Message-ID: <20110214225940.GB6230@shutemov.name>
Date:	Tue, 15 Feb 2011 00:59:40 +0200
From:	"Kirill A. Shutemov" <kirill@...temov.name>
To:	Matt Helsley <matthltc@...ibm.com>
Cc:	Paul Menage <menage@...gle.com>, Li Zefan <lizf@...fujitsu.com>,
	containers@...ts.linux-foundation.org,
	jacob.jun.pan@...ux.intel.com,
	Arjan van de Ven <arjan@...ux.intel.com>,
	linux-kernel@...r.kernel.org,
	Andrew Morton <akpm@...ux-foundation.org>,
	linux-api@...r.kernel.org
Subject: Re: [PATCH, v6 3/3] cgroups: introduce timer slack controller

On Mon, Feb 14, 2011 at 05:59:26AM -0800, Matt Helsley wrote:
> On Mon, Feb 14, 2011 at 03:06:27PM +0200, Kirill A. Shutsemov wrote:
> > From: Kirill A. Shutemov <kirill@...temov.name>
> > 
> > Every task_struct has timer_slack_ns value. This value uses to round up
> > poll() and select() timeout values. This feature can be useful in
> > mobile environment where combined wakeups are desired.
> > 
> > cgroup subsys "timer_slack" implement timer slack controller. It
> > provides a way to group tasks by timer slack value and manage the
> > value of group's tasks.
> > 
> > Idea-by: Jacob Pan <jacob.jun.pan@...ux.intel.com>
> > Signed-off-by: Kirill A. Shutemov <kirill@...temov.name>
> > ---
> >  Documentation/cgroups/timer_slack.txt |   93 +++++++++++
> >  include/linux/cgroup_subsys.h         |    6 +
> >  init/Kconfig                          |   10 ++
> >  kernel/Makefile                       |    1 +
> >  kernel/cgroup_timer_slack.c           |  285 +++++++++++++++++++++++++++++++++
> >  5 files changed, 395 insertions(+), 0 deletions(-)
> >  create mode 100644 Documentation/cgroups/timer_slack.txt
> >  create mode 100644 kernel/cgroup_timer_slack.c
> > 
> > diff --git a/Documentation/cgroups/timer_slack.txt b/Documentation/cgroups/timer_slack.txt
> > new file mode 100644
> > index 0000000..e7ec2f3
> > --- /dev/null
> > +++ b/Documentation/cgroups/timer_slack.txt
> > @@ -0,0 +1,93 @@
> > +Timer Slack Controller
> > +=====================
> > +
> > +Overview
> > +--------
> > +
> > +Every task_struct has timer_slack_ns value. This value uses to round up
> > +poll() and select() timeout values. This feature can be useful in
> > +mobile environment where combined wakeups are desired.
> > +
> > +cgroup subsys "timer_slack" implement timer slack controller. It
> > +provides a way to group tasks by timer slack value and manage the
> > +value of group's tasks.
> > +
> > +
> > +User interface
> > +--------------
> > +
> > +To get timer slack controller functionality you need to enable it in
> > +kernel configuration:
> > +
> > +CONFIG_CGROUP_TIMER_SLACK=y
> > +
> > +or if you want to compile it as module:
> > +
> > +CONFIG_CGROUP_TIMER_SLACK=m
> > +
> > +The controller provides three files in cgroup directory:
> > +
> > +# mount -t cgroup -o timer_slack none /sys/fs/cgroup
> > +# ls /sys/fs/cgroup/timer_slack.*
> > +/sys/fs/cgroup/timer_slack.max_slack_ns
> > +/sys/fs/cgroup/timer_slack.min_slack_ns
> > +/sys/fs/cgroup/timer_slack.set_slack_ns
> > +
> > +timer_slack.min_slack_ns and timer_slack.set_slack_ns specify allowed
