for list removal purposes */
struct cgroup *owner;
/* for delayed destruction */
struct delayed_work destroy_dwork;
};
/*
* The following two functions "fix" the issue where there are more pids
* than kmalloc will give memory for; in such cases, we use vmalloc/vfree.
* TODO: replace with a kernel-wide solution to this problem
*/
#define PIDLIST_TOO_LARGE(c) ((c) * sizeof(pid_t) > (PAGE_SIZE * 2))
static void *pidlist_allocate(int count)
{
if (PIDLIST_TOO_LARGE(count))
return vmalloc(count * sizeof(pid_t));
else
return kmalloc(count * sizeof(pid_t), GFP_KERNEL);
}
static void pidlist_free(void *p)
{
kvfree(p);
}
/*
* Used to destroy all pidlists lingering waiting for destroy timer. None
* should be left afterwards.
*/
void cgroup1_pidlist_destroy_all(struct cgroup *cgrp)
{
struct cgroup_pidlist *l, *tmp_l;
mutex_lock(&cgrp->pidlist_mutex);
list_for_each_entry_safe(l, tmp_l, &cgrp->pidlists, links)
mod_delayed_work(cgroup_pidlist_destroy_wq, &l->destroy_dwork, 0);
mutex_unlock(&cgrp->pidlist_mutex);
flush_workqueue(cgroup_pidlist_destroy_wq);
BUG_ON(!list_empty(&cgrp->pidlists));
}
static void cgroup_pidlist_destroy_work_fn(struct work_struct *work)
{
struct delayed_work *dwork = to_delayed_work(work);
struct cgroup_pidlist *l = container_of(dwork, struct cgroup_pidlist,
destroy_dwork);
struct cgroup_pidlist *tofree = NULL;
mutex_lock(&l->owner->pidlist_mutex);
/*
* Destroy iff we didn't get queued again. The state won't change
* as destroy_dwork can only be queued while locked.
*/
if (!delayed_work_pending(dwork)) {
list_del(&l->links);
pidlist_free(l->list);
put_pid_ns(l->key.ns);
tofree = l;
}
mutex_unlock(&l->owner->pidlist_mutex);
kfree(tofree);
}
/*
* pidlist_uniq - given a kmalloc()ed list, strip out all duplicate entries
* Returns the number of unique elements.
*/
static int pidlist_uniq(pid_t *list, int length)
{
int src, dest = 1;
/*
* we presume the 0th element is unique, so i starts at 1. trivial
* edge cases first; no work needs to be done for either
*/
if (length == 0 || length == 1)
return length;
/* src and dest walk down the list; dest counts unique elements */
for (src = 1; src < length; src++) {
/* find next unique element */
while (list[src] == list[src-1]) {
src++;
if (src == length)
goto after;
}
/* dest always points to where the next unique element goes */
list[dest] = list[src];
dest++;
}
after:
return dest;
}
/*
* The two pid files - task and cgroup.procs - guaranteed that the result
* is sorted, which forced this whole pidlist fiasco. As pid order is
* different per namespace, each namespace needs differently sorted list,
* making it impossible to use, for example, single rbtree of member tasks
* sorted by task pointer. As pidlists can be fairly large, allocating one
* per open file is dangerous, so cgroup had to implement shared pool of
* pidlists keyed by cgroup and namespace.
*/
static int cmppid(const void *a, const void *b)
{
return *(pid_t *)a - *(pid_t *)b;
}
static struct cgroup_pidlist *cgroup_pidlist_find(struct cgroup *cgrp,
enum cgroup_filetype type)
{
struct cgroup_pidlist *l;
/* don't need task_nsproxy() if we're looking at ourself */
struct pid_namespace *ns = task_active_pid_ns(current);
lockdep_assert_held(&cgrp->pidlist_mutex);
list_for_each_entry(l, &cgrp->pidlists, links)
if (l->key.type == type && l->key.ns == ns)
return l;
return NULL;
}
/*
* find the appropriate pidlist for our purpose (given procs vs tasks)
* returns with the lock on that pidlist already held, and takes care
* of the use count, or returns NULL with no locks held if we're out of
* memory.
