Evolution-X-Devices/kernel_google_msm-4.9
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

cgroup/cpuset: remove circular dependency deadlock

Browse Source 
Remove circular dependency deadlock in a scenario where hotplug of CPU is
being done while there is updation in cgroup and cpuset triggered from
userspace.

Process A => kthreadd => Process B => Process C => Process A

Process A
cpu_subsys_offline();
   cpu_down();
     _cpu_down();
       mutex_lock(&cpuhotplug.lock); //held
       __cpu_notify();
         workqueue_cpu_down_callback();
           queue_work_on(system_highpri_wq);
             __queue_work();
               insert_work();
                 wake_up_worker(); //pool->nr_running = 0
           flush_work();
           wait_for_completion();

worker_thread();
  need_more_worker(); // returns true
  manage_workers();
    maybe_create_worker();
	  create_worker();
	    kthread_create_on_node();
		  wake_up_process(kthreadd_task);

kthreadd
kthreadd();
  kernel_thread();
    do_fork();
      copy_process();
        percpu_down_read(&cgroup_threadgroup_rwsem);
          __rwsem_down_read_failed_common(); //waiting

Process B
kernfs_fop_write();
  cgroup_file_write();
    cgroup_tasks_write();
      percpu_down_write(&cgroup_threadgroup_rwsem); //held
      cgroup_attach_task();
        cgroup_migrate();
          cgroup_taskset_migrate();
            cpuset_can_attach();
              mutex_lock(&cpuset_mutex); //waiting

Process C
kernfs_fop_write();
  cgroup_file_write();
    cpuset_write_resmask();
      mutex_lock(&cpuset_mutex); //held
      update_cpumask();
        update_cpumasks_hier();
          rebuild_sched_domains_locked();
            get_online_cpus();
              mutex_lock(&cpuhotplug.lock); //waiting

Eliminate this dependecy by reordering locking of cpuset_mutex
and cpuhotplug.lock.

Change-Id: Ifd76373d717c53b531623a3be76b7d32e0d959fd
Signed-off-by: Prateek Sood <prsood@codeaurora.org>
This commit is contained in:
Prateek Sood
2017-09-08 13:10:55 +05:30
committed by Gerrit - the friendly Code Review server
parent 1a60f3f6c3
commit e9e7b7541a
3 changed files with 26 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/linux/cpu.h
View File

@@ -194,6 +194,7 @@ extern struct bus_type cpu_subsys;
extern void cpu_hotplug_begin(void);
extern void cpu_hotplug_done(void);
extern void get_online_cpus(void);
extern void cpu_hotplug_mutex_held(void);
extern void put_online_cpus(void);
extern void cpu_hotplug_disable(void);
extern void cpu_hotplug_enable(void);
@@ -216,6 +217,7 @@ static inline void cpu_hotplug_done(void) {}
#define cpu_hotplug_enable() do { } while (0)
#define hotcpu_notifier(fn, pri) do { (void)(fn); } while (0)
#define __hotcpu_notifier(fn, pri) do { (void)(fn); } while (0)
#define cpu_hotplug_mutex_held() do { } while (0)
/* These aren't inline functions due to a GCC bug. */
#define register_hotcpu_notifier(nb) ({ (void)(nb); 0; })
#define __register_hotcpu_notifier(nb) ({ (void)(nb); 0; })

5
kernel/cpu.c
View File

@@ -239,6 +239,11 @@ static struct {
#define cpuhp_lock_acquire() lock_map_acquire(&cpu_hotplug.dep_map)
#define cpuhp_lock_release() lock_map_release(&cpu_hotplug.dep_map)
void cpu_hotplug_mutex_held(void)
{
lockdep_assert_held(&cpu_hotplug.lock);
}
EXPORT_SYMBOL(cpu_hotplug_mutex_held);
void get_online_cpus(void)
{

