Sched: Add a tracepoint to track rq->nr_running
PATCH 1/1 Sched: Add a tracepoint to track rq->nr_running
Add a bare tracepoint trace_sched_update_nr_running_tp which tracks ->nr_running CPU's rq. This is used to accurately trace this data and provide a visualization of scheduler imbalances in, for example, the form of a heat map. The tracepoint is accessed by loading an external kernel module. An example module (forked from Qais' module and including the pelt related tracepoints) can be found at:
https://github.com/auldp/tracepoints-helpers.git
A script to turn the trace-cmd report output into a heatmap plot can be found at:
https://github.com/jirvoz/plot-nr-running
The tracepoints are added to add_nr_running() and sub_nr_running() which are in kernel/sched/sched.h. Since sched.h includes trace/events/tlb.h via mmu_context.h we had to limit when CREATE_TRACE_POINTS is defined.
--- include/trace/events/sched.h | 4 ++++ kernel/sched/core.c | 9 ++++----- kernel/sched/fair.c | 2 -- kernel/sched/pelt.c | 2 -- kernel/sched/sched.h | 12 ++++++++++++ 5 files changed, 20 insertions(+), 9 deletions(-)
include/trace/events/sched.h +4 -0
@@ -634,6 +634,10 @@ DECLARE_TRACE(sched_overutilized_tp, TP_PROTO(struct root_domain *rd, bool overutilized), TP_ARGS(rd, overutilized)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(int cpu, int change, unsigned int nr_running), +TP_ARGS (cpu, change, nr_running)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */
kernel/sched/core.c +4 -5
@@ -6,7 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ + +#define SCHED_CREATE_TRACE_POINTS #include "sched.h" +#undef SCHED_CREATE_TRACE_POINTS#include <linux/nospec.h>
@@ -21,9 +24,6 @@#include "pelt.h" -#define CREATE_TRACE_POINTS -#include <trace/events/sched.h> -/* * Export tracepoints that act as a bare tracehook (ie: have no trace event * associated with them) to allow external modules to probe them.
@@ -34,6 +34,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(pelt_dl_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_irq_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_se_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_overutilized_tp); +EXPORT_TRACEPOINT_SYMBOL_GPL (sched_update_nr_running_tp); DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
@@ -7969,5 +7970,3 @@ const u32 sched_prio_to_wmult[40] = {/* 10 */ 39045157 ,49367440 ,61356676 ,76695844 ,95443717 , /* 15 */ 119304647, 148102320, 186737708, 238609294, 286331153, }; - -#undef CREATE_TRACE_POINTS
kernel/sched/fair.c +0 -2
@@ -22,8 +22,6 @@*/ #include "sched.h" -#include <trace/events/sched.h> -/* * Targeted preemption latency for CPU-bound tasks: *
kernel/sched/pelt.c +0 -2
@@ -28,8 +28,6 @@#include "sched.h" #include "pelt.h" -#include <trace/events/sched.h> -/* * Approximate: * val * y^n, where y^32 ~= 0.5 (~1 scheduling period)
kernel/sched/sched.h +12 -0
@@ -75,6 +75,15 @@#include "cpupri.h" #include "cpudeadline.h" +#ifdef SCHED_CREATE_TRACE_POINTS +#define CREATE_TRACE_POINTS +#endif +#include <trace/events/sched.h> + +#ifdef SCHED_CREATE_TRACE_POINTS +#undef CREATE_TRACE_POINTS +#endif +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else
@@ -1959,6 +1968,7 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +trace_sched_update_nr_running_tp (cpu_of (rq), count, rq->nr_running);#ifdef CONFIG_SMP if (prev_nr <2 && rq->nr_running >=2 ) {
@@ -1973,6 +1983,8 @@ static inline void add_nr_running(struct rq *rq, unsigned count)static inline void sub_nr_running (struct rq *rq,unsigned count) { rq->nr_running -= count; +trace_sched_update_nr_running_tp (cpu_of (rq), -count, rq->nr_running); +/* Check if we still need preemption */ sched_update_tick_dependency (rq); }
Signature · 1 line
2.18.0
Steven Rostedt
maintainerQuoted 21 lines · On Fri, 19 Jun 2020 10:11:20 -0400 Phil Auld <[email protected]> wrote:
include/trace/events/sched.h +4
-0@@ -634,6 +634,10 @@ DECLARE_TRACE(sched_overutilized_tp, TP_PROTO(struct root_domain *rd, bool overutilized), TP_ARGS(rd, overutilized)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(int cpu, int change, unsigned int nr_running), +TP_ARGS (cpu, change, nr_running)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */ kernel/sched/core.c +0
-0
kernel/sched/core.c line 12
Quoted 7 lines
kernel/sched/core.c +3
-0@@ -6,7 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ + +#define SCHED_CREATE_TRACE_POINTS #include "sched.h" +#undef SCHED_CREATE_TRACE_POINTS
Because of the macro magic, and really try not to have trace events defined in any headers. Otherwise, we have weird defines like you are doing, and it doesn't fully protect it if a C file adds this header and defines CREATE_TRACE_POINTS first.
