Re: [PATCH 05/10] Make total_forks per-cgroup

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Sun, 2011-10-02 at 23:21 +0400, Glauber Costa wrote:
> This patch counts the total number of forks per-cgroup.
> The information is propagated to the parent, so the total
> number of forks in the system, is the parent cgroup's one.
> 
> To achieve that, total_forks is made per-cpu. There is no
> particular reason to do that, but by doing this, we are
> able to bundle it inside the cpustat structure already
> present.

I think fweisbec is also doing something with forks and cgroups.

> Signed-off-by: Glauber Costa <glommer@parallels.com>
> ---
>  include/linux/kernel_stat.h |    1 +
>  include/linux/sched.h       |    1 +
>  kernel/fork.c               |    7 ++-----
>  kernel/sched.c              |    9 ++++++++-
>  4 files changed, 12 insertions(+), 6 deletions(-)
> 
> diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
> index 71a69a0..9fb0dda 100644
> --- a/include/linux/kernel_stat.h
> +++ b/include/linux/kernel_stat.h
> @@ -29,6 +29,7 @@ enum cpu_usage_stat {
>  	GUEST_NICE,
>  	IDLE_BASE,
>  	IOWAIT_BASE,
> +	TOTAL_FORKS,
>  	NR_STATS,
>  };
>  
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index 64c5ba5..4ba9dde 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -2716,6 +2716,7 @@ struct cgroup;
>  struct cftype;
>  int cpu_cgroup_proc_stat(struct cgroup *cgrp, struct cftype *cft,
>  			 struct seq_file *p);
> +void task_group_new_fork(struct task_struct *p);
>  #endif /* __KERNEL__ */
>  
>  #endif
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 8e6b6f4..ec2b729 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -76,10 +76,6 @@
>  
>  #include <trace/events/sched.h>
>  
> -/*
> - * Protected counters by write_lock_irq(&tasklist_lock)
> - */
> -unsigned long total_forks;	/* Handle normal Linux uptimes. */
>  int nr_threads;			/* The idle threads do not count.. */
>  
>  int max_threads;		/* tunable limit on nr_threads */
> @@ -1372,7 +1368,8 @@ static struct task_struct *copy_process(unsigned long clone_flags,
>  		nr_threads++;
>  	}
>  
> -	total_forks++;
> +	task_group_new_fork(p);
> +
>  	spin_unlock(&current->sighand->siglock);
>  	write_unlock_irq(&tasklist_lock);
>  	proc_fork_connector(p);
> diff --git a/kernel/sched.c b/kernel/sched.c
> index 89d2248..5a8181e 100644
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -691,6 +691,11 @@ static inline void task_group_account_field(struct task_struct *p,
>  }
>  #endif /* CONFIG_CGROUP_SCHED */
>  
> +void task_group_new_fork(struct task_struct *p)
> +{
> +	task_group_account_field(p, 1, TOTAL_FORKS);
> +}
> +
>  static void update_rq_clock_task(struct rq *rq, s64 delta);
>  
>  static void update_rq_clock(struct rq *rq)
> @@ -9161,6 +9166,7 @@ int cpu_cgroup_proc_stat(struct cgroup *cgrp, struct cftype *cft,
>  	u64 guest, guest_nice;
>  	u64 sum = 0;
>  	u64 sum_softirq = 0;
> +	u64 total_forks = 0;
>  	unsigned int per_softirq_sums[NR_SOFTIRQS] = {0};
>  	struct timespec boottime;
>  #ifdef CONFIG_CGROUP_SCHED
> @@ -9201,6 +9207,7 @@ int cpu_cgroup_proc_stat(struct cgroup *cgrp, struct cftype *cft,
>  		steal += kstat->cpustat[STEAL];
>  		guest += kstat->cpustat[GUEST];
>  		guest_nice += kstat->cpustat[GUEST_NICE];
> +		total_forks += kstat->cpustat[TOTAL_FORKS];
>  		sum += kstat_cpu_irqs_sum(i);
>  		sum += arch_irq_stat_cpu(i);
>  
> @@ -9272,7 +9279,7 @@ int cpu_cgroup_proc_stat(struct cgroup *cgrp, struct cftype *cft,
>  	seq_printf(p,
>  		"\nctxt %llu\n"
>  		"btime %lu\n"
> -		"processes %lu\n"
> +		"processes %llu\n"
>  		"procs_running %lu\n"
>  		"procs_blocked %lu\n",
>  		nr_context_switches(),