2013-04-17 16:25:04

by Frederic Weisbecker

[permalink] [raw]
Subject: [GIT PULL] nohz: Full dynticks timekeeping and RCU improvement

Hi Ingo,

So with this we are mostly done with the RCU and timekeeping sides.
The interface is still a bit spartan:

* Need to keep the boot CPU outside full dynticks range.
* Need to have nohz_full= be a subset of rcu_nocbs=

Note that none of this is thought to be fixed on the rock. It's
just a KISS interface settlement so that we can focus on the
central things.

These constraints will likely be removed in the future.

Please pull from:

git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
timers/nohz

HEAD: da922312f6980cb9214a64f7303f6cb3cf93626f

Thanks.

Frederic Weisbecker (3):
nohz: Force boot CPU outside full dynticks range
nohz: Ensure full dynticks CPUs are RCU nocbs
nohz: New option to default all CPUs in full dynticks range

Documentation/kernel-parameters.txt | 6 ++-
include/linux/rcupdate.h | 7 +++
include/linux/tick.h | 2 +
init/main.c | 1 +
kernel/rcutree.c | 6 +-
kernel/rcutree.h | 1 -
kernel/rcutree_plugin.h | 13 ++----
kernel/time/Kconfig | 10 +++++
kernel/time/tick-sched.c | 77 ++++++++++++++++++----------------
9 files changed, 72 insertions(+), 51 deletions(-)

--
1.7.5.4


2013-04-17 16:25:00

by Frederic Weisbecker

[permalink] [raw]
Subject: [PATCH 1/3] nohz: Force boot CPU outside full dynticks range

The timekeeping job must be able to run early on boot
because there may be some pre-SMP (and thus pre-initcalls )
components that rely on it. The IO-APIC is one such users
as it tests the timer health by watching jiffies progression.

Given that it happens before we know the initial online
set, we can't rely on it to select a timekeeper. We need
one before SMP time otherwise we simply crash on boot.

To fix this and keep things simple for now, force the boot CPU
outside of the full dynticks range in any case and do this early
on kernel parameter parsing time.

We might want a trickier solution later, expecially for aSMP
architectures that need to assign housekeeping tasks to arbitrary
low power CPUs.

But it's still first pass KISS time for now.

Signed-off-by: Frederic Weisbecker <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Geoff Levand <[email protected]>
Cc: Gilad Ben Yossef <[email protected]>
Cc: Hakan Akkan <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Kevin Hilman <[email protected]>
Cc: Li Zhong <[email protected]>
Cc: Paul E. McKenney <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Thomas Gleixner <[email protected]>
---
Documentation/kernel-parameters.txt | 4 +-
kernel/time/tick-sched.c | 54 +++++++++-------------------------
2 files changed, 17 insertions(+), 41 deletions(-)

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 82365dd..887b297 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -1916,8 +1916,8 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
nohz_full= [KNL,BOOT]
In kernels built with CONFIG_NO_HZ_FULL=y, set
the specified list of CPUs whose tick will be stopped
- whenever possible. You need to keep at least one online
- CPU outside the range to maintain the timekeeping.
+ whenever possible. The boot CPU will be forced outside
+ the range to maintain the timekeeping.

noiotrap [SH] Disables trapped I/O port accesses.

diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index 369b576..2bac5ea 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -158,11 +158,21 @@ int tick_nohz_full_cpu(int cpu)
/* Parse the boot-time nohz CPU list from the kernel parameters. */
static int __init tick_nohz_full_setup(char *str)
{
+ int cpu;
+
alloc_bootmem_cpumask_var(&nohz_full_mask);
- if (cpulist_parse(str, nohz_full_mask) < 0)
+ if (cpulist_parse(str, nohz_full_mask) < 0) {
pr_warning("NOHZ: Incorrect nohz_full cpumask\n");
- else
- have_nohz_full_mask = true;
+ return 1;
+ }
+
+ cpu = smp_processor_id();
+ if (cpumask_test_cpu(cpu, nohz_full_mask)) {
+ pr_warning("NO_HZ: Clearing %d from nohz_full range for timekeeping\n", cpu);
+ cpumask_clear_cpu(cpu, nohz_full_mask);
+ }
+ have_nohz_full_mask = true;
+
return 1;
}
__setup("nohz_full=", tick_nohz_full_setup);
@@ -195,42 +205,8 @@ static char __initdata nohz_full_buf[NR_CPUS + 1];