Quoted 24 lines
kernel/sched/core.c +0
-0#include <linux/nospec.h> @@ -21,9 +24,6 @@kernel/sched/sched.h +9
-0@@ -75,6 +75,15 @@#include "cpupri.h" #include "cpudeadline.h" +#ifdef SCHED_CREATE_TRACE_POINTS +#define CREATE_TRACE_POINTS +#endif +#include <trace/events/sched.h> + +#ifdef SCHED_CREATE_TRACE_POINTS +#undef CREATE_TRACE_POINTS +#endif +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else
kernel/sched/sched.h static inline void add_nr_running(struct rq *rq, unsigned count) · line 1971
Quoted 4 lines
kernel/sched/sched.h +1
-0@@ -1959,6 +1968,7 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +trace_sched_update_nr_running_tp (cpu_of (rq), count, rq->nr_running);
Instead of having sched.h define CREATE_TRACE_POINTS, I would have the following:
if (trace_sched_update_nr_running_tp_enabled()) {
call_trace_sched_update_nr_runnig(rq, count);
}Then in sched/core.c:
void trace_sched_update_nr_running(struct rq *rq, int count)
{
trace_sched_update_nr_running_tp(cpu_of(rq), count, rq->nr_running);
}The trace_*_enabled() above uses static branches, where the if turns to a nop (pass through) when disabled and a jmp when enabled (same logic that trace points use themselves).
Then you don't need this macro dance, and risk having another C file define CREATE_TRACE_POINTS and spend hours debugging why it suddenly broke.
-- Steve
Quoted 81 lines · On Fri, Jun 19, 2020 at 12:46:41PM -0400 Steven Rostedt wrote:
Quoted 21 lines · On Fri, 19 Jun 2020 10:11:20 -0400 Phil Auld <[email protected]> wrote:
include/trace/events/sched.h +4
-0@@ -634,6 +634,10 @@ DECLARE_TRACE(sched_overutilized_tp, TP_PROTO(struct root_domain *rd, bool overutilized), TP_ARGS(rd, overutilized)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(int cpu, int change, unsigned int nr_running), +TP_ARGS (cpu, change, nr_running)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */ kernel/sched/core.c +0
-0kernel/sched/core.c line 12
Quoted 7 lines
kernel/sched/core.c +3
-0@@ -6,7 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ + +#define SCHED_CREATE_TRACE_POINTS #include "sched.h" +#undef SCHED_CREATE_TRACE_POINTSBecause of the macro magic, and really try not to have trace events defined in any headers. Otherwise, we have weird defines like you are doing, and it doesn't fully protect it if a C file adds this header and defines
CREATE_TRACE_POINTSfirst.Quoted 24 lines
kernel/sched/core.c +0
-0#include <linux/nospec.h> @@ -21,9 +24,6 @@kernel/sched/sched.h +9
-0@@ -75,6 +75,15 @@#include "cpupri.h" #include "cpudeadline.h" +#ifdef SCHED_CREATE_TRACE_POINTS +#define CREATE_TRACE_POINTS +#endif +#include <trace/events/sched.h> + +#ifdef SCHED_CREATE_TRACE_POINTS +#undef CREATE_TRACE_POINTS +#endif +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else kernel/sched/sched.h static inline void add_nr_running(struct rq *rq, unsigned count) · line 1971
Quoted 4 lines
kernel/sched/sched.h +1
-0@@ -1959,6 +1968,7 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +trace_sched_update_nr_running_tp (cpu_of (rq), count, rq->nr_running);Instead of having
sched.hdefineCREATE_TRACE_POINTS, I would have the following:if (trace_sched_update_nr_running_tp_enabled()) { call_trace_sched_update_nr_runnig(rq, count); }Then in
sched/core.c:void
trace_sched_update_nr_running(struct rq *rq, int count){ trace_sched_update_nr_running_tp(cpu_of(rq), count, rq->nr_running); }
Quoted 8 lines
The trace_*_enabled() above uses static branches, where the if turns to a nop (pass through) when disabled and a jmp when enabled (same logic that trace points use themselves).
