On Thu, Nov 29, 2007 at 11:30:35AM -0800, Andrew Morton wrote:
> > - Make the accounting scalable on SMP systems (perhaps
> > for 2.6.25)
>
> That sounds like a rather important todo. How bad is it now?
It is indeed an important todo. Right now we take a per-group global
lock on every accounting update (which can be very frequent) and hence
it is pretty bad.
Ingo had expressed the need to reintroduce this patch asap and hence I resent
it w/o attacking the scalability part. I will take a shot at scalability
enhancements tomorrow and send it as a separate patch.
> > +struct cpuacct {
> > + struct cgroup_subsys_state css;
> > + spinlock_t lock;
> > + /* total time used by this class (in nanoseconds) */
> > + u64 time;
> > +};
> > +
> > +struct cgroup_subsys cpuacct_subsys;
>
> This can be made static.
This symbol is needed in kernel/cgroup.c file, where it does this:
static struct cgroup_subsys *subsys[] = {
#include <linux/cgroup_subsys.h>
};
and hence it cant be static. Thanks for the rest of your comments. I
have fixed them in this version below:
Signed-off-by: Srivatsa Vaddagiri <[EMAIL PROTECTED]>
---
include/linux/cgroup_subsys.h | 6 ++
include/linux/cpu_acct.h | 14 +++++
init/Kconfig | 7 ++
kernel/Makefile | 1
kernel/cpu_acct.c | 101 ++++++++++++++++++++++++++++++++++++++++++
kernel/sched.c | 27 -----------
kernel/sched_fair.c | 5 ++
kernel/sched_rt.c | 1
8 files changed, 136 insertions(+), 26 deletions(-)
Index: current/include/linux/cgroup_subsys.h
===================================================================
--- current.orig/include/linux/cgroup_subsys.h
+++ current/include/linux/cgroup_subsys.h
@@ -11,6 +11,12 @@
SUBSYS(cpuset)
#endif
+#ifdef CONFIG_CGROUP_CPUACCT
+SUBSYS(cpuacct)
+#endif
+
+/* */
+
/* */
#ifdef CONFIG_CGROUP_DEBUG
Index: current/include/linux/cpu_acct.h
===================================================================
--- /dev/null
+++ current/include/linux/cpu_acct.h
@@ -0,0 +1,14 @@
+
+#ifndef _LINUX_CPU_ACCT_H
+#define _LINUX_CPU_ACCT_H
+
+#include <linux/cgroup.h>
+#include <asm/cputime.h>
+
+#ifdef CONFIG_CGROUP_CPUACCT
+extern void cpuacct_charge(struct task_struct *tsk, u64 cputime);
+#else
+static inline void cpuacct_charge(struct task_struct *tsk, u64 cputime) {}
+#endif
+
+#endif
Index: current/init/Kconfig
===================================================================
--- current.orig/init/Kconfig
+++ current/init/Kconfig
@@ -354,6 +354,13 @@ config FAIR_CGROUP_SCHED
endchoice
+config CGROUP_CPUACCT
+ bool "Simple CPU accounting cgroup subsystem"
+ depends on CGROUPS
+ help
+ Provides a simple Resource Controller for monitoring the
+ total CPU consumed by the tasks in a cgroup
+
config SYSFS_DEPRECATED
bool "Create deprecated sysfs files"
default y
Index: current/kernel/Makefile
===================================================================
--- current.orig/kernel/Makefile
+++ current/kernel/Makefile
@@ -40,6 +40,7 @@ obj-$(CONFIG_COMPAT) += compat.o
obj-$(CONFIG_CGROUPS) += cgroup.o
obj-$(CONFIG_CGROUP_DEBUG) += cgroup_debug.o
obj-$(CONFIG_CPUSETS) += cpuset.o
+obj-$(CONFIG_CGROUP_CPUACCT) += cpu_acct.o
obj-$(CONFIG_CGROUP_NS) += ns_cgroup.o
obj-$(CONFIG_IKCONFIG) += configs.o
obj-$(CONFIG_STOP_MACHINE) += stop_machine.o
Index: current/kernel/cpu_acct.c
===================================================================
--- /dev/null
+++ current/kernel/cpu_acct.c
@@ -0,0 +1,101 @@
+/*
+ * kernel/cpu_acct.c - CPU accounting cgroup subsystem
+ *
+ * Copyright (C) Google Inc, 2006
+ *
+ * Developed by Paul Menage ([EMAIL PROTECTED]) and Balbir Singh
+ * ([EMAIL PROTECTED])
+ *
+ */
+
+/*
+ * Example cgroup subsystem for reporting total CPU usage of tasks in a
+ * cgroup.