*/
static struct cgroup_pidlist *cgroup_pidlist_find_create(struct cgroup *cgrp,
enum cgroup_filetype type)
{
struct cgroup_pidlist *l;
lockdep_assert_held(&cgrp->pidlist_mutex);
l = cgroup_pidlist_find(cgrp, type);
if (l)
return l;
/* entry not found; create a new one */
l = kzalloc(sizeof(struct cgroup_pidlist), GFP_KERNEL);
if (!l)
return l;
INIT_DELAYED_WORK(&l->destroy_dwork, cgroup_pidlist_destroy_work_fn);
l->key.type = type;
/* don't need task_nsproxy() if we're looking at ourself */
l->key.ns = get_pid_ns(task_active_pid_ns(current));
l->owner = cgrp;
list_add(&l->links, &cgrp->pidlists);
return l;
}
/**
* cgroup_task_count - count the number of tasks in a cgroup.
* @cgrp: the cgroup in question
*/
int cgroup_task_count(const struct cgroup *cgrp)
{
int count = 0;
struct cgrp_cset_link *link;
spin_lock_irq(&css_set_lock);
list_for_each_entry(link, &cgrp->cset_links, cset_link)
count += link->cset->nr_tasks;
spin_unlock_irq(&css_set_lock);
return count;
}
/*
* Load a cgroup's pidarray with either procs' tgids or tasks' pids
*/
static int pidlist_array_load(struct cgroup *cgrp, enum cgroup_filetype type,
struct cgroup_pidlist **lp)
{
pid_t *array;
int length;
int pid, n = 0; /* used for populating the array */
struct css_task_iter it;
struct task_struct *tsk;
struct cgroup_pidlist *l;
lockdep_assert_held(&cgrp->pidlist_mutex);
/*
* If cgroup gets more users after we read count, we won't have
* enough space - tough. This race is indistinguishable to the
* caller from the case that the additional cgroup users didn't
* show up until sometime later on.
*/
length = cgroup_task_count(cgrp);
array = pidlist_allocate(length);
if (!array)
return -ENOMEM;
/* now, populate the array */
css_task_iter_start(&cgrp->self, 0, &it);
while ((tsk = css_task_iter_next(&it))) {
if (unlikely(n == length))
break;
/* get tgid or pid for procs or tasks file respectively */
if (type == CGROUP_FILE_PROCS)
pid = task_tgid_vnr(tsk);
else
pid = task_pid_vnr(tsk);
if (pid > 0) /* make sure to only use valid results */
array[n++] = pid;
}
css_task_iter_end(&it);
length = n;
/* now sort & (if procs) strip out duplicates */
sort(array, length, sizeof(pid_t), cmppid, NULL);
if (type == CGROUP_FILE_PROCS)
length = pidlist_uniq(array, length);
l = cgroup_pidlist_find_create(cgrp, type);
if (!l) {
pidlist_free(array);
return -ENOMEM;
}
/* store array, freeing old if necessary */
pidlist_free(l->list);
l->list = array;
l->length = length;
*lp = l;
return 0;
}
/*
* seq_file methods for the tasks/procs files. The seq_file position is the
* next pid to display; the seq_file iterator is a pointer to the pid
* in the cgroup->l->list array.
*/
static void *cgroup_pidlist_start(struct seq_file *s, loff_t *pos)
{
/*
* Initially we receive a position value that corresponds to
* one more than the last pid shown (or 0 on the first call or
* after a seek to the start). Use a binary-search to find the
* next pid to display, if any
*/
struct kernfs_open_file *of = s->private;
struct cgroup *cgrp = seq_css(s)->cgroup;
struct cgroup_pidlist *l;
enum cgroup_filetype type = seq_cft(s)->private;
int index = 0, pid = *pos;
int *iter, ret;
mutex_lock(&cgrp->pidlist_mutex);
/*
* !NULL @of->priv indicates that this isn't the first start()
* after open. If the matching pidlist is around, we can use that.
* Look for it. Note that @of->priv can't be used directly. It
* could already have been destroyed.
*/
if (of->priv)
of->priv = cgroup_pidlist_find(cgrp, type);
/*
* Either this is the first start() after open or the matching
* pidlist has been destroyed inbetween. Create a new one.