Then you don't need this macro dance, and risk having another C file define
CREATE_TRACE_POINTSand spend hours debugging why it suddenly broke.
Awesome, thanks Steve. I was really hoping there was a better way to do that. I try it this way.
Cheers,
Phil
Quoted 2 lines
-- Steve
Quoted 133 lines · Phil Auld
Add a bare tracepoint
trace_sched_update_nr_running_tpwhich tracks ->nr_running CPU's rq. This is used to accurately trace this data and provide a visualization of scheduler imbalances in, for example, the form of a heat map. The tracepoint is accessed by loading an external kernel module. An example module (forked from Qais' module and including the pelt related tracepoints) can be found at:https://github.com/auldp/tracepoints-helpers.git
A script to turn the trace-cmd report output into a heatmap plot can be found at:
https://github.com/jirvoz/plot-nr-running
The tracepoints are added to
add_nr_running()andsub_nr_running()which are inkernel/sched/sched.h. Sincesched.hincludestrace/events/tlb.hviammu_context.hwe had to limit whenCREATE_TRACE_POINTSis defined.Signed-off-by Phil Auld <[email protected]>CC Qais Yousef <[email protected]>CC Ingo Molnar <[email protected]>CC Peter Zijlstra <[email protected]>CC Vincent Guittot <[email protected]>--- include/trace/events/sched.h | 4 ++++ kernel/sched/core.c | 9 ++++----- kernel/sched/fair.c | 2 -- kernel/sched/pelt.c | 2 -- kernel/sched/sched.h | 12 ++++++++++++ 5 files changed, 20 insertions(+), 9 deletions(-)include/trace/events/sched.h +4
-0@@ -634,6 +634,10 @@ DECLARE_TRACE(sched_overutilized_tp, TP_PROTO(struct root_domain *rd, bool overutilized), TP_ARGS(rd, overutilized)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(int cpu, int change, unsigned int nr_running), +TP_ARGS (cpu, change, nr_running)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */ kernel/sched/core.c +4
-5@@ -6,7 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ + +#define SCHED_CREATE_TRACE_POINTS #include "sched.h" +#undef SCHED_CREATE_TRACE_POINTS#include <linux/nospec.h> @@ -21,9 +24,6 @@#include "pelt.h" -#define CREATE_TRACE_POINTS -#include <trace/events/sched.h> -/* * Export tracepoints that act as a bare tracehook (ie: have no trace event * associated with them) to allow external modules to probe them. @@ -34,6 +34,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(pelt_dl_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_irq_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_se_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_overutilized_tp); +EXPORT_TRACEPOINT_SYMBOL_GPL (sched_update_nr_running_tp); DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);@@ -7969,5 +7970,3 @@ const u32 sched_prio_to_wmult[40] = {/* 10 */ 39045157 ,49367440 ,61356676 ,76695844 ,95443717 , /* 15 */ 119304647, 148102320, 186737708, 238609294, 286331153, }; - -#undef CREATE_TRACE_POINTSkernel/sched/fair.c +0
-2@@ -22,8 +22,6 @@*/ #include "sched.h" -#include <trace/events/sched.h> -/* * Targeted preemption latency for CPU-bound tasks: * kernel/sched/pelt.c +0
-2@@ -28,8 +28,6 @@#include "sched.h" #include "pelt.h" -#include <trace/events/sched.h> -/* * Approximate: * val * y^n, where y^32 ~= 0.5 (~1 scheduling period) kernel/sched/sched.h +9
-0@@ -75,6 +75,15 @@#include "cpupri.h" #include "cpudeadline.h" +#ifdef SCHED_CREATE_TRACE_POINTS +#define CREATE_TRACE_POINTS +#endif +#include <trace/events/sched.h> + +#ifdef SCHED_CREATE_TRACE_POINTS +#undef CREATE_TRACE_POINTS +#endif +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else
kernel/sched/sched.h static inline void add_nr_running(struct rq *rq, unsigned count) · line 1971
Quoted 4 lines
kernel/sched/sched.h +1
-0@@ -1959,6 +1968,7 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +trace_sched_update_nr_running_tp (cpu_of (rq), count, rq->nr_running);
This is a very specific call site, so I guess it looks fine to pass very specific info too.