+ */
+
+#include <linux/module.h>
+#include <linux/cgroup.h>
+#include <linux/fs.h>
+#include <linux/rcupdate.h>
+
+struct cpuacct {
+ struct cgroup_subsys_state css;
+ spinlock_t lock;
+ /* total time used by this class (in nanoseconds) */
+ u64 time;
+};
+
+struct cgroup_subsys cpuacct_subsys;
+
+static inline struct cpuacct *cgroup_ca(struct cgroup *cont)
+{
+ return container_of(cgroup_subsys_state(cont, cpuacct_subsys_id),
+ struct cpuacct, css);
+}
+
+static inline struct cpuacct *task_ca(struct task_struct *task)
+{
+ return container_of(task_subsys_state(task, cpuacct_subsys_id),
+ struct cpuacct, css);
+}
+
+static struct cgroup_subsys_state *cpuacct_create(
+ struct cgroup_subsys *ss, struct cgroup *cont)
+{
+ struct cpuacct *ca = kzalloc(sizeof(*ca), GFP_KERNEL);
+
+ if (!ca)
+ return ERR_PTR(-ENOMEM);
+ spin_lock_init(&ca->lock);
+ return &ca->css;
+}
+
+static void cpuacct_destroy(struct cgroup_subsys *ss,
+ struct cgroup *cont)
+{
+ kfree(cgroup_ca(cont));
+}
+
+static u64 cpuusage_read(struct cgroup *cont, struct cftype *cft)
+{
+ struct cpuacct *ca = cgroup_ca(cont);
+
+ return ca->time;
+}
+
+static struct cftype files[] = {
+ {
+ .name = "usage",
+ .read_uint = cpuusage_read,
+ },
+};
+
+static int cpuacct_populate(struct cgroup_subsys *ss, struct cgroup *cont)
+{
+ return cgroup_add_files(cont, ss, files, ARRAY_SIZE(files));
+}
+
+void cpuacct_charge(struct task_struct *task, u64 cputime)
+{
+ struct cpuacct *ca;
+ unsigned long flags;
+
+ if (!cpuacct_subsys.active)
+ return;
+ rcu_read_lock();
+ ca = task_ca(task);
+ if (ca) {
+ spin_lock_irqsave(&ca->lock, flags);
+ ca->time += cputime;
+ spin_unlock_irqrestore(&ca->lock, flags);
+ }
+ rcu_read_unlock();
+}
+
+struct cgroup_subsys cpuacct_subsys = {
+ .name = "cpuacct",
+ .create = cpuacct_create,
+ .destroy = cpuacct_destroy,
+ .populate = cpuacct_populate,
+ .subsys_id = cpuacct_subsys_id,
+};
Index: current/kernel/sched.c
===================================================================
--- current.orig/kernel/sched.c
+++ current/kernel/sched.c
@@ -52,6 +52,7 @@
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/percpu.h>
+#include <linux/cpu_acct.h>
#include <linux/kthread.h>
#include <linux/seq_file.h>
#include <linux/sysctl.h>
@@ -7221,38 +7222,12 @@ static u64 cpu_shares_read_uint(struct c
return (u64) tg->shares;
}
-static u64 cpu_usage_read(struct cgroup *cgrp, struct cftype *cft)
-{
- struct task_group *tg = cgroup_tg(cgrp);
- unsigned long flags;
- u64 res = 0;
- int i;
-
- for_each_possible_cpu(i) {
- /*
- * Lock to prevent races with updating 64-bit counters
- * on 32-bit arches.
- */
- spin_lock_irqsave(&cpu_rq(i)->lock, flags);
- res += tg->se[i]->sum_exec_runtime;
- spin_unlock_irqrestore(&cpu_rq(i)->lock, flags);
- }
- /* Convert from ns to ms */
- do_div(res, NSEC_PER_MSEC);
-
- return res;
-}
-
static struct cftype cpu_files[] = {
{
.name = "shares",
.read_uint = cpu_shares_read_uint,
.write_uint = cpu_shares_write_uint,
},
- {
- .name = "usage",
- .read_uint = cpu_usage_read,
- },
};
static int cpu_cgroup_populate(struct cgroup_subsys *ss, struct cgroup *cont)
Index: current/kernel/sched_fair.c
===================================================================
--- current.orig/kernel/sched_fair.c
+++ current/kernel/sched_fair.c
@@ -351,6 +351,11 @@ static void update_curr(struct cfs_rq *c
__update_curr(cfs_rq, curr, delta_exec);
curr->exec_start = now;
+ if (entity_is_task(curr)) {
+ struct task_struct *curtask = task_of(curr);
+
+ cpuacct_charge(curtask, delta_exec);
+ }
}
static inline void
Index: current/kernel/sched_rt.c
===================================================================
--- current.orig/kernel/sched_rt.c
+++ current/kernel/sched_rt.c
@@ -23,6 +23,7 @@ static void update_curr_rt(struct rq *rq
curr->se.sum_exec_runtime += delta_exec;
curr->se.exec_start = rq->clock;
+ cpuacct_charge(curr, delta_exec);
}
static void enqueue_task_rt(struct rq *rq, struct task_struct *p, int wakeup)
--
Regards,
vatsa
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