*/
if (!of->priv) {
ret = pidlist_array_load(cgrp, type,
(struct cgroup_pidlist **)&of->priv);
if (ret)
return ERR_PTR(ret);
}
l = of->priv;
if (pid) {
int end = l->length;
while (index < end) {
int mid = (index + end) / 2;
if (l->list[mid] == pid) {
index = mid;
break;
} else if (l->list[mid] <= pid)
index = mid + 1;
else
end = mid;
}
}
/* If we're off the end of the array, we're done */
if (index >= l->length)
return NULL;
/* Update the abstract position to be the actual pid that we found */
iter = l->list + index;
*pos = *iter;
return iter;
}
static void cgroup_pidlist_stop(struct seq_file *s, void *v)
{
struct kernfs_open_file *of = s->private;
struct cgroup_pidlist *l = of->priv;
if (l)
mod_delayed_work(cgroup_pidlist_destroy_wq, &l->destroy_dwork,
CGROUP_PIDLIST_DESTROY_DELAY);
mutex_unlock(&seq_css(s)->cgroup->pidlist_mutex);
}
static void *cgroup_pidlist_next(struct seq_file *s, void *v, loff_t *pos)
{
struct kernfs_open_file *of = s->private;
struct cgroup_pidlist *l = of->priv;
pid_t *p = v;
pid_t *end = l->list + l->length;
/*
* Advance to the next pid in the array. If this goes off the
* end, we're done
*/
p++;
if (p >= end) {
return NULL;
} else {
*pos = *p;
return p;
}
}
static int cgroup_pidlist_show(struct seq_file *s, void *v)
{
seq_printf(s, "%d\n", *(int *)v);
return 0;
}
static ssize_t __cgroup1_procs_write(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off,
bool threadgroup)
{
struct cgroup *cgrp;
struct task_struct *task;
const struct cred *cred, *tcred;
ssize_t ret;
cgrp = cgroup_kn_lock_live(of->kn, false);
if (!cgrp)
return -ENODEV;
task = cgroup_procs_write_start(buf, threadgroup);
ret = PTR_ERR_OR_ZERO(task);
if (ret)
goto out_unlock;
/*
* Even if we're attaching all tasks in the thread group, we only
* need to check permissions on one of them.
*/
cred = current_cred();
tcred = get_task_cred(task);
if (!uid_eq(cred->euid, GLOBAL_ROOT_UID) &&
!uid_eq(cred->euid, tcred->uid) &&
!uid_eq(cred->euid, tcred->suid))
ret = -EACCES;
put_cred(tcred);
if (ret)
goto out_finish;
ret = cgroup_attach_task(cgrp, task, threadgroup);
out_finish:
cgroup_procs_write_finish(task);
out_unlock:
cgroup_kn_unlock(of->kn);
return ret ?: nbytes;
}
static ssize_t cgroup1_procs_write(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
return __cgroup1_procs_write(of, buf, nbytes, off, true);
}
static ssize_t cgroup1_tasks_write(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
return __cgroup1_procs_write(of, buf, nbytes, off, false);
}
static ssize_t cgroup_release_agent_write(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
struct cgroup *cgrp;
BUILD_BUG_ON(sizeof(cgrp->root->release_agent_path) < PATH_MAX);
cgrp = cgroup_kn_lock_live(of->kn, false);
if (!cgrp)
return -ENODEV;
spin_lock(&release_agent_path_lock);
strlcpy(cgrp->root->release_agent_path, strstrip(buf),
sizeof(cgrp->root->release_agent_path));
spin_unlock(&release_agent_path_lock);
cgroup_kn_unlock(of->kn);
return nbytes;
}
static int cgroup_release_agent_show(struct seq_file *seq, void *v)
{
struct cgroup *cgrp = seq_css(seq)->cgroup;
spin_lock(&release_agent_path_lock);
seq_puts(seq, cgrp->root->release_agent_path);
spin_unlock(&release_agent_path_lock);
seq_putc(seq, '\n');
return 0;
}
static int cgroup_sane_behavior_show(struct seq_file *seq, void *v)
{
seq_puts(seq, "0\n");
return 0;
}
static u64 cgroup_read_notify_on_release(struct cgroup_subsys_state *css,
struct cftype *cft)
{
return notify_on_release(css->cgroup);
}