But I think we can do better by just passing struct rq and add a new helper sched_trace_rq_nr_running() (see the bottom of fair.c for a similar helper functions for tracepoints).
This will allow the user to extract, cpu, nr_running and potentially other info while only pass a single argument to the tracepoint. Potentially extending its future usefulness.
The count can be inferred by storing the last nr_running and taking the diff when a new call happens.
... cpu = sched_trace_rq_cpu(rq); nr_running = sched_trace_rq_nr_running(rq); count = last_nr_running[cpu] - nr_running; last_nr_running[cpu] = nr_running; ...
I haven't looked at BTF, but it could potentially allow us to access members of unexported structs reliably without having to export all these helper functions. It's been something I wanted to look into but no time yet.
Thanks
--
Qais Yousef
Quoted 15 lines
kernel/sched/sched.h +2
-0#ifdef CONFIG_SMP if (prev_nr <2 && rq->nr_running >=2 ) {@@ -1973,6 +1983,8 @@ static inline void add_nr_running(struct rq *rq, unsigned count)static inline void sub_nr_running (struct rq *rq,unsigned count) { rq->nr_running -= count; +trace_sched_update_nr_running_tp (cpu_of (rq), -count, rq->nr_running); +/* Check if we still need preemption */ sched_update_tick_dependency (rq); }Signature · 1 line
2.18.0
Hi Qais,
Quoted 142 lines · On Mon, Jun 22, 2020 at 01:17:47PM +0100 Qais Yousef wrote:
Quoted 133 lines · Phil Auld
Add a bare tracepoint
trace_sched_update_nr_running_tpwhich tracks ->nr_running CPU's rq. This is used to accurately trace this data and provide a visualization of scheduler imbalances in, for example, the form of a heat map. The tracepoint is accessed by loading an external kernel module. An example module (forked from Qais' module and including the pelt related tracepoints) can be found at:https://github.com/auldp/tracepoints-helpers.git
A script to turn the trace-cmd report output into a heatmap plot can be found at:
https://github.com/jirvoz/plot-nr-running
The tracepoints are added to
add_nr_running()andsub_nr_running()which are inkernel/sched/sched.h. Sincesched.hincludestrace/events/tlb.hviammu_context.hwe had to limit whenCREATE_TRACE_POINTSis defined.Signed-off-by Phil Auld <[email protected]>CC Qais Yousef <[email protected]>CC Ingo Molnar <[email protected]>CC Peter Zijlstra <[email protected]>CC Vincent Guittot <[email protected]>--- include/trace/events/sched.h | 4 ++++ kernel/sched/core.c | 9 ++++----- kernel/sched/fair.c | 2 -- kernel/sched/pelt.c | 2 -- kernel/sched/sched.h | 12 ++++++++++++ 5 files changed, 20 insertions(+), 9 deletions(-)include/trace/events/sched.h +4
-0@@ -634,6 +634,10 @@ DECLARE_TRACE(sched_overutilized_tp, TP_PROTO(struct root_domain *rd, bool overutilized), TP_ARGS(rd, overutilized)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(int cpu, int change, unsigned int nr_running), +TP_ARGS (cpu, change, nr_running)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */ kernel/sched/core.c +4
-5@@ -6,7 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ + +#define SCHED_CREATE_TRACE_POINTS #include "sched.h" +#undef SCHED_CREATE_TRACE_POINTS#include <linux/nospec.h> @@ -21,9 +24,6 @@#include "pelt.h" -#define CREATE_TRACE_POINTS -#include <trace/events/sched.h> -/* * Export tracepoints that act as a bare tracehook (ie: have no trace event * associated with them) to allow external modules to probe them. @@ -34,6 +34,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(pelt_dl_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_irq_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_se_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_overutilized_tp); +EXPORT_TRACEPOINT_SYMBOL_GPL (sched_update_nr_running_tp); DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);@@ -7969,5 +7970,3 @@ const u32 sched_prio_to_wmult[40] = {/* 10 */ 39045157 ,49367440 ,61356676 ,76695844 ,95443717 , /* 15 */ 119304647, 148102320, 186737708, 238609294, 286331153, }; - -#undef CREATE_TRACE_POINTSkernel/sched/fair.c +0
-2@@ -22,8 +22,6 @@*/ #include "sched.h" -#include <trace/events/sched.h> -/* * Targeted preemption latency for CPU-bound tasks: * kernel/sched/pelt.c +0