static int __init init_tick_nohz_full(void)
{
- cpumask_var_t online_nohz;
- int cpu;
-
- if (!have_nohz_full_mask)
- return 0;
-
- cpu_notifier(tick_nohz_cpu_down_callback, 0);
-
- if (!zalloc_cpumask_var(&online_nohz, GFP_KERNEL)) {
- pr_warning("NO_HZ: Not enough memory to check full nohz mask\n");
- return -ENOMEM;
- }
-
- /*
- * CPUs can probably not be concurrently offlined on initcall time.
- * But we are paranoid, aren't we?
- */
- get_online_cpus();
-
- /* Ensure we keep a CPU outside the dynticks range for timekeeping */
- cpumask_and(online_nohz, cpu_online_mask, nohz_full_mask);
- if (cpumask_equal(online_nohz, cpu_online_mask)) {
- pr_warning("NO_HZ: Must keep at least one online CPU "
- "out of nohz_full range\n");
- /*
- * We know the current CPU doesn't have its tick stopped.
- * Let's use it for the timekeeping duty.
- */
- preempt_disable();
- cpu = smp_processor_id();
- pr_warning("NO_HZ: Clearing %d from nohz_full range\n", cpu);
- cpumask_clear_cpu(cpu, nohz_full_mask);
- preempt_enable();
- }
- put_online_cpus();
- free_cpumask_var(online_nohz);
+ if (have_nohz_full_mask)
+ cpu_notifier(tick_nohz_cpu_down_callback, 0);

cpulist_scnprintf(nohz_full_buf, sizeof(nohz_full_buf), nohz_full_mask);
pr_info("NO_HZ: Full dynticks CPUs: %s.\n", nohz_full_buf);
--
1.7.5.4

2013-04-17 16:25:33

by Frederic Weisbecker

[permalink] [raw]
Subject: [PATCH 2/3] nohz: Ensure full dynticks CPUs are RCU nocbs

We need full dynticks CPU to also be RCU nocb so
that we don't have to keep the tick to handle RCU
callbacks.

Make sure the range passed to nohz_full= boot
parameter is a subset of rcu_nocbs=

The CPUs that fail to meet this requirement will be
excluded from the nohz_full range. This is checked
early in boot time, before any CPU has the opportunity
to stop its tick.

Suggested-by: Steven Rostedt <[email protected]>
Signed-off-by: Frederic Weisbecker <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Geoff Levand <[email protected]>
Cc: Gilad Ben Yossef <[email protected]>
Cc: Hakan Akkan <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Kevin Hilman <[email protected]>
Cc: Li Zhong <[email protected]>
Cc: Paul E. McKenney <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Thomas Gleixner <[email protected]>
---
Documentation/kernel-parameters.txt | 2 ++
include/linux/rcupdate.h | 7 +++++++
include/linux/tick.h | 2 ++
init/main.c | 1 +
kernel/rcutree.c | 6 +++---
kernel/rcutree.h | 1 -
kernel/rcutree_plugin.h | 13 ++++---------
kernel/time/tick-sched.c | 22 ++++++++++++++++------
8 files changed, 35 insertions(+), 19 deletions(-)

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 887b297..4865e9b 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -1918,6 +1918,8 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
the specified list of CPUs whose tick will be stopped
whenever possible. The boot CPU will be forced outside
the range to maintain the timekeeping.
+ The CPUs in this range must also be included in the
+ rcu_nocbs= set.

noiotrap [SH] Disables trapped I/O port accesses.

diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index b758ce1..8e0948c 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -999,4 +999,11 @@ static inline notrace void rcu_read_unlock_sched_notrace(void)
#define kfree_rcu(ptr, rcu_head) \
__kfree_rcu(&((ptr)->rcu_head), offsetof(typeof(*(ptr)), rcu_head))

+#ifdef CONFIG_RCU_NOCB_CPU
+extern bool rcu_is_nocb_cpu(int cpu);
+#else
+static inline bool rcu_is_nocb_cpu(int cpu) { return false; }
+#endif /* #else #ifdef CONFIG_RCU_NOCB_CPU */
+
+
#endif /* __LINUX_RCUPDATE_H */
diff --git a/include/linux/tick.h b/include/linux/tick.h
index b4e3b0c..0b6873c 100644
--- a/include/linux/tick.h
+++ b/include/linux/tick.h
@@ -158,8 +158,10 @@ static inline u64 get_cpu_iowait_time_us(int cpu, u64 *unused) { return -1; }
# endif /* !CONFIG_NO_HZ_COMMON */

#ifdef CONFIG_NO_HZ_FULL
+extern void tick_nohz_init(void);
extern int tick_nohz_full_cpu(int cpu);
#else
+static inline void tick_nohz_init(void) { }
static inline int tick_nohz_full_cpu(int cpu) { return 0; }
#endif

diff --git a/init/main.c b/init/main.c
index 63534a1..2acb5bb 100644
--- a/init/main.c
+++ b/init/main.c
@@ -547,6 +547,7 @@ asmlinkage void __init start_kernel(void)
idr_init_cache();
perf_event_init();
rcu_init();
+ tick_nohz_init();
radix_tree_init();
/* init some links before init_ISA_irqs() */
early_irq_init();
diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index f5ab502..1d4ceff 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -1695,7 +1695,7 @@ rcu_send_cbs_to_orphanage(int cpu, struct rcu_state *rsp,
struct rcu_node *rnp, struct rcu_data *rdp)
{
/* No-CBs CPUs do not have orphanable callbacks. */
- if (is_nocb_cpu(rdp->cpu))
+ if (rcu_is_nocb_cpu(rdp->cpu))
return;