> > +range of timer slack for tasks in cgroup. By default it unlimited:
> > +
> > +# cat /sys/fs/cgroup/timer_slack.min_slack_ns
> > +0
> > +# cat /sys/fs/cgroup/timer_slack.max_slack_ns
> > +4294967295
> > +
> > +You can specify limits you want:
> > +
> > +# echo 50000 > /sys/fs/cgroup/timer_slack.min_slack_ns
> > +# echo 1000000 > /sys/fs/cgroup/timer_slack.max_slack_ns
> > +# cat /sys/fs/cgroup/timer_slack.{min,max}_slack_ns
> > +50000
> > +1000000
> > +
> > +Timer slack value of all tasks of the cgroup will be adjusted to fit
> > +min-max range.
> > +
> > +If a task will try to call prctl() to change timer slack value out of
> > +the range it get -EPERM.
> > +
> > +You can change timer slack value of all tasks of the cgroup at once:
> > +
> > +# echo 70000 > /sys/fs/cgroup/timer_slack.set_slack_ns
> > +
> > +Timer slack controller supports hierarchical groups. The only rule:
> > +parent's limit range should be wider or equal to child's. Sibling
> > +cgroups can have overlapping min-max range.
> > +
> > +		(root: 50000 - 1000000)
> > +		 /		     \
> > +	(a: 50000 - 50000)	(b: 500000 - 1000000)
> > +				 /		   \
> > +			(c: 500000 - 900000) (d: 700000 - 800000)
> > +
> > +# mkdir /sys/fs/cgroup/a
> > +# echo 50000 > /sys/fs/cgroup/a/timer_slack.max_slack_ns
> > +# cat /sys/fs/cgroup/a/timer_slack.{min,max}_slack_ns
> > +50000
> > +50000
> > +# mkdir /sys/fs/cgroup/b
> > +# echo 500000 > /sys/fs/cgroup/b/timer_slack.min_slack_ns
> > +# cat /sys/fs/cgroup/b/timer_slack.{min,max}_slack_ns
> > +500000
> > +1000000
> > +# mkdir /sys/fs/cgroup/b/c
> > +# echo 900000 > /sys/fs/cgroup/b/c/timer_slack.max_slack_ns
> > +# cat /sys/fs/cgroup/b/c/timer_slack.{min,max}_slack_ns
> > +500000
> > +900000
> > +# mkdir /sys/fs/cgroup/b/d
> > +# echo 700000 > /sys/fs/cgroup/b/d/timer_slack.min_slack_ns
> > +# echo 800000 > /sys/fs/cgroup/b/d/timer_slack.max_slack_ns
> > +# cat /sys/fs/cgroup/b/d/timer_slack.{min,max}_slack_ns
> > +700000
> > +800000
> > +
> > diff --git a/include/linux/cgroup_subsys.h b/include/linux/cgroup_subsys.h
> > index ccefff0..e399228 100644
> > --- a/include/linux/cgroup_subsys.h
> > +++ b/include/linux/cgroup_subsys.h
> > @@ -66,3 +66,9 @@ SUBSYS(blkio)
> >  #endif
> > 
> >  /* */
> > +
> > +#ifdef CONFIG_CGROUP_TIMER_SLACK
> > +SUBSYS(timer_slack)
> > +#endif
> > +
> > +/* */
> > diff --git a/init/Kconfig b/init/Kconfig
> > index be788c0..bb54ae9 100644
> > --- a/init/Kconfig
> > +++ b/init/Kconfig
> > @@ -596,6 +596,16 @@ config CGROUP_FREEZER
> >  	  Provides a way to freeze and unfreeze all tasks in a
> >  	  cgroup.
> > 
> > +config CGROUP_TIMER_SLACK
> > +	tristate "Timer slack cgroup controller"
> > +	help
> > +	  Provides a way of tasks grouping by timer slack value.
> > +	  Every timer slack cgroup has min and max slack value. Task's
> > +	  timer slack value will be adjusted to fit min-max range once
> > +	  the task attached to the cgroup.
> > +	  It's useful in mobile devices where certain background apps
> > +	  are attached to a cgroup and combined wakeups are desired.
> > +
> >  config CGROUP_DEVICE
> >  	bool "Device controller for cgroups"
> >  	help