-2@@ -28,8 +28,6 @@#include "sched.h" #include "pelt.h" -#include <trace/events/sched.h> -/* * Approximate: * val * y^n, where y^32 ~= 0.5 (~1 scheduling period) kernel/sched/sched.h +9
-0@@ -75,6 +75,15 @@#include "cpupri.h" #include "cpudeadline.h" +#ifdef SCHED_CREATE_TRACE_POINTS +#define CREATE_TRACE_POINTS +#endif +#include <trace/events/sched.h> + +#ifdef SCHED_CREATE_TRACE_POINTS +#undef CREATE_TRACE_POINTS +#endif +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else kernel/sched/sched.h static inline void add_nr_running(struct rq *rq, unsigned count) · line 1971
Quoted 4 lines
kernel/sched/sched.h +1
-0@@ -1959,6 +1968,7 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +trace_sched_update_nr_running_tp (cpu_of (rq), count, rq->nr_running);This is a very specific call site, so I guess it looks fine to pass very specific info too.
Quoted 7 lines
But I think we can do better by just passing struct rq and add a new helper
sched_trace_rq_nr_running()(see the bottom offair.cfor a similar helper functions for tracepoints).This will allow the user to extract, cpu,
nr_runningand potentially other info while only pass a single argument to the tracepoint. Potentially extending its future usefulness.
I can certainly add a sched_trace_rq_nr_running helper and pass the *rq if you think that is really important.
I'd prefer to keep the count field though as that is the only way to tell if this is an add_nr_running or sub_nr_running from looking at a single trace event.
I could make it two different tracepoints. Would that be better? To me that seemed more complicated though. The tooling would need to look at it different events and there would be more kernel change.
Thanks,
Phil
Quoted 38 lines
The count can be inferred by storing the last
nr_runningand taking the diff when a new call happens.... cpu = sched_trace_rq_cpu(rq); nr_running = sched_trace_rq_nr_running(rq); count = last_nr_running[cpu] - nr_running; last_nr_running[cpu] = nr_running; ...I haven't looked at BTF, but it could potentially allow us to access members of unexported structs reliably without having to export all these helper functions. It's been something I wanted to look into but no time yet.
Thanks
--
Qais YousefQuoted 15 lines
kernel/sched/sched.h +2
-0#ifdef CONFIG_SMP if (prev_nr <2 && rq->nr_running >=2 ) {@@ -1973,6 +1983,8 @@ static inline void add_nr_running(struct rq *rq, unsigned count)static inline void sub_nr_running (struct rq *rq,unsigned count) { rq->nr_running -= count; +trace_sched_update_nr_running_tp (cpu_of (rq), -count, rq->nr_running); +/* Check if we still need preemption */ sched_update_tick_dependency (rq); }Signature · 1 line
2.18.0
Hi Phil
On 06/23/20 15:38, Phil Auld wrote:
[...]
Quoted 7 lines
Quoted 7 lines
This is a very specific call site, so I guess it looks fine to pass very specific info too.
But I think we can do better by just passing struct rq and add a new helper
sched_trace_rq_nr_running()(see the bottom offair.cfor a similar helper functions for tracepoints).
Quoted 6 lines
Quoted 3 lines
This will allow the user to extract, cpu,
nr_runningand potentially other info while only pass a single argument to the tracepoint. Potentially extending its future usefulness.I can certainly add a
sched_trace_rq_nr_runninghelper and pass the *rq if you think that is really important.
As I said, this is a very specific call site, so passing specific info should be fine, so not really important.
My general view on this (which is influenced by what Peter asked for when we first introduced this) is that it's better to allow a trace point to extract more signals from this specific call site by passing generic info and let the event code/module do what it wants.
But the idea behind these tracepoints is that they can evolve when they need to. So I don't think we should hung up on this if it makes things unnecessarily complex.
Quoted 4 lines
I'd prefer to keep the count field though as that is the only way to tell if this is an
add_nr_runningorsub_nr_runningfrom looking at a single trace event.
