2009-04-06 14:03:18

by Geunsik Lim

[permalink] [raw]
Subject: [PATCH] Update old Mini-HOWTO for ftrace

Dear Steven,

The author of ftrace(Internal kernel Function Tracer) feature
renamed the ftrace tracer
to function to avoid further confusion between the ftrace
infrastructure and the
function tracer on Oct-20, 2008.

But, Update of Mini-HOWTO document are still not working
.
Most of the developers often utilize "/debug/tracing/README" file
as Mini-HOWTO.
They will confuse definition and usage about ftrace between
README(ftrace) file
and available_tracers(function) file after linux-2.6.27 like me.
And, I want you to append "chrt -f 5 sleep 1" command as simple
example for latency.

Signed-off-by: GeunSik Lim <[email protected]>
Signed-off-by: Steven Rostedt <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
---
kernel/trace/trace.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index a0174a4..15f0eec 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -2358,7 +2358,7 @@ static const char readme_msg[] =
"# mkdir /debug\n"
"# mount -t debugfs nodev /debug\n\n"
"# cat /debug/tracing/available_tracers\n"
- "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
+ "wakeup preemptirqsoff preemptoff irqsoff function
sched_switch none\n\n"
"# cat /debug/tracing/current_tracer\n"
"none\n"
"# echo sched_switch > /debug/tracing/current_tracer\n"
@@ -2368,6 +2368,7 @@ static const char readme_msg[] =
"noprint-parent nosym-offset nosym-addr noverbose\n"
"# echo print-parent > /debug/tracing/trace_options\n"
"# echo 1 > /debug/tracing/tracing_enabled\n"
+ "# chrt -f 5 sleep 1\n"
"# cat /debug/tracing/trace > /tmp/trace.txt\n"
"echo 0 > /debug/tracing/tracing_enabled\n"
;



--
Regards,
GeunSik Lim


Attachments:
update-minihowto-ftrace.patch (1.93 kB)

2009-04-06 14:17:55

by Geunsik Lim

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace

I attached patch file again because no-operation is changed
from "none"word to "nop" word.

for example,
- "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
+ "wakeup preemptirqsoff preemptoff irqsoff function sched_switch nop\n\n"

./trace_nop.c:struct tracer nop_trace __read_mostly =
.name = "nop",
.init = nop_trace_init,
.reset = nop_trace_reset,
.selftest = trace_selftest_startup_nop,
.flags = &nop_flags,
.set_flag = nop_set_flag



2009/4/6 GeunSik Lim <[email protected]>:
>   Dear Steven,
>
>   The author of ftrace(Internal kernel Function Tracer) feature
> renamed the ftrace tracer
>    to function to avoid further confusion between the ftrace
> infrastructure and the
>    function tracer on Oct-20, 2008.
>
>    But, Update of Mini-HOWTO document are still not working
>    .
>    Most of the developers often utilize "/debug/tracing/README" file
> as Mini-HOWTO.
>    They will confuse definition and usage about ftrace  between
> README(ftrace) file
>    and available_tracers(function) file after linux-2.6.27 like me.
>    And, I want you to append "chrt -f 5 sleep 1" command as simple
> example for latency.
>
>    Signed-off-by: GeunSik Lim <[email protected]>
>    Signed-off-by: Steven Rostedt <[email protected]>
>    Signed-off-by: Ingo Molnar <[email protected]>
> ---
>  kernel/trace/trace.c |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index a0174a4..15f0eec 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -2358,7 +2358,7 @@ static const char readme_msg[] =
>        "# mkdir /debug\n"
>        "# mount -t debugfs nodev /debug\n\n"
>        "# cat /debug/tracing/available_tracers\n"
> -       "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
> +       "wakeup preemptirqsoff preemptoff irqsoff function
> sched_switch none\n\n"
>        "# cat /debug/tracing/current_tracer\n"
>        "none\n"
>        "# echo sched_switch > /debug/tracing/current_tracer\n"
> @@ -2368,6 +2368,7 @@ static const char readme_msg[] =
>        "noprint-parent nosym-offset nosym-addr noverbose\n"
>        "# echo print-parent > /debug/tracing/trace_options\n"
>        "# echo 1 > /debug/tracing/tracing_enabled\n"
> +       "# chrt -f 5 sleep 1\n"
>        "# cat /debug/tracing/trace > /tmp/trace.txt\n"
>        "echo 0 > /debug/tracing/tracing_enabled\n"
>  ;
>
>
>
> --
> Regards,
> GeunSik Lim
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>



--
Regards,
GeunSik Lim


Attachments:
update-minihowto-ftrace-nop.patch (1.93 kB)

2009-04-06 23:10:06

by Geunsik Lim

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace

Sorry,
I mistake because I don't understand how to submit patch exactly.
I just appended below "singned-off" contents for convenience.
Sorry again.