> > diff --git a/kernel/Makefile b/kernel/Makefile
> > index aa43e63..0e660c5 100644
> > --- a/kernel/Makefile
> > +++ b/kernel/Makefile
> > @@ -61,6 +61,7 @@ obj-$(CONFIG_BACKTRACE_SELF_TEST) += backtracetest.o
> >  obj-$(CONFIG_COMPAT) += compat.o
> >  obj-$(CONFIG_CGROUPS) += cgroup.o
> >  obj-$(CONFIG_CGROUP_FREEZER) += cgroup_freezer.o
> > +obj-$(CONFIG_CGROUP_TIMER_SLACK) += cgroup_timer_slack.o
> >  obj-$(CONFIG_CPUSETS) += cpuset.o
> >  obj-$(CONFIG_CGROUP_NS) += ns_cgroup.o
> >  obj-$(CONFIG_UTS_NS) += utsname.o
> > diff --git a/kernel/cgroup_timer_slack.c b/kernel/cgroup_timer_slack.c
> > new file mode 100644
> > index 0000000..ec63700
> > --- /dev/null
> > +++ b/kernel/cgroup_timer_slack.c
> > @@ -0,0 +1,285 @@
> > +/*
> > + * cgroup_timer_slack.c - control group timer slack subsystem
> > + *
> > + * Copyright Nokia Corparation, 2011
> > + * Author: Kirill A. Shutemov
> > + *
> > + * This program is free software; you can redistribute it and/or modify
> > + * it under the terms of the GNU General Public License as published by
> > + * the Free Software Foundation; either version 2 of the License, or
> > + * (at your option) any later version.
> > + *
> > + * This program is distributed in the hope that it will be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> > + * GNU General Public License for more details.
> > + */
> > +#include <linux/cgroup.h>
> > +#include <linux/init_task.h>
> > +#include <linux/module.h>
> > +#include <linux/slab.h>
> > +#include <linux/rcupdate.h>
> > +
> > +struct cgroup_subsys timer_slack_subsys;
> > +struct timer_slack_cgroup {
> > +	struct cgroup_subsys_state css;
> > +	unsigned long min_slack_ns;
> > +	unsigned long max_slack_ns;
> > +};
> > +
> > +enum {
> > +	TIMER_SLACK_MIN,
> > +	TIMER_SLACK_MAX,
> > +};
> > +
> > +static struct timer_slack_cgroup *cgroup_to_tslack_cgroup(struct cgroup *cgroup)
> > +{
> > +	struct cgroup_subsys_state *css;
> > +
> > +	css = cgroup_subsys_state(cgroup, timer_slack_subsys.subsys_id);
> > +	return container_of(css, struct timer_slack_cgroup, css);
> > +}
> > +
> > +static int is_timer_slack_allowed(struct timer_slack_cgroup *tslack_cgroup,
> > +		unsigned long slack_ns)
> > +{
> > +	if (slack_ns < tslack_cgroup->min_slack_ns ||
> > +			slack_ns > tslack_cgroup->max_slack_ns)
> > +		return false;
> > +	return true;
> > +}
> > +
> > +static int cgroup_timer_slack_check(struct notifier_block *nb,
> > +		unsigned long slack_ns, void *data)
> > +{
> > +	struct cgroup_subsys_state *css;
> > +	struct timer_slack_cgroup *tslack_cgroup;
> > +
> > +	/* XXX: lockdep false positive? */
> > +	rcu_read_lock();
> > +	css = task_subsys_state(current, timer_slack_subsys.subsys_id);
> > +	tslack_cgroup = container_of(css, struct timer_slack_cgroup, css);
> > +	rcu_read_unlock();
> > +
> > +	if (!is_timer_slack_allowed(tslack_cgroup, slack_ns))
> 
> I think this test -- or at least accesses to the cgroup's min/max values
> -- needs to be in the rcu_read_lock() else there is a race between the
> notifier call from the context of the task calling prctl() and the context
> of the task writing to the cgroup's (min|max)_slack_ns files.