Passing the count field is fine by me...
Quoted 4 lines
I could make it two different tracepoints. Would that be better? To me that seemed more complicated though. The tooling would need to look at it different events and there would be more kernel change.
... but splitting the tracepoint doesn't look pretty.
If passing the rq and the count is enough for you, I'd vote this is better. If not, then I won't insist into twisting things too much for the sake of it.
Thanks
--
Qais Yousef
Add a bare tracepoint trace_sched_update_nr_running_tp which tracks ->nr_running CPU's rq. This is used to accurately trace this data and provide a visualization of scheduler imbalances in, for example, the form of a heat map. The tracepoint is accessed by loading an external kernel module. An example module (forked from Qais' module and including the pelt related tracepoints) can be found at:
https://github.com/auldp/tracepoints-helpers.git
A script to turn the trace-cmd report output into a heatmap plot can be found at:
https://github.com/jirvoz/plot-nr-running
The tracepoints are added to add_nr_running() and sub_nr_running() which are in kernel/sched/sched.h. In order to avoid CREATE_TRACE_POINTS in the header a wrapper call is used and the trace/events/sched.h include is moved before sched.h in kernel/sched/core.
---
V2: Fix use of tracepoint in header from Steven. Pass rq* and use helper to get nr_running field, from Qais.
include/linux/sched.h | 1 + include/trace/events/sched.h | 4 ++++ kernel/sched/core.c | 13 +++++++++---- kernel/sched/fair.c | 8 ++++++-- kernel/sched/pelt.c | 2 -- kernel/sched/sched.h | 10 ++++++++++ 6 files changed, 30 insertions(+), 8 deletions(-)
include/linux/sched.h +1 -0
@@ -2015,6 +2015,7 @@ const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);const struct sched_avg *sched_trace_rq_avg_irq (struct rq *rq);int sched_trace_rq_cpu (struct rq *rq); +int sched_trace_rq_nr_running (struct rq *rq);const struct cpumask *sched_trace_rd_span (struct root_domain *rd);
include/trace/events/sched.h +4 -0
@@ -634,6 +634,10 @@ DECLARE_TRACE(sched_overutilized_tp, TP_PROTO(struct root_domain *rd, bool overutilized), TP_ARGS(rd, overutilized)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(struct rq *rq, int change), +TP_ARGS (rq, change)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */
kernel/sched/core.c +9 -4
@@ -6,6 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ +#define CREATE_TRACE_POINTS +#include <trace/events/sched.h> +#undef CREATE_TRACE_POINTS +#include "sched.h" #include <linux/nospec.h>
@@ -21,9 +25,6 @@#include "pelt.h" -#define CREATE_TRACE_POINTS -#include <trace/events/sched.h> -/* * Export tracepoints that act as a bare tracehook (ie: have no trace event * associated with them) to allow external modules to probe them.
@@ -34,6 +35,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(pelt_dl_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_irq_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (pelt_se_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_overutilized_tp); +EXPORT_TRACEPOINT_SYMBOL_GPL (sched_update_nr_running_tp); DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
@@ -7970,4 +7972,7 @@ const u32 sched_prio_to_wmult[40] = { /* 15 */ 119304647, 148102320, 186737708, 238609294, 286331153, }; -#undef CREATE_TRACE_POINTS +void call_trace_sched_update_nr_running (struct rq *rq,int count) +{ +trace_sched_update_nr_running_tp (rq, count); +}
kernel/sched/fair.c +6 -2
@@ -22,8 +22,6 @@*/ #include "sched.h" -#include <trace/events/sched.h> -/* * Targeted preemption latency for CPU-bound tasks: *
@@ -11293,3 +11291,9 @@ const struct cpumask *sched_trace_rd_span(struct root_domain *rd)#endif }EXPORT_SYMBOL_GPL (sched_trace_rd_span); + +int sched_trace_rq_nr_running (struct rq *rq) +{ +return rq ? rq->nr_running :-1 ; +} +EXPORT_SYMBOL_GPL (sched_trace_rq_nr_running);