static int cgroup_write_notify_on_release(struct cgroup_subsys_state *css,
struct cftype *cft, u64 val)
{
if (val)
set_bit(CGRP_NOTIFY_ON_RELEASE, &css->cgroup->flags);
else
clear_bit(CGRP_NOTIFY_ON_RELEASE, &css->cgroup->flags);
return 0;
}
static u64 cgroup_clone_children_read(struct cgroup_subsys_state *css,
struct cftype *cft)
{
return test_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags);
}
static int cgroup_clone_children_write(struct cgroup_subsys_state *css,
struct cftype *cft, u64 val)
{
if (val)
set_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags);
else
clear_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags);
return 0;
}
/* cgroup core interface files for the legacy hierarchies */
struct cftype cgroup1_base_files[] = {
{
.name = "cgroup.procs",
.seq_start = cgroup_pidlist_start,
.seq_next = cgroup_pidlist_next,
.seq_stop = cgroup_pidlist_stop,
.seq_show = cgroup_pidlist_show,
.private = CGROUP_FILE_PROCS,
.write = cgroup1_procs_write,
},
{
.name = "cgroup.clone_children",
.read_u64 = cgroup_clone_children_read,
.write_u64 = cgroup_clone_children_write,
},
{
.name = "cgroup.sane_behavior",
.flags = CFTYPE_ONLY_ON_ROOT,
.seq_show = cgroup_sane_behavior_show,
},
{
.name = "tasks",
.seq_start = cgroup_pidlist_start,
.seq_next = cgroup_pidlist_next,
.seq_stop = cgroup_pidlist_stop,
.seq_show = cgroup_pidlist_show,
.private = CGROUP_FILE_TASKS,
.write = cgroup1_tasks_write,
},
{
.name = "notify_on_release",
.read_u64 = cgroup_read_notify_on_release,
.write_u64 = cgroup_write_notify_on_release,
},
{
.name = "release_agent",
.flags = CFTYPE_ONLY_ON_ROOT,
.seq_show = cgroup_release_agent_show,
.write = cgroup_release_agent_write,
.max_write_len = PATH_MAX - 1,
},
{ } /* terminate */
};
/* Display information about each subsystem and each hierarchy */
static int proc_cgroupstats_show(struct seq_file *m, void *v)
{
struct cgroup_subsys *ss;
int i;
seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n");
/*
* ideally we don't want subsystems moving around while we do this.
* cgroup_mutex is also necessary to guarantee an atomic snapshot of
* subsys/hierarchy state.
*/
mutex_lock(&cgroup_mutex);
for_each_subsys(ss, i)
seq_printf(m, "%s\t%d\t%d\t%d\n",
ss->legacy_name, ss->root->hierarchy_id,
atomic_read(&ss->root->nr_cgrps),
cgroup_ssid_enabled(i));
mutex_unlock(&cgroup_mutex);
return 0;
}
static int cgroupstats_open(struct inode *inode, struct file *file)
{
return single_open(file, proc_cgroupstats_show, NULL);
}
const struct file_operations proc_cgroupstats_operations = {
.open = cgroupstats_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
/**
* cgroupstats_build - build and fill cgroupstats
* @stats: cgroupstats to fill information into
* @dentry: A dentry entry belonging to the cgroup for which stats have
* been requested.
*
* Build and fill cgroupstats so that taskstats can export it to user
* space.
*/
int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry)
{
struct kernfs_node *kn = kernfs_node_from_dentry(dentry);
struct cgroup *cgrp;
struct css_task_iter it;
struct task_struct *tsk;
/* it should be kernfs_node belonging to cgroupfs and is a directory */
if (dentry->d_sb->s_type != &cgroup_fs_type || !kn ||
kernfs_type(kn) != KERNFS_DIR)
return -EINVAL;
mutex_lock(&cgroup_mutex);
/*
* We aren't being called from kernfs and there's no guarantee on
* @kn->priv's validity. For this and css_tryget_online_from_dir(),
* @kn->priv is RCU safe. Let's do the RCU dancing.