I'll rework this code.

> > +		return notifier_from_errno(-EPERM);
> > +	return NOTIFY_OK;
> > +}
> > +
> > +static struct notifier_block cgroup_timer_slack_nb = {
> > +	.notifier_call = cgroup_timer_slack_check,
> > +};
> > +
> > +static struct cgroup_subsys_state *
> > +tslack_cgroup_create(struct cgroup_subsys *subsys, struct cgroup *cgroup)
> > +{
> > +	struct timer_slack_cgroup *tslack_cgroup;
> > +
> > +	tslack_cgroup = kmalloc(sizeof(*tslack_cgroup), GFP_KERNEL);
> > +	if (!tslack_cgroup)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	if (cgroup->parent) {
> > +		struct timer_slack_cgroup *parent;
> > +		parent = cgroup_to_tslack_cgroup(cgroup->parent);
> > +		tslack_cgroup->min_slack_ns = parent->min_slack_ns;
> > +		tslack_cgroup->max_slack_ns = parent->max_slack_ns;
> > +	} else {
> > +		tslack_cgroup->min_slack_ns = 0UL;
> > +		tslack_cgroup->max_slack_ns = ULONG_MAX;
> > +	}
> > +
> > +	return &tslack_cgroup->css;
> > +}
> > +
> > +static void tslack_cgroup_destroy(struct cgroup_subsys *subsys,
> > +		struct cgroup *cgroup)
> > +{
> > +	kfree(cgroup_to_tslack_cgroup(cgroup));
> > +}
> > +
> > +/*
> > + * Adjust ->timer_slack_ns and ->default_max_slack_ns of the task to fit
> > + * limits of the cgroup.
> > + */
> > +static void tslack_adjust_task(struct timer_slack_cgroup *tslack_cgroup,
> > +		struct task_struct *tsk)
> > +{
> > +	if (tslack_cgroup->min_slack_ns > tsk->timer_slack_ns)
> > +		tsk->timer_slack_ns = tslack_cgroup->min_slack_ns;
> > +	else if (tslack_cgroup->max_slack_ns < tsk->timer_slack_ns)
> > +		tsk->timer_slack_ns = tslack_cgroup->max_slack_ns;
> 
> It occured to me there's a macro for this in include/linux/kernel.h:
> 
> tsk->timer_slack_ns = clamp(tsk->timer_slack_ns,
> 				tslack_cgroup->min_slack_ns,
> 				tslack_cgroup->max_slack_ns);

Nice. Thank you.

> > +
> > +	if (tslack_cgroup->min_slack_ns > tsk->default_timer_slack_ns)
> > +		tsk->default_timer_slack_ns = tslack_cgroup->min_slack_ns;
> > +	else if (tslack_cgroup->max_slack_ns < tsk->default_timer_slack_ns)
> > +		tsk->default_timer_slack_ns = tslack_cgroup->max_slack_ns;
> 
> tsk->default_timer_slack_ns = clamp(tsk->default_timer_slack_ns,
> 					tslack_cgroup->min_slack_ns,
> 					tslack_cgroup->max_slack_ns);
> 
> > +}
> > +
> > +static void tslack_cgroup_attach(struct cgroup_subsys *subsys,
> > +		struct cgroup *cgroup, struct cgroup *prev,
> > +		struct task_struct *tsk, bool threadgroup)
> > +{
> > +	tslack_adjust_task(cgroup_to_tslack_cgroup(cgroup), tsk);
> > +}
> > +
> > +static int tslack_write_set_slack_ns(struct cgroup *cgroup, struct cftype *cft,
> > +		u64 val)
> > +{
> > +	struct timer_slack_cgroup *tslack_cgroup;
> > +	struct cgroup_iter it;
> > +	struct task_struct *task;
> > +
> > +	tslack_cgroup = cgroup_to_tslack_cgroup(cgroup);
> > +	if (!is_timer_slack_allowed(cgroup_to_tslack_cgroup(cgroup), val))
> > +		return -EPERM;
> > +
> > +	/* Change timer slack value for all tasks in the cgroup */
> > +	cgroup_iter_start(cgroup, &it);
> > +	while ((task = cgroup_iter_next(cgroup, &it)))
> > +		task->timer_slack_ns = val;
> > +	cgroup_iter_end(cgroup, &it);
> > +
> > +	return 0;
> > +}
> > +
> > +static u64 tslack_read_range(struct cgroup *cgroup, struct cftype *cft)
> > +{
> > +	struct timer_slack_cgroup *tslack_cgroup;
> > +
> > +	tslack_cgroup = cgroup_to_tslack_cgroup(cgroup);
> > +	switch (cft->private) {
> > +	case TIMER_SLACK_MIN:
> > +		return tslack_cgroup->min_slack_ns;
> > +	case TIMER_SLACK_MAX:
> > +		return tslack_cgroup->max_slack_ns;
> > +	default:
> > +		BUG();
> > +	}
> > +}
> > +
> > +static int validate_change(struct cgroup *cgroup, u64 val, int type)
> > +{
> > +	struct timer_slack_cgroup *tslack_cgroup, *child;
> > +	struct cgroup *cur;
> > +
> > +	BUG_ON(type != TIMER_SLACK_MIN && type != TIMER_SLACK_MAX);
> > +
> > +	if (val > ULONG_MAX)
> > +		return -EINVAL;
> > +
> > +	if (cgroup->parent) {
> > +		struct timer_slack_cgroup *parent;
> > +		parent = cgroup_to_tslack_cgroup(cgroup->parent);
> > +		if (!is_timer_slack_allowed(parent, val))
> > +			return -EPERM;
> > +	}
> > +
> > +	tslack_cgroup = cgroup_to_tslack_cgroup(cgroup);
> > +	if (type == TIMER_SLACK_MIN && val > tslack_cgroup->max_slack_ns)
> > +		return -EINVAL;
> > +	if (type == TIMER_SLACK_MAX && val < tslack_cgroup->min_slack_ns)
> > +		return -EINVAL;
> > +
> > +	list_for_each_entry(cur, &cgroup->children, sibling) {
> > +		child = cgroup_to_tslack_cgroup(cur);
> > +		if (type == TIMER_SLACK_MIN && val > child->min_slack_ns)
> > +			return -EBUSY;
> > +		if (type == TIMER_SLACK_MAX && val < child->max_slack_ns)
> > +			return -EBUSY;
> > +	}
> 
> This doesn't look right. Child cgroups should not constrain their
> parents. Instead you should allow the change and propagate the
> constraint to the children.

