[PATCH] trace/events/sched: Remove unit on printing vruntime value

[PATCH] trace/events/sched: Remove unit on printing vruntime value

[Gildong Hong]

vruntime is not actual time so remove unit on printing

Signed-off-by: Gildong Hong <roaming2man@gmail.com>
---
 include/trace/events/sched.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
index fbb99a61f714..26ef85a8d3cd 100644
--- a/include/trace/events/sched.h
+++ b/include/trace/events/sched.h
@@ -511,7 +511,7 @@ DECLARE_EVENT_CLASS(sched_stat_runtime,
 		__entry->vruntime	= vruntime;
 	),
 
-	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
+	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu",
 			__entry->comm, __entry->pid,
 			(unsigned long long)__entry->runtime,
 			(unsigned long long)__entry->vruntime)
-- 
2.25.1

Re: [PATCH] trace/events/sched: Remove unit on printing vruntime value

[Steven Rostedt]

On Thu, 16 Mar 2023 20:56:55 +0900
Gildong Hong <roaming2man@gmail.com> wrote:

> vruntime is not actual time so remove unit on printing
> 
> Signed-off-by: Gildong Hong <roaming2man@gmail.com>
> ---
>  include/trace/events/sched.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
> index fbb99a61f714..26ef85a8d3cd 100644
> --- a/include/trace/events/sched.h
> +++ b/include/trace/events/sched.h
> @@ -511,7 +511,7 @@ DECLARE_EVENT_CLASS(sched_stat_runtime,
>  		__entry->vruntime	= vruntime;
>  	),
>  
> -	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
> +	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu",

What about calling it "[vns]" for "virtual nanosecond" ?

;-)

-- Steve

>  			__entry->comm, __entry->pid,
>  			(unsigned long long)__entry->runtime,
>  			(unsigned long long)__entry->vruntime)

Re: [PATCH] trace/events/sched: Remove unit on printing vruntime value

[Gildong Hong]

On Thu, Mar 16, 2023 at 10:11:37AM -0400, Steven Rostedt wrote:
> On Thu, 16 Mar 2023 20:56:55 +0900
> Gildong Hong <roaming2man@gmail.com> wrote:
> 
> > vruntime is not actual time so remove unit on printing
> > 
> > Signed-off-by: Gildong Hong <roaming2man@gmail.com>
> > ---
> >  include/trace/events/sched.h | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
> > index fbb99a61f714..26ef85a8d3cd 100644
> > --- a/include/trace/events/sched.h
> > +++ b/include/trace/events/sched.h
> > @@ -511,7 +511,7 @@ DECLARE_EVENT_CLASS(sched_stat_runtime,
> >  		__entry->vruntime	= vruntime;
> >  	),
> >  
> > -	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
> > +	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu",
> 
> What about calling it "[vns]" for "virtual nanosecond" ?
> 
> ;-)
> 
> -- Steve
> 
> >  			__entry->comm, __entry->pid,
> >  			(unsigned long long)__entry->runtime,
> >  			(unsigned long long)__entry->vruntime)
> 

Dear Steve

"vns" sounds nice.
But some people may hard to understand the meaning.
"ns" is a global standard time unit where "vns" is not.
So I still prefer to remove the unit.

Thanks for your review.
Gildong

Re: [PATCH] trace/events/sched: Remove unit on printing vruntime value

[Steven Rostedt]

On Sun, 19 Mar 2023 12:05:20 +0900
Gildong Hong <roaming2man@gmail.com> wrote:

> > > -	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
> > > +	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu",  

> 
> "vns" sounds nice.
> But some people may hard to understand the meaning.
> "ns" is a global standard time unit where "vns" is not.
> So I still prefer to remove the unit.

Really? You have "runtime" (which most people understand) and labeled
"ns", but then you have "vruntime" which some people may find hard to
understand the meaning of. If they do understand the meaning of
"vruntime" then they should have the means to understand "vns".

Otherwise, no label is meaningless too!

