Hook up group scheduler with control groups

Enable "cgroup" (formerly containers) based fair group scheduling.  This
will let administrator create arbitrary groups of tasks (using "cgroup"
pseudo filesystem) and control their cpu bandwidth usage.

[akpm@linux-foundation.org: fix cpp condition]
Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Signed-off-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

3 files changed, 137 insertions, 0 deletions

diff --git a/include/linux/cgroup_subsys.h b/include/linux/cgroup_subsys.h
index 651ff0869b2d..0b9bfbde8168 100644
--- a/include/linux/cgroup_subsys.h
+++ b/include/linux/cgroup_subsys.h
@@ -30,3 +30,9 @@ SUBSYS(ns)
 #endif
 
 /* */
+
+#ifdef CONFIG_FAIR_CGROUP_SCHED
+SUBSYS(cpu_cgroup)
+#endif
+
+/* */
diff --git a/init/Kconfig b/init/Kconfig
index 0007d1b5e867..541382d539ad 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -334,6 +334,16 @@ config FAIR_USER_SCHED
 	  This option will choose userid as the basis for grouping
 	  tasks, thus providing equal CPU bandwidth to each user.
 
+config FAIR_CGROUP_SCHED
+	bool "Control groups"
+ 	depends on CGROUPS
+ 	help
+	  This option allows you to create arbitrary task groups
+	  using the "cgroup" pseudo filesystem and control
+	  the cpu bandwidth allocated to each such task group.
+	  Refer to Documentation/cgroups.txt for more information
+	  on "cgroup" pseudo filesystem.
+
 endchoice
 
 config SYSFS_DEPRECATED
diff --git a/kernel/sched.c b/kernel/sched.c
index 4071306e1088..afe76ec2e7fe 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -155,10 +155,15 @@ struct rt_prio_array {
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
 
+#include <linux/cgroup.h>
+
 struct cfs_rq;
 
 /* task group related information */
 struct task_group {
+#ifdef CONFIG_FAIR_CGROUP_SCHED
+	struct cgroup_subsys_state css;
+#endif
 	/* schedulable entities of this group on each cpu */
 	struct sched_entity **se;
 	/* runqueue "owned" by this group on each cpu */
@@ -199,6 +204,9 @@ static inline struct task_group *task_group(struct task_struct *p)
 
 #ifdef CONFIG_FAIR_USER_SCHED
 	tg = p->user->tg;
+#elif defined(CONFIG_FAIR_CGROUP_SCHED)
+	tg = container_of(task_subsys_state(p, cpu_cgroup_subsys_id),
+				struct task_group, css);
 #else
 	tg  = &init_task_group;
 #endif
@@ -7091,3 +7099,116 @@ unsigned long sched_group_shares(struct task_group *tg)
 }
 
 #endif	/* CONFIG_FAIR_GROUP_SCHED */
+
+#ifdef CONFIG_FAIR_CGROUP_SCHED
+
+/* return corresponding task_group object of a cgroup */
+static inline struct task_group *cgroup_tg(struct cgroup *cont)
+{
+	return container_of(cgroup_subsys_state(cont, cpu_cgroup_subsys_id),
+					 struct task_group, css);
+}
+
+static struct cgroup_subsys_state *
+cpu_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
+{
+	struct task_group *tg;
+
+	if (!cont->parent) {
+		/* This is early initialization for the top cgroup */
+		init_task_group.css.cgroup = cont;
+		return &init_task_group.css;
+	}
+
+	/* we support only 1-level deep hierarchical scheduler atm */
+	if (cont->parent->parent)
+		return ERR_PTR(-EINVAL);
+
+	tg = sched_create_group();
+	if (IS_ERR(tg))
+		return ERR_PTR(-ENOMEM);
+
+	/* Bind the cgroup to task_group object we just created */
+	tg->css.cgroup = cont;
+
+	return &tg->css;
+}
+
+static void cpu_cgroup_destroy(struct cgroup_subsys *ss,
+					struct cgroup *cont)
+{
+	struct task_group *tg = cgroup_tg(cont);
+
+	sched_destroy_group(tg);
+}
+
+static int cpu_cgroup_can_attach(struct cgroup_subsys *ss,
+			     struct cgroup *cont, struct task_struct *tsk)
+{
+	/* We don't support RT-tasks being in separate groups */
+	if (tsk->sched_class != &fair_sched_class)
+		return -EINVAL;
+
+	return 0;
+}
+
+static void
+cpu_cgroup_attach(struct cgroup_subsys *ss, struct cgroup *cont,
+			struct cgroup *old_cont, struct task_struct *tsk)
+{
+	sched_move_task(tsk);
+}
+
+static ssize_t cpu_shares_write(struct cgroup *cont, struct cftype *cftype,
+				struct file *file, const char __user *userbuf,
+				size_t nbytes, loff_t *ppos)
+{
+	unsigned long shareval;
+	struct task_group *tg = cgroup_tg(cont);
+	char buffer[2*sizeof(unsigned long) + 1];
+	int rc;
+
+	if (nbytes > 2*sizeof(unsigned long))	/* safety check */
+		return -E2BIG;
+
+	if (copy_from_user(buffer, userbuf, nbytes))
+		return -EFAULT;
+
+	buffer[nbytes] = 0;	/* nul-terminate */
+	shareval = simple_strtoul(buffer, NULL, 10);
+
+	rc = sched_group_set_shares(tg, shareval);
+
+	return (rc < 0 ? rc : nbytes);
+}
+
+static u64 cpu_shares_read_uint(struct cgroup *cont, struct cftype *cft)
+{
+	struct task_group *tg = cgroup_tg(cont);
+
+	return (u64) tg->shares;
+}
+
+static struct cftype cpu_shares = {
+	.name = "shares",
+	.read_uint = cpu_shares_read_uint,
+	.write = cpu_shares_write,
+};
+
+static int cpu_cgroup_populate(struct cgroup_subsys *ss, struct cgroup *cont)
+{
+	return cgroup_add_file(cont, ss, &cpu_shares);
+}
+
+struct cgroup_subsys cpu_cgroup_subsys = {
+	.name 	    	= "cpu",
+	.create	    	= cpu_cgroup_create,
+	.destroy    	= cpu_cgroup_destroy,
+	.can_attach 	= cpu_cgroup_can_attach,
+	.attach     	= cpu_cgroup_attach,
+	.populate   	= cpu_cgroup_populate,
+	.subsys_id  	= cpu_cgroup_subsys_id,
+	.early_init	= 1,
+};
+
+#endif	/* CONFIG_FAIR_CGROUP_SCHED */