/*
@@ -2757,10 +2757,10 @@ static void _rcu_barrier(struct rcu_state *rsp)
* corresponding CPU's preceding callbacks have been invoked.
*/
for_each_possible_cpu(cpu) {
- if (!cpu_online(cpu) && !is_nocb_cpu(cpu))
+ if (!cpu_online(cpu) && !rcu_is_nocb_cpu(cpu))
continue;
rdp = per_cpu_ptr(rsp->rda, cpu);
- if (is_nocb_cpu(cpu)) {
+ if (rcu_is_nocb_cpu(cpu)) {
_rcu_barrier_trace(rsp, "OnlineNoCB", cpu,
rsp->n_barrier_done);
atomic_inc(&rsp->barrier_cpu_count);
diff --git a/kernel/rcutree.h b/kernel/rcutree.h
index f993c0a..38acc49 100644
--- a/kernel/rcutree.h
+++ b/kernel/rcutree.h
@@ -529,7 +529,6 @@ static void print_cpu_stall_info(struct rcu_state *rsp, int cpu);
static void print_cpu_stall_info_end(void);
static void zero_cpu_stall_ticks(struct rcu_data *rdp);
static void increment_cpu_stall_ticks(void);
-static bool is_nocb_cpu(int cpu);
static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
bool lazy);
static bool rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
index a5745e9..0cd91cc 100644
--- a/kernel/rcutree_plugin.h
+++ b/kernel/rcutree_plugin.h
@@ -2167,7 +2167,7 @@ static int __init parse_rcu_nocb_poll(char *arg)
early_param("rcu_nocb_poll", parse_rcu_nocb_poll);

/* Is the specified CPU a no-CPUs CPU? */
-static bool is_nocb_cpu(int cpu)
+bool rcu_is_nocb_cpu(int cpu)
{
if (have_rcu_nocb_mask)
return cpumask_test_cpu(cpu, rcu_nocb_mask);
@@ -2225,7 +2225,7 @@ static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
bool lazy)
{

- if (!is_nocb_cpu(rdp->cpu))
+ if (!rcu_is_nocb_cpu(rdp->cpu))
return 0;
__call_rcu_nocb_enqueue(rdp, rhp, &rhp->next, 1, lazy);
return 1;
@@ -2242,7 +2242,7 @@ static bool __maybe_unused rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
long qll = rsp->qlen_lazy;

/* If this is not a no-CBs CPU, tell the caller to do it the old way. */
- if (!is_nocb_cpu(smp_processor_id()))
+ if (!rcu_is_nocb_cpu(smp_processor_id()))
return 0;
rsp->qlen = 0;
rsp->qlen_lazy = 0;
@@ -2282,7 +2282,7 @@ static bool nocb_cpu_expendable(int cpu)
* If there are no no-CB CPUs or if this CPU is not a no-CB CPU,
* then offlining this CPU is harmless. Let it happen.
*/
- if (!have_rcu_nocb_mask || is_nocb_cpu(cpu))
+ if (!have_rcu_nocb_mask || rcu_is_nocb_cpu(cpu))
return 1;

/* If no memory, play it safe and keep the CPU around. */
@@ -2464,11 +2464,6 @@ static void __init rcu_init_nocb(void)

#else /* #ifdef CONFIG_RCU_NOCB_CPU */

-static bool is_nocb_cpu(int cpu)
-{
- return false;
-}
-
static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
bool lazy)
{
diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index 2bac5ea..d71a5f2 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -203,17 +203,27 @@ static int __cpuinit tick_nohz_cpu_down_callback(struct notifier_block *nfb,
*/
static char __initdata nohz_full_buf[NR_CPUS + 1];

-static int __init init_tick_nohz_full(void)
+void __init tick_nohz_init(void)
{
- if (have_nohz_full_mask)
- cpu_notifier(tick_nohz_cpu_down_callback, 0);
+ int cpu;
+
+ if (!have_nohz_full_mask)
+ return;
+
+ cpu_notifier(tick_nohz_cpu_down_callback, 0);
+
+ /* Make sure full dynticks CPU are also RCU nocbs */
+ for_each_cpu(cpu, nohz_full_mask) {
+ if (!rcu_is_nocb_cpu(cpu)) {
+ pr_warning("NO_HZ: CPU %d is not RCU nocb: "
+ "cleared from nohz_full range", cpu);
+ cpumask_clear_cpu(cpu, nohz_full_mask);
+ }
+ }

cpulist_scnprintf(nohz_full_buf, sizeof(nohz_full_buf), nohz_full_mask);
pr_info("NO_HZ: Full dynticks CPUs: %s.\n", nohz_full_buf);
-
- return 0;
}
-core_initcall(init_tick_nohz_full);
#else
#define have_nohz_full_mask (0)
#endif
--
1.7.5.4

2013-04-17 16:25:29

by Frederic Weisbecker

[permalink] [raw]
Subject: [PATCH 3/3] nohz: New option to default all CPUs in full dynticks range

Provide a new kernel config that defaults all CPUs to be part
of the full dynticks range, except the boot one for timekeeping.