See discussion with Thomas.
 
> One thing that might make reviewing such changes to these patches easier
> would be to split out the arbitrary-depth hierarchy support into a
> follow-on patch. You can do it like blkio does in the _create() function:
> 
>         /* Currently we do not support hierarchy deeper than two level */
>         if (parent != cgroup->top_cgroup)
>                 return ERR_PTR(-EPERM);
> 
> > +
> > +	return 0;
> > +}
> > +
> > +static int tslack_write_range(struct cgroup *cgroup, struct cftype *cft,
> > +		u64 val)
> > +{
> > +	struct timer_slack_cgroup *tslack_cgroup;
> > +	struct cgroup_iter it;
> > +	struct task_struct *task;
> > +	int err;
> > +
> > +	err = validate_change(cgroup, val, cft->private);
> > +	if (err)
> > +		return err;
> > +
> > +	tslack_cgroup = cgroup_to_tslack_cgroup(cgroup);
> > +	if (cft->private == TIMER_SLACK_MIN)
> > +		tslack_cgroup->min_slack_ns = val;
> > +	else
> > +		tslack_cgroup->max_slack_ns = val;
> > +
> > +	/*
> > +	 * Adjust timer slack value for all tasks in the cgroup to fit
> > +	 * min-max range.
> > +	 */
> > +	cgroup_iter_start(cgroup, &it);
> > +	while ((task = cgroup_iter_next(cgroup, &it)))
> > +		tslack_adjust_task(tslack_cgroup, task);
> > +	cgroup_iter_end(cgroup, &it);
> 
> So, you should "adjust" child cgroups too rather than return -EBUSY from
> validate_change().

Ditto.

> > +
> > +	return 0;
> > +}
> > +
> > +static struct cftype files[] = {
> > +	{
> > +		.name = "set_slack_ns",
> > +		.write_u64 = tslack_write_set_slack_ns,
> > +	},
> > +	{
> > +		.name = "min_slack_ns",
> > +		.private = TIMER_SLACK_MIN,
> > +		.read_u64 = tslack_read_range,
> > +		.write_u64 = tslack_write_range,
> > +	},
> > +	{
> > +		.name = "max_slack_ns",
> > +		.private = TIMER_SLACK_MAX,
> > +		.read_u64 = tslack_read_range,
> > +		.write_u64 = tslack_write_range,
> > +	},
> 
> I didn't get a reply on how a max_slack_ns is useful. It seems
> prudent to add as little interface as possible and only when
> we clearly see the utility of it.

For example, you can create two groups (excluding root cgroup):

default - timer slack range 50000-50000
relaxed - timer slack range 500000-unlimited.

Now you can drag tasks between these group without need to reset value on
relaxed -> default transition.

-- 
 Kirill A. Shutemov
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists