Skip to contentSkip to conversation

sched/nohz: add debugfs control over sched_tick_max_deferment

PATCH In review Kevin Hilman 3 participants 9 messages Dec 17, 2013 – Jan 25, 2014 Scheduler

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 sched/nohz: add debugfs control over sched_tick_max_deferment

Allow debugfs override of sched_tick_max_deferment in order to ease finding/fixing the remaining issues with full nohz.

The value to be written is in jiffies, and -1 means the max deferment is disabled (scheduler_tick_max_deferment() returns KTIME_MAX.)

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
kernel/sched/core.c +15 -1
index 5ac63c9a995a..4b1fe3e69fe4 100644
@@ -2175,6 +2175,8 @@ void scheduler_tick(void)
 }
 
 #ifdef CONFIG_NO_HZ_FULL
+static u32 sched_tick_max_deferment = HZ;
+
 /**
  * scheduler_tick_max_deferment
  *
@@ -2193,13 +2195,25 @@ u64 scheduler_tick_max_deferment(void)
 	struct rq *rq = this_rq();
 	unsigned long next, now = ACCESS_ONCE(jiffies);
 
-	next = rq->last_sched_tick + HZ;
+	if (sched_tick_max_deferment == -1)
+		return KTIME_MAX;
+
+	next = rq->last_sched_tick + sched_tick_max_deferment;
 
 	if (time_before_eq(next, now))
 		return 0;
 
 	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
 }
+
+static __init int sched_nohz_full_init_debug(void)
+{
+	debugfs_create_u32("sched_tick_max_deferment", 0644, NULL,
+			   &sched_tick_max_deferment);
+
+	return 0;
+}
+late_initcall(sched_nohz_full_init_debug);
 #endif
 
 notrace unsigned long get_parent_ip(unsigned long addr)
Signature · 1 line
1.8.3
#1 lore raw

PATCH 2/2 sched/nohz: fix overflow error in scheduler_tick_max_deferment()

The conversion of the max deferment from usecs to nsecs can easily overflow on platforms where a long is 32-bits. To fix, cast the usecs value to u64 before multiplying by NSECS_PER_USEC.

This was discovered on 32-bit ARM platform when extending the max deferment value.

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/core.c +1 -1
index 4b1fe3e69fe4..3d7c80e1c4d9 100644
@@ -2203,7 +2203,7 @@ u64 scheduler_tick_max_deferment(void)
 	if (time_before_eq(next, now))
 		return 0;
 
-	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
+	return (u64)jiffies_to_usecs(next - now) * NSEC_PER_USEC;
 }
 
 static __init int sched_nohz_full_init_debug(void)
Signature · 1 line
1.8.3
#2 lore raw

Re: [PATCH 2/2] sched/nohz: fix overflow error in scheduler_tick_max_deferment()

Quoted 18 lines · Kevin Hilman

The conversion of the max deferment from usecs to nsecs can easily overflow on platforms where a long is 32-bits. To fix, cast the usecs value to u64 before multiplying by NSECS_PER_USEC.

This was discovered on 32-bit ARM platform when extending the max deferment value.

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/core.c +0 -0
index 4b1fe3e69fe4..3d7c80e1c4d9 100644

kernel/sched/core.c u64 scheduler_tick_max_deferment(void) · line 2206

Quoted 5 lines
kernel/sched/core.c +1 -1
@@ -2203,7 +2203,7 @@ u64 scheduler_tick_max_deferment(void)
 	if (time_before_eq(next, now))
 		return 0;
 
-	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
+	return (u64)jiffies_to_usecs(next - now) * NSEC_PER_USEC;

Just to be sure I understand the issue. The problem is that jiffies_to_usecs() return an unsigned int which is then multiplied by NSEC_PER_USEC. If the result of the mul is too big to be stored in an unsigned int, we overflow and may loose some high part of the result. Right?

Quoted 6 lines
kernel/sched/core.c +0 -0
 }
 
 static __init int sched_nohz_full_init_debug(void)
Signature · 1 line
1.8.3
#3 lore raw

Re: [PATCH 2/2] sched/nohz: fix overflow error in scheduler_tick_max_deferment()

Quoted 22 lines · Frederic Weisbecker
Quoted 21 lines · Kevin Hilman

The conversion of the max deferment from usecs to nsecs can easily overflow on platforms where a long is 32-bits. To fix, cast the usecs value to u64 before multiplying by NSECS_PER_USEC.

This was discovered on 32-bit ARM platform when extending the max deferment value.

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/core.c +0 -0
index 4b1fe3e69fe4..3d7c80e1c4d9 100644
@@ -2203,7 +2203,7 @@ u64 scheduler_tick_max_deferment(void)
 	if (time_before_eq(next, now))
 		return 0;
 
Quoted 7 lines

kernel/sched/core.c u64 scheduler_tick_max_deferment(void) · line 2206

Quoted 2 lines
kernel/sched/core.c +1 -1
-	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
+	return (u64)jiffies_to_usecs(next - now) * NSEC_PER_USEC;

Just to be sure I understand the issue. The problem is that jiffies_to_usecs() return an unsigned int which is then multiplied by NSEC_PER_USEC. If the result of the mul is too big to be stored in an unsigned int, we overflow and may loose some high part of the result. Right?

Correct.

Kevin

#4 lore raw

[tip:timers/urgent] sched/nohz: Fix overflow error in scheduler_tick_max_deferment()

In reply to Kevin Hilman
sched/nohz: Fix overflow error in scheduler_tick_max_deferment()
Commit-ID:  8fe8ff09ce3b5750e1f3e45a1f4a81d59c7ff1f1
Gitweb:     http://git.kernel.org/tip/8fe8ff09ce3b5750e1f3e45a1f4a81d59c7ff1f1
Author:     Kevin Hilman <[email protected]>
AuthorDate: Wed, 15 Jan 2014 14:51:38 +0100
Committer:  Frederic Weisbecker <[email protected]>
CommitDate: Thu, 16 Jan 2014 00:08:12 +0100

sched/nohz: Fix overflow error in scheduler_tick_max_deferment()

While calculating the scheduler tick max deferment, the delta is converted from microseconds to nanoseconds through a multiplication against NSEC_PER_USEC.

But this microseconds operand is an unsigned int, thus the result may likely overflow. The result is cast to u64 but only once the operation is completed, which is too late to avoid overflown result.

This is currently not a problem because the scheduler tick max deferment is 1 second. But this may become an issue as we plan to make this value tunable.

So lets fix this by casting the usecs value to u64 before multiplying by NSECS_PER_USEC.

Also to prevent from this kind of mistake to happen again, move this ad-hoc jiffies -> nsecs conversion to a new helper.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Alex Shi <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Paul E. McKenney <[email protected]>
Cc John Stultz <[email protected]>
Cc Kevin Hilman <[email protected]>
[move ad-hoc conversion to jiffies_to_nsecs helper]
---
 include/linux/jiffies.h | 6 ++++++
 kernel/sched/core.c     | 2 +-
 2 files changed, 7 insertions(+), 1 deletion(-)
include/linux/jiffies.h +6 -0
index d235e88..1f44466 100644
@@ -294,6 +294,12 @@ extern unsigned long preset_lpj;
  */
 extern unsigned int jiffies_to_msecs(const unsigned long j);
 extern unsigned int jiffies_to_usecs(const unsigned long j);
+
+static inline u64 jiffies_to_nsecs(const unsigned long j)
+{
+	return (u64)jiffies_to_usecs(j) * NSEC_PER_USEC;
+}
+
 extern unsigned long msecs_to_jiffies(const unsigned int m);
 extern unsigned long usecs_to_jiffies(const unsigned int u);
 extern unsigned long timespec_to_jiffies(const struct timespec *value);
kernel/sched/core.c +1 -1
index a88f4a4..61e601f 100644
@@ -2325,7 +2325,7 @@ u64 scheduler_tick_max_deferment(void)
 	if (time_before_eq(next, now))
 		return 0;
 
-	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
+	return jiffies_to_nsecs(next - now);
 }
 #endif
 
#5 lore raw
In reply to Kevin Hilman
Quoted 39 lines · Kevin Hilman

Allow debugfs override of sched_tick_max_deferment in order to ease finding/fixing the remaining issues with full nohz.