This default setting is overriden by the nohz_full= boot option
if passed by the user.

This is helpful for those who don't need a finegrained range
of full dynticks CPU and also for automated testing.

Suggested-by: Ingo Molnar <[email protected]>
Signed-off-by: Frederic Weisbecker <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Geoff Levand <[email protected]>
Cc: Gilad Ben Yossef <[email protected]>
Cc: Hakan Akkan <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Kevin Hilman <[email protected]>
Cc: Li Zhong <[email protected]>
Cc: Paul E. McKenney <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Thomas Gleixner <[email protected]>
---
kernel/time/Kconfig | 10 ++++++++++
kernel/time/tick-sched.c | 23 +++++++++++++++++++++--
2 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/kernel/time/Kconfig b/kernel/time/Kconfig
index 358d601..99c3f13 100644
--- a/kernel/time/Kconfig
+++ b/kernel/time/Kconfig
@@ -128,6 +128,16 @@ config NO_HZ_FULL

endchoice

+config NO_HZ_FULL_ALL
+ bool "Full dynticks system on all CPUs by default"
+ depends on NO_HZ_FULL
+ help
+ If the user doesn't pass the nohz_full boot option to
+ define the range of full dynticks CPUs, consider that all
+ CPUs in the system are full dynticks by default.
+ Note the boot CPU will still be kept outside the range to
+ handle the timekeeping duty.
+
config NO_HZ
bool "Old Idle dynticks config"
depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS
diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index d71a5f2..a76e090 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -203,12 +203,31 @@ static int __cpuinit tick_nohz_cpu_down_callback(struct notifier_block *nfb,
*/
static char __initdata nohz_full_buf[NR_CPUS + 1];

+static int tick_nohz_init_all(void)
+{
+ int err = -1;
+
+#ifdef CONFIG_NO_HZ_FULL_ALL
+ if (!alloc_cpumask_var(&nohz_full_mask, GFP_KERNEL)) {
+ pr_err("NO_HZ: Can't allocate full dynticks cpumask\n");
+ return err;
+ }
+ err = 0;
+ cpumask_setall(nohz_full_mask);
+ cpumask_clear_cpu(smp_processor_id(), nohz_full_mask);
+ have_nohz_full_mask = true;
+#endif
+ return err;
+}
+
void __init tick_nohz_init(void)
{
int cpu;

- if (!have_nohz_full_mask)
- return;
+ if (!have_nohz_full_mask) {
+ if (tick_nohz_init_all() < 0)
+ return;
+ }

cpu_notifier(tick_nohz_cpu_down_callback, 0);

--
1.7.5.4

2013-04-17 19:27:44

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH 1/3] nohz: Force boot CPU outside full dynticks range

On Wed, Apr 17, 2013 at 06:24:47PM +0200, Frederic Weisbecker wrote:
> The timekeeping job must be able to run early on boot
> because there may be some pre-SMP (and thus pre-initcalls )
> components that rely on it. The IO-APIC is one such users
> as it tests the timer health by watching jiffies progression.
>
> Given that it happens before we know the initial online
> set, we can't rely on it to select a timekeeper. We need
> one before SMP time otherwise we simply crash on boot.
>
> To fix this and keep things simple for now, force the boot CPU
> outside of the full dynticks range in any case and do this early
> on kernel parameter parsing time.
>
> We might want a trickier solution later, expecially for aSMP
> architectures that need to assign housekeeping tasks to arbitrary
> low power CPUs.
>
> But it's still first pass KISS time for now.
>
> Signed-off-by: Frederic Weisbecker <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Chris Metcalf <[email protected]>
> Cc: Christoph Lameter <[email protected]>
> Cc: Geoff Levand <[email protected]>
> Cc: Gilad Ben Yossef <[email protected]>
> Cc: Hakan Akkan <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Kevin Hilman <[email protected]>
> Cc: Li Zhong <[email protected]>
> Cc: Paul E. McKenney <[email protected]>
> Cc: Paul Gortmaker <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Steven Rostedt <[email protected]>
> Cc: Thomas Gleixner <[email protected]>

Reviewed-by: Paul E. McKenney <[email protected]>

(Plus updated the documentation accordingly.)

> ---
> Documentation/kernel-parameters.txt | 4 +-
> kernel/time/tick-sched.c | 54 +++++++++-------------------------
> 2 files changed, 17 insertions(+), 41 deletions(-)
>
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index 82365dd..887b297 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -1916,8 +1916,8 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
> nohz_full= [KNL,BOOT]
> In kernels built with CONFIG_NO_HZ_FULL=y, set
> the specified list of CPUs whose tick will be stopped
> - whenever possible. You need to keep at least one online
> - CPU outside the range to maintain the timekeeping.
> + whenever possible. The boot CPU will be forced outside
> + the range to maintain the timekeeping.
>
> noiotrap [SH] Disables trapped I/O port accesses.
>
> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
> index 369b576..2bac5ea 100644
> --- a/kernel/time/tick-sched.c
> +++ b/kernel/time/tick-sched.c
> @@ -158,11 +158,21 @@ int tick_nohz_full_cpu(int cpu)
> /* Parse the boot-time nohz CPU list from the kernel parameters. */
> static int __init tick_nohz_full_setup(char *str)
> {
> + int cpu;
> +
> alloc_bootmem_cpumask_var(&nohz_full_mask);
> - if (cpulist_parse(str, nohz_full_mask) < 0)
> + if (cpulist_parse(str, nohz_full_mask) < 0) {
> pr_warning("NOHZ: Incorrect nohz_full cpumask\n");
> - else
> - have_nohz_full_mask = true;
> + return 1;
> + }
> +
> + cpu = smp_processor_id();
> + if (cpumask_test_cpu(cpu, nohz_full_mask)) {
> + pr_warning("NO_HZ: Clearing %d from nohz_full range for timekeeping\n", cpu);
> + cpumask_clear_cpu(cpu, nohz_full_mask);
> + }
> + have_nohz_full_mask = true;
> +
> return 1;
> }
> __setup("nohz_full=", tick_nohz_full_setup);
> @@ -195,42 +205,8 @@ static char __initdata nohz_full_buf[NR_CPUS + 1];
>
> static int __init init_tick_nohz_full(void)
> {
> - cpumask_var_t online_nohz;
> - int cpu;
> -
> - if (!have_nohz_full_mask)
> - return 0;
> -
> - cpu_notifier(tick_nohz_cpu_down_callback, 0);
> -
> - if (!zalloc_cpumask_var(&online_nohz, GFP_KERNEL)) {
> - pr_warning("NO_HZ: Not enough memory to check full nohz mask\n");
> - return -ENOMEM;
> - }
> -
> - /*
> - * CPUs can probably not be concurrently offlined on initcall time.
> - * But we are paranoid, aren't we?
> - */
> - get_online_cpus();
> -
> - /* Ensure we keep a CPU outside the dynticks range for timekeeping */
> - cpumask_and(online_nohz, cpu_online_mask, nohz_full_mask);
> - if (cpumask_equal(online_nohz, cpu_online_mask)) {
> - pr_warning("NO_HZ: Must keep at least one online CPU "
> - "out of nohz_full range\n");
> - /*
> - * We know the current CPU doesn't have its tick stopped.
> - * Let's use it for the timekeeping duty.
> - */
> - preempt_disable();
> - cpu = smp_processor_id();
> - pr_warning("NO_HZ: Clearing %d from nohz_full range\n", cpu);
> - cpumask_clear_cpu(cpu, nohz_full_mask);
> - preempt_enable();
> - }
> - put_online_cpus();
> - free_cpumask_var(online_nohz);
> + if (have_nohz_full_mask)
> + cpu_notifier(tick_nohz_cpu_down_callback, 0);
>
> cpulist_scnprintf(nohz_full_buf, sizeof(nohz_full_buf), nohz_full_mask);
> pr_info("NO_HZ: Full dynticks CPUs: %s.\n", nohz_full_buf);
> --
> 1.7.5.4
>

2013-04-17 19:59:10

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH 2/3] nohz: Ensure full dynticks CPUs are RCU nocbs

On Wed, Apr 17, 2013 at 06:24:48PM +0200, Frederic Weisbecker wrote:
> We need full dynticks CPU to also be RCU nocb so
> that we don't have to keep the tick to handle RCU
> callbacks.
>
> Make sure the range passed to nohz_full= boot
> parameter is a subset of rcu_nocbs=
>
> The CPUs that fail to meet this requirement will be
> excluded from the nohz_full range. This is checked
> early in boot time, before any CPU has the opportunity
> to stop its tick.
>
> Suggested-by: Steven Rostedt <[email protected]>
> Signed-off-by: Frederic Weisbecker <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Chris Metcalf <[email protected]>
> Cc: Christoph Lameter <[email protected]>
> Cc: Geoff Levand <[email protected]>
> Cc: Gilad Ben Yossef <[email protected]>
> Cc: Hakan Akkan <[email protected]>
> Cc: Ingo Molnar <mi[email protected]>
> Cc: Kevin Hilman <[email protected]>
> Cc: Li Zhong <[email protected]>
> Cc: Paul E. McKenney <[email protected]>
> Cc: Paul Gortmaker <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Steven Rostedt <[email protected]>
> Cc: Thomas Gleixner <[email protected]>

Reviewed-by: Paul E. McKenney <[email protected]>

(And again, also updated documentation.)

> ---
> Documentation/kernel-parameters.txt | 2 ++
> include/linux/rcupdate.h | 7 +++++++
> include/linux/tick.h | 2 ++
> init/main.c | 1 +
> kernel/rcutree.c | 6 +++---
> kernel/rcutree.h | 1 -
> kernel/rcutree_plugin.h | 13 ++++---------
> kernel/time/tick-sched.c | 22 ++++++++++++++++------
> 8 files changed, 35 insertions(+), 19 deletions(-)
>
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index 887b297..4865e9b 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -1918,6 +1918,8 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
> the specified list of CPUs whose tick will be stopped
> whenever possible. The boot CPU will be forced outside
> the range to maintain the timekeeping.
> + The CPUs in this range must also be included in the
> + rcu_nocbs= set.
>
> noiotrap [SH] Disables trapped I/O port accesses.
>
> diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
> index b758ce1..8e0948c 100644
> --- a/include/linux/rcupdate.h
> +++ b/include/linux/rcupdate.h
> @@ -999,4 +999,11 @@ static inline notrace void rcu_read_unlock_sched_notrace(void)
> #define kfree_rcu(ptr, rcu_head) \
> __kfree_rcu(&((ptr)->rcu_head), offsetof(typeof(*(ptr)), rcu_head))
>
> +#ifdef CONFIG_RCU_NOCB_CPU
> +extern bool rcu_is_nocb_cpu(int cpu);
> +#else
> +static inline bool rcu_is_nocb_cpu(int cpu) { return false; }
> +#endif /* #else #ifdef CONFIG_RCU_NOCB_CPU */
> +
> +
> #endif /* __LINUX_RCUPDATE_H */
> diff --git a/include/linux/tick.h b/include/linux/tick.h
> index b4e3b0c..0b6873c 100644
> --- a/include/linux/tick.h
> +++ b/include/linux/tick.h
> @@ -158,8 +158,10 @@ static inline u64 get_cpu_iowait_time_us(int cpu, u64 *unused) { return -1; }
> # endif /* !CONFIG_NO_HZ_COMMON */
>
> #ifdef CONFIG_NO_HZ_FULL
> +extern void tick_nohz_init(void);
> extern int tick_nohz_full_cpu(int cpu);
> #else
> +static inline void tick_nohz_init(void) { }
> static inline int tick_nohz_full_cpu(int cpu) { return 0; }
> #endif
>
> diff --git a/init/main.c b/init/main.c
> index 63534a1..2acb5bb 100644
> --- a/init/main.c
> +++ b/init/main.c
> @@ -547,6 +547,7 @@ asmlinkage void __init start_kernel(void)
> idr_init_cache();
> perf_event_init();
> rcu_init();
> + tick_nohz_init();
> radix_tree_init();
> /* init some links before init_ISA_irqs() */
> early_irq_init();
> diff --git a/kernel/rcutree.c b/kernel/rcutree.c
> index f5ab502..1d4ceff 100644
> --- a/kernel/rcutree.c
> +++ b/kernel/rcutree.c
> @@ -1695,7 +1695,7 @@ rcu_send_cbs_to_orphanage(int cpu, struct rcu_state *rsp,
> struct rcu_node *rnp, struct rcu_data *rdp)
> {
> /* No-CBs CPUs do not have orphanable callbacks. */
> - if (is_nocb_cpu(rdp->cpu))
> + if (rcu_is_nocb_cpu(rdp->cpu))
> return;
>
> /*
> @@ -2757,10 +2757,10 @@ static void _rcu_barrier(struct rcu_state *rsp)
> * corresponding CPU's preceding callbacks have been invoked.
> */
> for_each_possible_cpu(cpu) {
> - if (!cpu_online(cpu) && !is_nocb_cpu(cpu))
> + if (!cpu_online(cpu) && !rcu_is_nocb_cpu(cpu))
> continue;
> rdp = per_cpu_ptr(rsp->rda, cpu);
> - if (is_nocb_cpu(cpu)) {
> + if (rcu_is_nocb_cpu(cpu)) {
> _rcu_barrier_trace(rsp, "OnlineNoCB", cpu,
> rsp->n_barrier_done);
> atomic_inc(&rsp->barrier_cpu_count);
> diff --git a/kernel/rcutree.h b/kernel/rcutree.h
> index f993c0a..38acc49 100644
> --- a/kernel/rcutree.h
> +++ b/kernel/rcutree.h
> @@ -529,7 +529,6 @@ static void print_cpu_stall_info(struct rcu_state *rsp, int cpu);
> static void print_cpu_stall_info_end(void);
> static void zero_cpu_stall_ticks(struct rcu_data *rdp);
> static void increment_cpu_stall_ticks(void);
> -static bool is_nocb_cpu(int cpu);
> static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
> bool lazy);
> static bool rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
> diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
> index a5745e9..0cd91cc 100644
> --- a/kernel/rcutree_plugin.h
> +++ b/kernel/rcutree_plugin.h
> @@ -2167,7 +2167,7 @@ static int __init parse_rcu_nocb_poll(char *arg)
> early_param("rcu_nocb_poll", parse_rcu_nocb_poll);
>
> /* Is the specified CPU a no-CPUs CPU? */
> -static bool is_nocb_cpu(int cpu)
> +bool rcu_is_nocb_cpu(int cpu)
> {
> if (have_rcu_nocb_mask)
> return cpumask_test_cpu(cpu, rcu_nocb_mask);
> @@ -2225,7 +2225,7 @@ static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
> bool lazy)
> {
>
> - if (!is_nocb_cpu(rdp->cpu))
> + if (!rcu_is_nocb_cpu(rdp->cpu))
> return 0;
> __call_rcu_nocb_enqueue(rdp, rhp, &rhp->next, 1, lazy);
> return 1;
> @@ -2242,7 +2242,7 @@ static bool __maybe_unused rcu_nocb_adopt_orphan_cbs(struct rcu_state *rsp,
> long qll = rsp->qlen_lazy;
>
> /* If this is not a no-CBs CPU, tell the caller to do it the old way. */
> - if (!is_nocb_cpu(smp_processor_id()))
> + if (!rcu_is_nocb_cpu(smp_processor_id()))
> return 0;
> rsp->qlen = 0;
> rsp->qlen_lazy = 0;
> @@ -2282,7 +2282,7 @@ static bool nocb_cpu_expendable(int cpu)
> * If there are no no-CB CPUs or if this CPU is not a no-CB CPU,
> * then offlining this CPU is harmless. Let it happen.
> */
> - if (!have_rcu_nocb_mask || is_nocb_cpu(cpu))
> + if (!have_rcu_nocb_mask || rcu_is_nocb_cpu(cpu))
> return 1;
>
> /* If no memory, play it safe and keep the CPU around. */
> @@ -2464,11 +2464,6 @@ static void __init rcu_init_nocb(void)
>
> #else /* #ifdef CONFIG_RCU_NOCB_CPU */
>
> -static bool is_nocb_cpu(int cpu)
> -{
> - return false;
> -}
> -
> static bool __call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *rhp,
> bool lazy)
> {
> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
> index 2bac5ea..d71a5f2 100644
> --- a/kernel/time/tick-sched.c
> +++ b/kernel/time/tick-sched.c
> @@ -203,17 +203,27 @@ static int __cpuinit tick_nohz_cpu_down_callback(struct notifier_block *nfb,
> */
> static char __initdata nohz_full_buf[NR_CPUS + 1];
>
> -static int __init init_tick_nohz_full(void)
> +void __init tick_nohz_init(void)
> {
> - if (have_nohz_full_mask)
> - cpu_notifier(tick_nohz_cpu_down_callback, 0);
> + int cpu;
> +
> + if (!have_nohz_full_mask)
> + return;
> +
> + cpu_notifier(tick_nohz_cpu_down_callback, 0);
> +
> + /* Make sure full dynticks CPU are also RCU nocbs */
> + for_each_cpu(cpu, nohz_full_mask) {
> + if (!rcu_is_nocb_cpu(cpu)) {
> + pr_warning("NO_HZ: CPU %d is not RCU nocb: "
> + "cleared from nohz_full range", cpu);
> + cpumask_clear_cpu(cpu, nohz_full_mask);
> + }
> + }
>
> cpulist_scnprintf(nohz_full_buf, sizeof(nohz_full_buf), nohz_full_mask);
> pr_info("NO_HZ: Full dynticks CPUs: %s.\n", nohz_full_buf);
> -
> - return 0;
> }
> -core_initcall(init_tick_nohz_full);
> #else
> #define have_nohz_full_mask (0)
> #endif
> --
> 1.7.5.4
>

2013-04-17 20:03:27

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH 3/3] nohz: New option to default all CPUs in full dynticks range

On Wed, Apr 17, 2013 at 06:24:49PM +0200, Frederic Weisbecker wrote:
> Provide a new kernel config that defaults all CPUs to be part
> of the full dynticks range, except the boot one for timekeeping.
>
> This default setting is overriden by the nohz_full= boot option
> if passed by the user.
>
> This is helpful for those who don't need a finegrained range
> of full dynticks CPU and also for automated testing.
>
> Suggested-by: Ingo Molnar <[email protected]>
> Signed-off-by: Frederic Weisbecker <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Chris Metcalf <[email protected]>
> Cc: Christoph Lameter <[email protected]>
> Cc: Geoff Levand <[email protected]>
> Cc: Gilad Ben Yossef <[email protected]>
> Cc: Hakan Akkan <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Kevin Hilman <[email protected]>
> Cc: Li Zhong <[email protected]>
> Cc: Paul E. McKenney <[email protected]>
> Cc: Paul Gortmaker <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Steven Rostedt <[email protected]>
> Cc: Thomas Gleixner <[email protected]>

Reviewed-by: Paul E. McKenney <[email protected]>

> ---
> kernel/time/Kconfig | 10 ++++++++++
> kernel/time/tick-sched.c | 23 +++++++++++++++++++++--
> 2 files changed, 31 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/time/Kconfig b/kernel/time/Kconfig
> index 358d601..99c3f13 100644
> --- a/kernel/time/Kconfig
> +++ b/kernel/time/Kconfig
> @@ -128,6 +128,16 @@ config NO_HZ_FULL
>
> endchoice
>
> +config NO_HZ_FULL_ALL
> + bool "Full dynticks system on all CPUs by default"
> + depends on NO_HZ_FULL
> + help
> + If the user doesn't pass the nohz_full boot option to
> + define the range of full dynticks CPUs, consider that all
> + CPUs in the system are full dynticks by default.
> + Note the boot CPU will still be kept outside the range to
> + handle the timekeeping duty.
> +
> config NO_HZ
> bool "Old Idle dynticks config"
> depends on !ARCH_USES_GETTIMEOFFSET && GENERIC_CLOCKEVENTS
> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
> index d71a5f2..a76e090 100644
> --- a/kernel/time/tick-sched.c
> +++ b/kernel/time/tick-sched.c
> @@ -203,12 +203,31 @@ static int __cpuinit tick_nohz_cpu_down_callback(struct notifier_block *nfb,
> */
> static char __initdata nohz_full_buf[NR_CPUS + 1];
>
> +static int tick_nohz_init_all(void)
> +{
> + int err = -1;
> +
> +#ifdef CONFIG_NO_HZ_FULL_ALL
> + if (!alloc_cpumask_var(&nohz_full_mask, GFP_KERNEL)) {
> + pr_err("NO_HZ: Can't allocate full dynticks cpumask\n");
> + return err;
> + }
> + err = 0;
> + cpumask_setall(nohz_full_mask);
> + cpumask_clear_cpu(smp_processor_id(), nohz_full_mask);
> + have_nohz_full_mask = true;
> +#endif
> + return err;
> +}
> +
> void __init tick_nohz_init(void)
> {
> int cpu;
>
> - if (!have_nohz_full_mask)
> - return;
> + if (!have_nohz_full_mask) {
> + if (tick_nohz_init_all() < 0)
> + return;
> + }
>
> cpu_notifier(tick_nohz_cpu_down_callback, 0);
>
> --
> 1.7.5.4
>

2013-04-19 12:04:57

by Frederic Weisbecker

[permalink] [raw]
Subject: Re: [GIT PULL] nohz: Full dynticks timekeeping and RCU improvement

2013/4/17 Frederic Weisbecker <[email protected]>:
> Hi Ingo,
>
> So with this we are mostly done with the RCU and timekeeping sides.
> The interface is still a bit spartan:
>
> * Need to keep the boot CPU outside full dynticks range.
> * Need to have nohz_full= be a subset of rcu_nocbs=
>
> Note that none of this is thought to be fixed on the rock. It's
> just a KISS interface settlement so that we can focus on the
> central things.
>
> These constraints will likely be removed in the future.
>
> Please pull from:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
> timers/nohz

And there is the branch "timers/nohz-reviewed" which contain the exact
same patches but rebased to add Paul's Reviewed-by: tags.

Thanks.

2013-04-21 09:05:16

by Ingo Molnar

[permalink] [raw]
Subject: Re: [GIT PULL] nohz: Full dynticks timekeeping and RCU improvement


* Frederic Weisbecker <[email protected]> wrote:

> 2013/4/17 Frederic Weisbecker <[email protected]>:
> > Hi Ingo,
> >
> > So with this we are mostly done with the RCU and timekeeping sides.
> > The interface is still a bit spartan:
> >
> > * Need to keep the boot CPU outside full dynticks range.
> > * Need to have nohz_full= be a subset of rcu_nocbs=
> >
> > Note that none of this is thought to be fixed on the rock. It's
> > just a KISS interface settlement so that we can focus on the
> > central things.
> >
> > These constraints will likely be removed in the future.
> >
> > Please pull from:
> >
> > git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
> > timers/nohz
>
> And there is the branch "timers/nohz-reviewed" which contain the exact
> same patches but rebased to add Paul's Reviewed-by: tags.

Pulled, thanks Frederic!

Ingo