*/
rcu_read_lock();
cgrp = rcu_dereference(*(void __rcu __force **)&kn->priv);
if (!cgrp || cgroup_is_dead(cgrp)) {
rcu_read_unlock();
mutex_unlock(&cgroup_mutex);
return -ENOENT;
}
rcu_read_unlock();
css_task_iter_start(&cgrp->self, 0, &it);
while ((tsk = css_task_iter_next(&it))) {
switch (tsk->state) {
case TASK_RUNNING:
stats->nr_running++;
break;
case TASK_INTERRUPTIBLE:
stats->nr_sleeping++;
break;
case TASK_UNINTERRUPTIBLE:
stats->nr_uninterruptible++;
break;
case TASK_STOPPED:
stats->nr_stopped++;
break;
default:
if (delayacct_is_task_waiting_on_io(tsk))
stats->nr_io_wait++;
break;
}
}
css_task_iter_end(&it);
mutex_unlock(&cgroup_mutex);
return 0;
}
void cgroup1_check_for_release(struct cgroup *cgrp)
{
if (notify_on_release(cgrp) && !cgroup_is_populated(cgrp) &&
!css_has_online_children(&cgrp->self) && !cgroup_is_dead(cgrp))
schedule_work(&cgrp->release_agent_work);
}
/*
* Notify userspace when a cgroup is released, by running the
* configured release agent with the name of the cgroup (path
* relative to the root of cgroup file system) as the argument.
*
* Most likely, this user command will try to rmdir this cgroup.
*
* This races with the possibility that some other task will be
* attached to this cgroup before it is removed, or that some other
* user task will 'mkdir' a child cgroup of this cgroup. That's ok.
* The presumed 'rmdir' will fail quietly if this cgroup is no longer
* unused, and this cgroup will be reprieved from its death sentence,
* to continue to serve a useful existence. Next time it's released,
* we will get notified again, if it still has 'notify_on_release' set.
*
* The final arg to call_usermodehelper() is UMH_WAIT_EXEC, which
* means only wait until the task is successfully execve()'d. The
* separate release agent task is forked by call_usermodehelper(),
* then control in this thread returns here, without waiting for the
* release agent task. We don't bother to wait because the caller of
* this routine has no use for the exit status of the release agent
* task, so no sense holding our caller up for that.
*/
void cgroup1_release_agent(struct work_struct *work)
{
struct cgroup *cgrp =
container_of(work, struct cgroup, release_agent_work);
char *pathbuf = NULL, *agentbuf = NULL;
char *argv[3], *envp[3];
int ret;
mutex_lock(&cgroup_mutex);
pathbuf = kmalloc(PATH_MAX, GFP_KERNEL);
agentbuf = kstrdup(cgrp->root->release_agent_path, GFP_KERNEL);
if (!pathbuf || !agentbuf)
goto out;
spin_lock_irq(&css_set_lock);
ret = cgroup_path_ns_locked(cgrp, pathbuf, PATH_MAX, &init_cgroup_ns);
spin_unlock_irq(&css_set_lock);
if (ret < 0 || ret >= PATH_MAX)
goto out;
argv[0] = agentbuf;
argv[1] = pathbuf;
argv[2] = NULL;
/* minimal command environment */
envp[0] = "HOME=/";
envp[1] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin";
envp[2] = NULL;
mutex_unlock(&cgroup_mutex);
call_usermodehelper(argv[0], argv, envp, UMH_WAIT_EXEC);
goto out_free;
out:
mutex_unlock(&cgroup_mutex);
out_free:
kfree(agentbuf);
kfree(pathbuf);
}
/*
* cgroup_rename - Only allow simple rename of directories in place.
*/
static int cgroup1_rename(struct kernfs_node *kn, struct kernfs_node *new_parent,
const char *new_name_str)
{
struct cgroup *cgrp = kn->priv;
int ret;
if (kernfs_type(kn) != KERNFS_DIR)
return -ENOTDIR;
if (kn->parent != new_parent)
return -EIO;
/*
* We're gonna grab cgroup_mutex which nests outside kernfs
* active_ref. kernfs_rename() doesn't require active_ref
* protection. Break them before grabbing cgroup_mutex.
*/
kernfs_break_active_protection(new_parent);
kernfs_break_active_protection(kn);
mutex_lock(&cgroup_mutex);
ret = kernfs_rename(kn, new_parent, new_name_str);
if (!ret)
trace_cgroup_rename(cgrp);
mutex_unlock(&cgroup_mutex);
kernfs_unbreak_active_protection(kn);
kernfs_unbreak_active_protection(new_parent);
return ret;
}
static int cgroup1_show_options(struct seq_file *seq, struct kernfs_root *kf_root)
{
struct cgroup_root *root = cgroup_root_from_kf(kf_root);
struct cgroup_subsys *ss;
int ssid;
for_each_subsys(ss, ssid)
if (root->subsys_mask & (1 << ssid))
seq_show_option(seq, ss->legacy_name, NULL);
if (root->flags & CGRP_ROOT_NOPREFIX)
seq_puts(seq, ",noprefix");
if (root->flags & CGRP_ROOT_XATTR)
seq_puts(seq, ",xattr");
if (root->flags & CGRP_ROOT_CPUSET_V2_MODE)
seq_puts(seq, ",cpuset_v2_mode");
spin_lock(&release_agent_path_lock);
if (strlen(root->release_agent_path))
seq_show_option(seq, "release_agent",
root->release_agent_path);
spin_unlock(&release_agent_path_lock);
if (test_bit(CGRP_CPUSET_CLONE_CHILDREN, &root->cgrp.flags))
seq_puts(seq, ",clone_children");
if (strlen(root->name))
seq_show_option(seq, "name", root->name);
return 0;
}
static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts)
{
char *token, *o = data;
bool all_ss = false, one_ss = false;
u16 mask = U16_MAX;
struct cgroup_subsys *ss;
int nr_opts = 0;
int i;
#ifdef CONFIG_CPUSETS
mask = ~((u16)1 << cpuset_cgrp_id);
#endif
memset(opts, 0, sizeof(*opts));
while ((token = strsep(&o, ",")) != NULL) {
nr_opts++;
if (!*token)
return -EINVAL;
if (!strcmp(token, "none")) {
/* Explicitly have no subsystems */
opts->none = true;
continue;
}
if (!strcmp(token, "all")) {
/* Mutually exclusive option 'all' + subsystem name */
if (one_ss)
return -EINVAL;
all_ss = true;
continue;
}
if (!strcmp(token, "noprefix")) {
opts->flags |= CGRP_ROOT_NOPREFIX;
continue;
}
if (!strcmp(token, "clone_children")) {
opts->cpuset_clone_children = true;
continue;
}
if (!strcmp(token, "cpuset_v2_mode")) {
opts->flags |= CGRP_ROOT_CPUSET_V2_MODE;
continue;
}
if (!strcmp(token, "xattr")) {
opts->flags |= CGRP_ROOT_XATTR;
continue;
}
if (!strncmp(token, "release_agent=", 14)) {
/* Specifying two release agents is forbidden */
if (opts->release_agent)
return -EINVAL;
opts->release_agent =
kstrndup(token + 14, PATH_MAX - 1, GFP_KERNEL);
if (!opts->release_agent)
return -ENOMEM;
continue;
}
if (!strncmp(token, "name=", 5)) {
const char *name = token + 5;
/* Can't specify an empty name */
if (!strlen(name))
return -EINVAL;
/* Must match [\w.-]+ */
for (i = 0; i < strlen(name); i++) {
char c = name[i];
if (isalnum(c))
continue;
if ((c == '.') || (c == '-') || (c == '_'))
continue;
return -EINVAL;
}
/* Specifying two names is forbidden */
if (opts->name)
return -EINVAL;
opts->name = kstrndup(name,
MAX_CGROUP_ROOT_NAMELEN - 1,
GFP_KERNEL);
if (!opts->name)
return -ENOMEM;
continue;
}
for_each_subsys(ss, i) {
if (strcmp(token, ss->legacy_name))
continue;
if (!cgroup_ssid_enabled(i))
continue;
if (cgroup1_ssid_disabled(i))
continue;
/* Mutually exclusive option 'all' + subsystem name */
if (all_ss)
return -EINVAL;
opts->subsys_mask |= (1 << i);
one_ss = true;
break;
}
if (i == CGROUP_SUBSYS_COUNT)
return -ENOENT;
}
/*
* If the 'all' option was specified select all the subsystems,
* otherwise if 'none', 'name=' and a subsystem name options were
* not specified, let's default to 'all'
*/
if (all_ss || (!one_ss && !opts->none && !opts->name))
for_each_subsys(ss, i)
if (cgroup_ssid_enabled(i) && !cgroup1_ssid_disabled(i))
opts->subsys_mask |= (1 << i);
/*
* We either have to specify by name or by subsystems. (So all
* empty hierarchies must have a name).
*/
if (!opts->subsys_mask && !opts->name)
return -EINVAL;
/*
* Option noprefix was introduced just for backward compatibility
* with the old cpuset, so we allow noprefix only if mounting just
* the cpuset subsystem.
*/
if ((opts->flags & CGRP_ROOT_NOPREFIX) && (opts->subsys_mask & mask))
return -EINVAL;
/* Can't specify "none" and some subsystems */
if (opts->subsys_mask && opts->none)
return -EINVAL;
return 0;
}
static int cgroup1_remount(struct kernfs_root *kf_root, int *flags, char *data)
{
int ret = 0;
struct cgroup_root *root = cgroup_root_from_kf(kf_root);
struct cgroup_sb_opts opts;
u16 added_mask, removed_mask;
cgroup_lock_and_drain_offline(&cgrp_dfl_root.cgrp);
/* See what subsystems are wanted */
ret = parse_cgroupfs_options(data, &opts);
if (ret)
goto out_unlock;
if (opts.subsys_mask != root->subsys_mask || opts.release_agent)
pr_warn("option changes via remount are deprecated (pid=%d comm=%s)\n",
task_tgid_nr(current), current->comm);
added_mask = opts.subsys_mask & ~root->subsys_mask;
removed_mask = root->subsys_mask & ~opts.subsys_mask;
/* Don't allow flags or name to change at remount */
if ((opts.flags ^ root->flags) ||
(opts.name && strcmp(opts.name, root->name))) {
pr_err("option or name mismatch, new: 0x%x \"%s\", old: 0x%x \"%s\"\n",
opts.flags, opts.name ?: "", root->flags, root->name);
ret = -EINVAL;
goto out_unlock;
}
/* remounting is not allowed for populated hierarchies */
if (!list_empty(&root->cgrp.self.children)) {
ret = -EBUSY;
goto out_unlock;
}
ret = rebind_subsystems(root, added_mask);
if (ret)
goto out_unlock;
WARN_ON(rebind_subsystems(&cgrp_dfl_root, removed_mask));
if (opts.release_agent) {
spin_lock(&release_agent_path_lock);
strcpy(root->release_agent_path, opts.release_agent);
spin_unlock(&release_agent_path_lock);
}
trace_cgroup_remount(root);
out_unlock:
kfree(opts.release_agent);
kfree(opts.name);
mutex_unlock(&cgroup_mutex);
return ret;
}
struct kernfs_syscall_ops cgroup1_kf_syscall_ops = {
.rename = cgroup1_rename,
.show_options = cgroup1_show_options,
.remount_fs = cgroup1_remount,
.mkdir = cgroup_mkdir,
.rmdir = cgroup_rmdir,
.show_path = cgroup_show_path,
};
struct dentry *cgroup1_mount(struct file_system_type *fs_type, int flags,
void *data, unsigned long magic,
struct cgroup_namespace *ns)
{
struct super_block *pinned_sb = NULL;
struct cgroup_sb_opts opts;
struct cgroup_root *root;
struct cgroup_subsys *ss;
struct dentry *dentry;
int i, ret;
bool new_root = false;
cgroup_lock_and_drain_offline(&cgrp_dfl_root.cgrp);
/* First find the desired set of subsystems */
ret = parse_cgroupfs_options(data, &opts);
if (ret)
goto out_unlock;
/*
* Destruction of cgroup root is asynchronous, so subsystems may
* still be dying after the previous unmount. Let's drain the
* dying subsystems. We just need to ensure that the ones
* unmounted previously finish dying and don't care about new ones
* starting. Testing ref liveliness is good enough.
*/
for_each_subsys(ss, i) {
if (!(opts.subsys_mask & (1 << i)) ||
ss->root == &cgrp_dfl_root)
continue;
if (!percpu_ref_tryget_live(&ss->root->cgrp.self.refcnt)) {
mutex_unlock(&cgroup_mutex);
msleep(10);
ret = restart_syscall();
goto out_free;
}
cgroup_put(&ss->root->cgrp);
}
for_each_root(root) {
bool name_match = false;
if (root == &cgrp_dfl_root)
continue;
/*
* If we asked for a name then it must match. Also, if
* name matches but sybsys_mask doesn't, we should fail.
* Remember whether name matched.
*/
if (opts.name) {
if (strcmp(opts.name, root->name))
continue;
name_match = true;
}
/*
* If we asked for subsystems (or explicitly for no
* subsystems) then they must match.
*/
if ((opts.subsys_mask || opts.none) &&
(opts.subsys_mask != root->subsys_mask)) {
if (!name_match)
continue;
ret = -EBUSY;
goto out_unlock;
}
if (root->flags ^ opts.flags)
pr_warn("new mount options do not match the existing superblock, will be ignored\n");
/*
* We want to reuse @root whose lifetime is governed by its
* ->cgrp. Let's check whether @root is alive and keep it
* that way. As cgroup_kill_sb() can happen anytime, we
* want to block it by pinning the sb so that @root doesn't
* get killed before mount is complete.
*
* With the sb pinned, tryget_live can reliably indicate
* whether @root can be reused. If it's being killed,
* drain it. We can use wait_queue for the wait but this
* path is super cold. Let's just sleep a bit and retry.
*/
pinned_sb = kernfs_pin_sb(root->kf_root, NULL);
if (IS_ERR(pinned_sb) ||
!percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
mutex_unlock(&cgroup_mutex);
if (!IS_ERR_OR_NULL(pinned_sb))
deactivate_super(pinned_sb);
msleep(10);
ret = restart_syscall();
goto out_free;
}
ret = 0;
goto out_unlock;
}
/*
* No such thing, create a new one. name= matching without subsys
* specification is allowed for already existing hierarchies but we
* can't create new one without subsys specification.
*/
if (!opts.subsys_mask && !opts.none) {
ret = -EINVAL;
goto out_unlock;
}
/* Hierarchies may only be created in the initial cgroup namespace. */
if (ns != &init_cgroup_ns) {
ret = -EPERM;
goto out_unlock;
}
root = kzalloc(sizeof(*root), GFP_KERNEL);
if (!root) {
ret = -ENOMEM;
goto out_unlock;
}
new_root = true;
init_cgroup_root(root, &opts);
ret = cgroup_setup_root(root, opts.subsys_mask, PERCPU_REF_INIT_DEAD);
if (ret)
cgroup_free_root(root);
out_unlock:
mutex_unlock(&cgroup_mutex);
out_free:
kfree(opts.release_agent);
kfree(opts.name);
if (ret)
return ERR_PTR(ret);
dentry = cgroup_do_mount(&cgroup_fs_type, flags, root,
CGROUP_SUPER_MAGIC, ns);
/*
* There's a race window after we release cgroup_mutex and before
* allocating a superblock. Make sure a concurrent process won't
* be able to re-use the root during this window by delaying the
* initialization of root refcnt.
*/
if (new_root) {
mutex_lock(&cgroup_mutex);
percpu_ref_reinit(&root->cgrp.self.refcnt);
mutex_unlock(&cgroup_mutex);
}
/*
* If @pinned_sb, we're reusing an existing root and holding an
* extra ref on its sb. Mount is complete. Put the extra ref.
*/
if (pinned_sb)
deactivate_super(pinned_sb);
return dentry;
}
static int __init cgroup1_wq_init(void)
{
/*
* Used to destroy pidlists and separate to serve as flush domain.
* Cap @max_active to 1 too.
*/
cgroup_pidlist_destroy_wq = alloc_workqueue("cgroup_pidlist_destroy",
0, 1);
BUG_ON(!cgroup_pidlist_destroy_wq);
return 0;
}
core_initcall(cgroup1_wq_init);
static int __init cgroup_no_v1(char *str)
{
struct cgroup_subsys *ss;
char *token;
int i;
while ((token = strsep(&str, ",")) != NULL) {
if (!*token)
continue;
if (!strcmp(token, "all")) {
cgroup_no_v1_mask = U16_MAX;
break;
}
for_each_subsys(ss, i) {
if (strcmp(token, ss->name) &&
strcmp(token, ss->legacy_name))
continue;
cgroup_no_v1_mask |= 1 << i;
}
}
return 1;
}
__setup("cgroup_no_v1=", cgroup_no_v1);