33
kernel/cpuset.c
View File

@@ -808,16 +808,15 @@ done:
* 'cpus' is removed, then call this routine to rebuild the
* scheduler's dynamic sched domains.
*
* Call with cpuset_mutex held. Takes get_online_cpus().
*/
static void rebuild_sched_domains_locked(void)
static void rebuild_sched_domains_unlocked(void)
{
struct sched_domain_attr *attr;
cpumask_var_t *doms;
int ndoms;
cpu_hotplug_mutex_held();
lockdep_assert_held(&cpuset_mutex);
get_online_cpus();
/*
* We have raced with CPU hotplug. Don't do anything to avoid
@@ -825,27 +824,27 @@ static void rebuild_sched_domains_locked(void)
* Anyways, hotplug work item will rebuild sched domains.
*/
if (!cpumask_equal(top_cpuset.effective_cpus, cpu_active_mask))
goto out;
return;
/* Generate domain masks and attrs */
ndoms = generate_sched_domains(&doms, &attr);
/* Have scheduler rebuild the domains */
partition_sched_domains(ndoms, doms, attr);
out:
put_online_cpus();
}
#else /* !CONFIG_SMP */
static void rebuild_sched_domains_locked(void)
static void rebuild_sched_domains_unlocked(void)
{
}
#endif /* CONFIG_SMP */
void rebuild_sched_domains(void)
{
get_online_cpus();
mutex_lock(&cpuset_mutex);
rebuild_sched_domains_locked();
rebuild_sched_domains_unlocked();
mutex_unlock(&cpuset_mutex);
put_online_cpus();
}
/**
@@ -877,7 +876,6 @@ static void update_tasks_cpumask(struct cpuset *cs)
*
* On legacy hierachy, effective_cpus will be the same with cpu_allowed.
*
* Called with cpuset_mutex held
*/
static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus)
{
@@ -932,7 +930,7 @@ static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus)
rcu_read_unlock();
if (need_rebuild_sched_domains)
rebuild_sched_domains_locked();
rebuild_sched_domains_unlocked();
}
/**
@@ -1282,7 +1280,7 @@ static int update_relax_domain_level(struct cpuset *cs, s64 val)
cs->relax_domain_level = val;
if (!cpumask_empty(cs->cpus_allowed) &&
is_sched_load_balance(cs))
rebuild_sched_domains_locked();
rebuild_sched_domains_unlocked();
}
return 0;
@@ -1313,7 +1311,6 @@ static void update_tasks_flags(struct cpuset *cs)
* cs: the cpuset to update
* turning_on: whether the flag is being set or cleared
*
* Call with cpuset_mutex held.
*/
static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs,
@@ -1348,7 +1345,7 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs,
spin_unlock_irq(&callback_lock);
if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed)
rebuild_sched_domains_locked();
rebuild_sched_domains_unlocked();
if (spread_flag_changed)
update_tasks_flags(cs);
@@ -1616,6 +1613,7 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft,
cpuset_filetype_t type = cft->private;
int retval = 0;
get_online_cpus();
mutex_lock(&cpuset_mutex);
if (!is_cpuset_online(cs)) {
retval = -ENODEV;
@@ -1653,6 +1651,7 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft,
}
out_unlock:
mutex_unlock(&cpuset_mutex);
put_online_cpus();
return retval;
}
@@ -1663,6 +1662,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft,
cpuset_filetype_t type = cft->private;
int retval = -ENODEV;
get_online_cpus();
mutex_lock(&cpuset_mutex);
if (!is_cpuset_online(cs))
goto out_unlock;
@@ -1677,6 +1677,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft,
}
out_unlock:
mutex_unlock(&cpuset_mutex);
put_online_cpus();
return retval;
}
@@ -1715,6 +1716,7 @@ static ssize_t cpuset_write_resmask(struct kernfs_open_file *of,
kernfs_break_active_protection(of->kn);
flush_work(&cpuset_hotplug_work);
get_online_cpus();
mutex_lock(&cpuset_mutex);
if (!is_cpuset_online(cs))
goto out_unlock;
@@ -1740,6 +1742,7 @@ static ssize_t cpuset_write_resmask(struct kernfs_open_file *of,
free_trial_cpuset(trialcs);
out_unlock:
mutex_unlock(&cpuset_mutex);
put_online_cpus();
kernfs_unbreak_active_protection(of->kn);
css_put(&cs->css);
flush_workqueue(cpuset_migrate_mm_wq);
@@ -2046,13 +2049,14 @@ out_unlock:
/*
* If the cpuset being removed has its flag 'sched_load_balance'
* enabled, then simulate turning sched_load_balance off, which
* will call rebuild_sched_domains_locked().
* will call rebuild_sched_domains_unlocked().
*/
static void cpuset_css_offline(struct cgroup_subsys_state *css)
{
struct cpuset *cs = css_cs(css);
get_online_cpus();
mutex_lock(&cpuset_mutex);
if (is_sched_load_balance(cs))
@@ -2062,6 +2066,7 @@ static void cpuset_css_offline(struct cgroup_subsys_state *css)
clear_bit(CS_ONLINE, &cs->flags);
mutex_unlock(&cpuset_mutex);
put_online_cpus();
}
static void cpuset_css_free(struct cgroup_subsys_state *css)