kernel/sched/pelt.c +0 -2
@@ -28,8 +28,6 @@#include "sched.h" #include "pelt.h" -#include <trace/events/sched.h> -/* * Approximate: * val * y^n, where y^32 ~= 0.5 (~1 scheduling period)
kernel/sched/sched.h +10 -0
@@ -75,6 +75,8 @@#include "cpupri.h" #include "cpudeadline.h" +#include <trace/events/sched.h> +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else
@@ -96,6 +98,7 @@ extern atomic_long_t calc_load_tasks;extern void calc_global_load_tick (struct rq *this_rq);extern long calc_load_fold_active (struct rq *this_rq,long adjust); +extern void call_trace_sched_update_nr_running (struct rq *rq,int count);/* * Helpers for converting nanosecond timing to jiffy resolution */
@@ -1959,6 +1962,9 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +if (trace_sched_update_nr_running_tp_enabled ()) { +call_trace_sched_update_nr_running (rq, count); + }#ifdef CONFIG_SMP if (prev_nr <2 && rq->nr_running >=2 ) {
@@ -1973,6 +1979,10 @@ static inline void add_nr_running(struct rq *rq, unsigned count)static inline void sub_nr_running (struct rq *rq,unsigned count) { rq->nr_running -= count; +if (trace_sched_update_nr_running_tp_enabled ()) { +call_trace_sched_update_nr_running (rq, count); + } +/* Check if we still need preemption */ sched_update_tick_dependency (rq); }
Signature · 4 lines
2.18.0 --
Qais Yousef
Reviewed-byQuoted 19 lines · Phil Auld
Add a bare tracepoint
trace_sched_update_nr_running_tpwhich tracks ->nr_running CPU's rq. This is used to accurately trace this data and provide a visualization of scheduler imbalances in, for example, the form of a heat map. The tracepoint is accessed by loading an external kernel module. An example module (forked from Qais' module and including the pelt related tracepoints) can be found at:https://github.com/auldp/tracepoints-helpers.git
A script to turn the trace-cmd report output into a heatmap plot can be found at:
https://github.com/jirvoz/plot-nr-running
The tracepoints are added to
add_nr_running()andsub_nr_running()which are inkernel/sched/sched.h. In order to avoidCREATE_TRACE_POINTSin the header a wrapper call is used and thetrace/events/sched.hinclude is moved beforesched.hinkernel/sched/core.
Quoted 8 lines
Signed-off-by Phil Auld <[email protected]>CC Qais Yousef <[email protected]>CC Ingo Molnar <[email protected]>CC Peter Zijlstra <[email protected]>CC Vincent Guittot <[email protected]>CC Steven Rostedt <[email protected]>---
LGTM.
Thanks
--
Qais Yousef
[tip: sched/core] sched: Add a tracepoint to track rq->nr_running
In reply to Phil AuldThe following commit has been merged into the sched/core branch of tip:
The following commit has been merged into the sched/core branch of tip:
Commit-ID: 9d246053a69196c7c27068870e9b4b66ac536f68 Gitweb: https://git.kernel.org/tip/9d246053a69196c7c27068870e9b4b66ac536f68 Author: Phil Auld <[email protected]> AuthorDate: Mon, 29 Jun 2020 15:23:03 -04:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Wed, 08 Jul 2020 11:39:02 +02:00
sched: Add a tracepoint to track rq->nr_running
Add a bare tracepoint trace_sched_update_nr_running_tp which tracks ->nr_running CPU's rq. This is used to accurately trace this data and provide a visualization of scheduler imbalances in, for example, the form of a heat map. The tracepoint is accessed by loading an external kernel module. An example module (forked from Qais' module and including the pelt related tracepoints) can be found at:
https://github.com/auldp/tracepoints-helpers.git
A script to turn the trace-cmd report output into a heatmap plot can be found at:
https://github.com/jirvoz/plot-nr-running
The tracepoints are added to add_nr_running() and sub_nr_running() which are in kernel/sched/sched.h. In order to avoid CREATE_TRACE_POINTS in the header a wrapper call is used and the trace/events/sched.h include is moved before sched.h in kernel/sched/core.