The value to be written is in jiffies, and -1 means the max deferment is disabled (scheduler_tick_max_deferment() returns KTIME_MAX.)

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
kernel/sched/core.c +6 -1
index 5ac63c9a995a..4b1fe3e69fe4 100644
@@ -2175,6 +2175,8 @@ void scheduler_tick(void)
 }
 
 #ifdef CONFIG_NO_HZ_FULL
+static u32 sched_tick_max_deferment = HZ;
+
 /**
  * scheduler_tick_max_deferment
  *
@@ -2193,13 +2195,25 @@ u64 scheduler_tick_max_deferment(void)
 	struct rq *rq = this_rq();
 	unsigned long next, now = ACCESS_ONCE(jiffies);
 
-	next = rq->last_sched_tick + HZ;
+	if (sched_tick_max_deferment == -1)
+		return KTIME_MAX;
+
+	next = rq->last_sched_tick + sched_tick_max_deferment;
 
 	if (time_before_eq(next, now))
 		return 0;
 
 	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;

kernel/sched/core.c u64 scheduler_tick_max_deferment(void) · line 2216

Quoted 10 lines
kernel/sched/core.c +9 -0
 }
+
+static __init int sched_nohz_full_init_debug(void)
+{
+	debugfs_create_u32("sched_tick_max_deferment", 0644, NULL,
+			   &sched_tick_max_deferment);
+
+	return 0;
+}
+late_initcall(sched_nohz_full_init_debug);

If the goal is mostly to turn off sched_tick_max_deferment (set to -1), we should perhaps make it a boolean sched feature (see kernel/sched/features.h) as it's a pretty well consolidated interface.

Quoted 6 lines
kernel/sched/core.c +0 -0
 #endif
 
 notrace unsigned long get_parent_ip(unsigned long addr)
Signature · 1 line
1.8.3
#6 lore raw
Quoted 39 lines · Frederic Weisbecker
Quoted 38 lines · Kevin Hilman

Allow debugfs override of sched_tick_max_deferment in order to ease finding/fixing the remaining issues with full nohz.

The value to be written is in jiffies, and -1 means the max deferment is disabled (scheduler_tick_max_deferment() returns KTIME_MAX.)

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
kernel/sched/core.c +6 -1
index 5ac63c9a995a..4b1fe3e69fe4 100644
@@ -2175,6 +2175,8 @@ void scheduler_tick(void)
 }
 
 #ifdef CONFIG_NO_HZ_FULL
+static u32 sched_tick_max_deferment = HZ;
+
 /**
  * scheduler_tick_max_deferment
  *
@@ -2193,13 +2195,25 @@ u64 scheduler_tick_max_deferment(void)
 	struct rq *rq = this_rq();
 	unsigned long next, now = ACCESS_ONCE(jiffies);
 
-	next = rq->last_sched_tick + HZ;
+	if (sched_tick_max_deferment == -1)
+		return KTIME_MAX;
+
+	next = rq->last_sched_tick + sched_tick_max_deferment;
 
 	if (time_before_eq(next, now))
 		return 0;
 
Quoted 15 lines

kernel/sched/core.c u64 scheduler_tick_max_deferment(void) · line 2216

Quoted 11 lines
kernel/sched/core.c +9 -0
 	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
 }
+
+static __init int sched_nohz_full_init_debug(void)
+{
+	debugfs_create_u32("sched_tick_max_deferment", 0644, NULL,
+			   &sched_tick_max_deferment);
+
+	return 0;
+}
+late_initcall(sched_nohz_full_init_debug);

If the goal is mostly to turn off sched_tick_max_deferment (set to -1), we should perhaps make it a boolean sched feature (see kernel/sched/features.h) as it's a pretty well consolidated interface.

Well, I suspect folks may want to set it to various values, depending on workload to experiment with the results.

Also, my first attempt was to add control over this via sysctl[1] (though not sched_features) and you suggested[2] I use debugfs instead since this should be a temporary hack until we can remove the 1Hz residual tick.

Kevin

[1] http://marc.info/?l=linux-kernel&m=137159992306877&w=2
[2] http://marc.info/?l=linux-kernel&m=137166737830821&w=2

#7 lore raw
Quoted 57 lines · Kevin Hilman
Quoted 39 lines · Frederic Weisbecker
Quoted 38 lines · Kevin Hilman

Allow debugfs override of sched_tick_max_deferment in order to ease finding/fixing the remaining issues with full nohz.

The value to be written is in jiffies, and -1 means the max deferment is disabled (scheduler_tick_max_deferment() returns KTIME_MAX.)

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
kernel/sched/core.c +6 -1
index 5ac63c9a995a..4b1fe3e69fe4 100644
@@ -2175,6 +2175,8 @@ void scheduler_tick(void)
 }
 
 #ifdef CONFIG_NO_HZ_FULL
+static u32 sched_tick_max_deferment = HZ;
+
 /**
  * scheduler_tick_max_deferment
  *
@@ -2193,13 +2195,25 @@ u64 scheduler_tick_max_deferment(void)
 	struct rq *rq = this_rq();
 	unsigned long next, now = ACCESS_ONCE(jiffies);
 
-	next = rq->last_sched_tick + HZ;
+	if (sched_tick_max_deferment == -1)
+		return KTIME_MAX;
+
+	next = rq->last_sched_tick + sched_tick_max_deferment;
 
 	if (time_before_eq(next, now))
 		return 0;
 
Quoted 15 lines

kernel/sched/core.c u64 scheduler_tick_max_deferment(void) · line 2216

Quoted 11 lines
kernel/sched/core.c +9 -0
 	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
 }
+
+static __init int sched_nohz_full_init_debug(void)
+{
+	debugfs_create_u32("sched_tick_max_deferment", 0644, NULL,
+			   &sched_tick_max_deferment);
+
+	return 0;
+}
+late_initcall(sched_nohz_full_init_debug);

If the goal is mostly to turn off sched_tick_max_deferment (set to -1), we should perhaps make it a boolean sched feature (see kernel/sched/features.h) as it's a pretty well consolidated interface.

Quoted 2 lines

Well, I suspect folks may want to set it to various values, depending on workload to experiment with the results.

Another option is to add an integer file in sched_features/ debugfs directory. But all other files there are boolean, so that wouldn't integrate there very well.

One of the things I would like to try is to offline sched_class(current[$CPU])::scheduler_tick() to the timekeeper or any housekeeping CPU.

So the housekeeper could handle the periodic tick on behalf of full dynticks CPUs. And then being able to tune the frequency of this sounds interesting.

So yeah having an tunable integer makes sense after all.

Quoted 4 lines

Also, my first attempt was to add control over this via sysctl[1] (though not sched_features) and you suggested[2] I use debugfs instead since this should be a temporary hack until we can remove the 1Hz residual tick.

Right, but SCHED_FEAT are debugfs :) And I thought we could either reuse it
or reuse the sched feature debugfs directory. But again I realize it's all made of bool values so it's not very welcoming for consistency.

Anyway thinking about it more, perhaps we should actually use your patch that use sysctl since the rest of the scheduler does that for tunable numbers.

Now since it's sysctl, I'm kind of more picky about correctness limits: what if people set high values, thinking the kernel can handle them just fine, while it can't yet obviously? Should we ignore values that goes too far? And how to we draw the line?

Thoughts?

Quoted 4 lines

Kevin

[1] http://marc.info/?l=linux-kernel&m=137159992306877&w=2
[2] http://marc.info/?l=linux-kernel&m=137166737830821&w=2

#8 lore raw
Quoted 81 lines · Frederic Weisbecker
Quoted 57 lines · Kevin Hilman
Quoted 39 lines · Frederic Weisbecker
Quoted 38 lines · Kevin Hilman

Allow debugfs override of sched_tick_max_deferment in order to ease finding/fixing the remaining issues with full nohz.

The value to be written is in jiffies, and -1 means the max deferment is disabled (scheduler_tick_max_deferment() returns KTIME_MAX.)

Cc Frederic Weisbecker <[email protected]>
---
 kernel/sched/core.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
kernel/sched/core.c +6 -1
index 5ac63c9a995a..4b1fe3e69fe4 100644
@@ -2175,6 +2175,8 @@ void scheduler_tick(void)
 }
 
 #ifdef CONFIG_NO_HZ_FULL
+static u32 sched_tick_max_deferment = HZ;
+
 /**
  * scheduler_tick_max_deferment
  *
@@ -2193,13 +2195,25 @@ u64 scheduler_tick_max_deferment(void)
 	struct rq *rq = this_rq();
 	unsigned long next, now = ACCESS_ONCE(jiffies);
 
-	next = rq->last_sched_tick + HZ;
+	if (sched_tick_max_deferment == -1)
+		return KTIME_MAX;
+
+	next = rq->last_sched_tick + sched_tick_max_deferment;
 
 	if (time_before_eq(next, now))
 		return 0;
 
Quoted 15 lines

kernel/sched/core.c u64 scheduler_tick_max_deferment(void) · line 2216

Quoted 11 lines
kernel/sched/core.c +9 -0
 	return jiffies_to_usecs(next - now) * NSEC_PER_USEC;
 }
+
+static __init int sched_nohz_full_init_debug(void)
+{
+	debugfs_create_u32("sched_tick_max_deferment", 0644, NULL,
+			   &sched_tick_max_deferment);
+
+	return 0;
+}
+late_initcall(sched_nohz_full_init_debug);

If the goal is mostly to turn off sched_tick_max_deferment (set to -1), we should perhaps make it a boolean sched feature (see kernel/sched/features.h) as it's a pretty well consolidated interface.

Quoted 2 lines

Well, I suspect folks may want to set it to various values, depending on workload to experiment with the results.

Another option is to add an integer file in sched_features/ debugfs directory. But all other files there are boolean, so that wouldn't integrate there very well.

One of the things I would like to try is to offline sched_class(current[$CPU])::scheduler_tick() to the timekeeper or any housekeeping CPU.

So the housekeeper could handle the periodic tick on behalf of full dynticks CPUs. And then being able to tune the frequency of this sounds interesting.

So yeah having an tunable integer makes sense after all.

Quoted 4 lines

Also, my first attempt was to add control over this via sysctl[1] (though not sched_features) and you suggested[2] I use debugfs instead since this should be a temporary hack until we can remove the 1Hz residual tick.

Right, but SCHED_FEAT are debugfs :) And I thought we could either reuse it
or reuse the sched feature debugfs directory. But again I realize it's all made of bool values so it's not very welcoming for consistency.

Quoted 8 lines

Anyway thinking about it more, perhaps we should actually use your patch that use sysctl since the rest of the scheduler does that for tunable numbers.

Now since it's sysctl, I'm kind of more picky about correctness limits: what if people set high values, thinking the kernel can handle them just fine, while it can't yet obviously? Should we ignore values that goes too far? And how to we draw the line?

Thoughts?

Well, since I think itt's more of a debug feature, I don't think we should draw a line. For example, the current patch lets you disable it completly by setting it to -1. Surely this will break some use cases, but breaking things is kinda the point so we are better able to see where the problems are.

Also, you previously made the case that it shouldn't be a sysctl option since this will (hopefully) be going away in the not too distant future. Isn't that still the case?

Kevin