-- Steve

Re: [PATCH] trace/events/sched: Remove unit on printing vruntime value

[Gildong Hong]

On Sun, Mar 19, 2023 at 02:07:38PM -0400, Steven Rostedt wrote:
> On Sun, 19 Mar 2023 12:05:20 +0900
> Gildong Hong <roaming2man@gmail.com> wrote:
> 
> > > > -	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
> > > > +	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu",  
> 
> > 
> > "vns" sounds nice.
> > But some people may hard to understand the meaning.
> > "ns" is a global standard time unit where "vns" is not.
> > So I still prefer to remove the unit.
> 
> Really? You have "runtime" (which most people understand) and labeled
> "ns", but then you have "vruntime" which some people may find hard to
> understand the meaning of. If they do understand the meaning of
> "vruntime" then they should have the means to understand "vns".
> 
> Otherwise, no label is meaningless too!
> 
> -- Steve

I agree with your opinion.
People can understand "vns" if they understand "vruntime".
Followings are the modified patch.
I add your name on "Suggested-by"
I or you could remove the line if you mind it.

Thanks,
Gildong

------------
From 8a8d1f08a2f217151f26f04795d6c1483eb1c48f Mon Sep 17 00:00:00 2001
From: Gildong Hong <roaming2man@gmail.com>
Date: Thu, 16 Mar 2023 20:05:48 +0900
Subject: [PATCH] trace/events/sched: Change unit on printing vruntime value

vruntime is not actual time so change unit on printing
from "ns" to "vns" which denotes "virtual nanosecond"

Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Gildong Hong <roaming2man@gmail.com>
---
 include/trace/events/sched.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
index fbb99a61f714..993986c6c3f6 100644
--- a/include/trace/events/sched.h
+++ b/include/trace/events/sched.h
@@ -511,7 +511,7 @@ DECLARE_EVENT_CLASS(sched_stat_runtime,
 		__entry->vruntime	= vruntime;
 	),
 
-	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
+	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [vns]",
 			__entry->comm, __entry->pid,
 			(unsigned long long)__entry->runtime,
 			(unsigned long long)__entry->vruntime)
-- 
2.25.1

Re: [PATCH] trace/events/sched: Remove unit on printing vruntime value

[Peter Zijlstra]

On Sun, Mar 19, 2023 at 02:07:38PM -0400, Steven Rostedt wrote:
> On Sun, 19 Mar 2023 12:05:20 +0900
> Gildong Hong <roaming2man@gmail.com> wrote:
> 
> > > > -	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu [ns]",
> > > > +	TP_printk("comm=%s pid=%d runtime=%Lu [ns] vruntime=%Lu",  
> 
> > 
> > "vns" sounds nice.
> > But some people may hard to understand the meaning.
> > "ns" is a global standard time unit where "vns" is not.
> > So I still prefer to remove the unit.
> 
> Really? You have "runtime" (which most people understand) and labeled
> "ns", but then you have "vruntime" which some people may find hard to
> understand the meaning of. If they do understand the meaning of
> "vruntime" then they should have the means to understand "vns".
> 
> Otherwise, no label is meaningless too!

I'm with Steve, if you're down to consuming these here numbers, you'd
better know wth you're doing.

Re: [PATCH] trace/events/sched: Remove unit on printing vruntime value

[Steven Rostedt]

On Mon, 20 Mar 2023 20:41:39 +0900
Gildong Hong <roaming2man@gmail.com> wrote:

> I agree with your opinion.
> People can understand "vns" if they understand "vruntime".
> Followings are the modified patch.
> I add your name on "Suggested-by"
> I or you could remove the line if you mind it.

I'm good.

Thanks,

-- Steve

> 
> Thanks,
> Gildong
> 
> ------------
> >From 8a8d1f08a2f217151f26f04795d6c1483eb1c48f Mon Sep 17 00:00:00 2001  
> From: Gildong Hong <roaming2man@gmail.com>
> Date: Thu, 16 Mar 2023 20:05:48 +0900
> Subject: [PATCH] trace/events/sched: Change unit on printing vruntime value
> 
> vruntime is not actual time so change unit on printing
> from "ns" to "vns" which denotes "virtual nanosecond"
> 
> Suggested-by: Steven Rostedt <rostedt@goodmis.org>
> Signed-off-by: Gildong Hong <roaming2man@gmail.com>