--- include/linux/sched.h | 1 + include/trace/events/sched.h | 4 ++++ kernel/sched/core.c | 13 +++++++++---- kernel/sched/fair.c | 8 ++++++-- kernel/sched/pelt.c | 2 -- kernel/sched/sched.h | 10 ++++++++++ 6 files changed, 30 insertions(+), 8 deletions(-)
include/linux/sched.h +1 -0
@@ -2044,6 +2044,7 @@ const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);const struct sched_avg *sched_trace_rq_avg_irq (struct rq *rq);int sched_trace_rq_cpu (struct rq *rq); +int sched_trace_rq_nr_running (struct rq *rq);const struct cpumask *sched_trace_rd_span (struct root_domain *rd);
include/trace/events/sched.h +4 -0
@@ -642,6 +642,10 @@ DECLARE_TRACE(sched_util_est_se_tp, TP_PROTO(struct sched_entity *se), TP_ARGS(se)); +DECLARE_TRACE (sched_update_nr_running_tp, + TP_PROTO(struct rq *rq, int change), +TP_ARGS (rq, change)); +#endif /* _TRACE_SCHED_H */ /* This part must be outside protection */
kernel/sched/core.c +9 -4
@@ -6,6 +6,10 @@* * Copyright (C) 1991-2002 Linus Torvalds */ +#define CREATE_TRACE_POINTS +#include <trace/events/sched.h> +#undef CREATE_TRACE_POINTS +#include "sched.h" #include <linux/nospec.h>
@@ -23,9 +27,6 @@#include "pelt.h" #include "smp.h" -#define CREATE_TRACE_POINTS -#include <trace/events/sched.h> -/* * Export tracepoints that act as a bare tracehook (ie: have no trace event * associated with them) to allow external modules to probe them.
@@ -38,6 +39,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(pelt_se_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_overutilized_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_util_est_cfs_tp);EXPORT_TRACEPOINT_SYMBOL_GPL (sched_util_est_se_tp); +EXPORT_TRACEPOINT_SYMBOL_GPL (sched_update_nr_running_tp); DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
@@ -8195,4 +8197,7 @@ const u32 sched_prio_to_wmult[40] = { /* 15 */ 119304647, 148102320, 186737708, 238609294, 286331153, }; -#undef CREATE_TRACE_POINTS +void call_trace_sched_update_nr_running (struct rq *rq,int count) +{ +trace_sched_update_nr_running_tp (rq, count); +}
kernel/sched/fair.c +6 -2
@@ -22,8 +22,6 @@*/ #include "sched.h" -#include <trace/events/sched.h> -/* * Targeted preemption latency for CPU-bound tasks: *
@@ -11296,3 +11294,9 @@ const struct cpumask *sched_trace_rd_span(struct root_domain *rd)#endif }EXPORT_SYMBOL_GPL (sched_trace_rd_span); + +int sched_trace_rq_nr_running (struct rq *rq) +{ +return rq ? rq->nr_running :-1 ; +} +EXPORT_SYMBOL_GPL (sched_trace_rq_nr_running);
kernel/sched/pelt.c +0 -2
@@ -28,8 +28,6 @@#include "sched.h" #include "pelt.h" -#include <trace/events/sched.h> -/* * Approximate: * val * y^n, where y^32 ~= 0.5 (~1 scheduling period)
kernel/sched/sched.h +10 -0
@@ -76,6 +76,8 @@#include "cpupri.h" #include "cpudeadline.h" +#include <trace/events/sched.h> +#ifdef CONFIG_SCHED_DEBUG # define SCHED_WARN_ON (x) WARN_ONCE(x, #x)#else
@@ -97,6 +99,7 @@ extern atomic_long_t calc_load_tasks;extern void calc_global_load_tick (struct rq *this_rq);extern long calc_load_fold_active (struct rq *this_rq,long adjust); +extern void call_trace_sched_update_nr_running (struct rq *rq,int count);/* * Helpers for converting nanosecond timing to jiffy resolution */
@@ -1973,6 +1976,9 @@ static inline void add_nr_running(struct rq *rq, unsigned count)unsigned prev_nr = rq->nr_running; rq->nr_running = prev_nr + count; +if (trace_sched_update_nr_running_tp_enabled ()) { +call_trace_sched_update_nr_running (rq, count); + }#ifdef CONFIG_SMP if (prev_nr <2 && rq->nr_running >=2 ) {
@@ -1987,6 +1993,10 @@ static inline void add_nr_running(struct rq *rq, unsigned count)static inline void sub_nr_running (struct rq *rq,unsigned count) { rq->nr_running -= count; +if (trace_sched_update_nr_running_tp_enabled ()) { +call_trace_sched_update_nr_running (rq, count); + } +/* Check if we still need preemption */ sched_update_tick_dependency (rq); }