Signed-off-by: GeunSik Lim <[email protected]>
Signed-off-by: Steven Rostedt <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>

2009/4/6 GeunSik Lim <[email protected]>:
> I attached patch file again because no-operation is changed
> from "none"word to "nop" word.
>
> for example,
> -       "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
> +       "wakeup preemptirqsoff preemptoff irqsoff function sched_switch nop\n\n"
>
> ./trace_nop.c:struct tracer nop_trace __read_mostly =
>        .name   = "nop",
>        .init           = nop_trace_init,
>        .reset  = nop_trace_reset,
>        .selftest       = trace_selftest_startup_nop,
>        .flags  = &nop_flags,
>        .set_flag       = nop_set_flag
>
>
>
> 2009/4/6 GeunSik Lim <[email protected]>:
>>   Dear Steven,
>>
>>   The author of ftrace(Internal kernel Function Tracer) feature
>> renamed the ftrace tracer
>>    to function to avoid further confusion between the ftrace
>> infrastructure and the
>>    function tracer on Oct-20, 2008.
>>
>>    But, Update of Mini-HOWTO document are still not working
>>    .
>>    Most of the developers often utilize "/debug/tracing/README" file
>> as Mini-HOWTO.
>>    They will confuse definition and usage about ftrace  between
>> README(ftrace) file
>>    and available_tracers(function) file after linux-2.6.27 like me.
>>    And, I want you to append "chrt -f 5 sleep 1" command as simple
>> example for latency.
>>
>>    Signed-off-by: GeunSik Lim <[email protected]>
>>    Signed-off-by: Steven Rostedt <[email protected]>
>>    Signed-off-by: Ingo Molnar <[email protected]>
>> ---
>>  kernel/trace/trace.c |    3 ++-
>>  1 files changed, 2 insertions(+), 1 deletions(-)
>>
>> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
>> index a0174a4..15f0eec 100644
>> --- a/kernel/trace/trace.c
>> +++ b/kernel/trace/trace.c
>> @@ -2358,7 +2358,7 @@ static const char readme_msg[] =
>>        "# mkdir /debug\n"
>>        "# mount -t debugfs nodev /debug\n\n"
>>        "# cat /debug/tracing/available_tracers\n"
>> -       "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
>> +       "wakeup preemptirqsoff preemptoff irqsoff function
>> sched_switch none\n\n"
>>        "# cat /debug/tracing/current_tracer\n"
>>        "none\n"
>>        "# echo sched_switch > /debug/tracing/current_tracer\n"
>> @@ -2368,6 +2368,7 @@ static const char readme_msg[] =
>>        "noprint-parent nosym-offset nosym-addr noverbose\n"
>>        "# echo print-parent > /debug/tracing/trace_options\n"
>>        "# echo 1 > /debug/tracing/tracing_enabled\n"
>> +       "# chrt -f 5 sleep 1\n"
>>        "# cat /debug/tracing/trace > /tmp/trace.txt\n"
>>        "echo 0 > /debug/tracing/tracing_enabled\n"
>>  ;
>>
>>
>>
>> --
>> Regards,
>> GeunSik Lim
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to [email protected]
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at  http://www.tux.org/lkml/
>>
>
>
>
> --
> Regards,
> GeunSik Lim
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>



--
Regards,
GeunSik Lim

2009-04-06 23:32:47

by Thomas Gleixner

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace

GeunSik,

On Tue, 7 Apr 2009, GeunSik Lim wrote:

> Sorry,
> I mistake because I don't understand how to submit patch exactly.

To submit a patch you need to have:

1) useful subject line

[PATCH] subsystem: what does the patch do

subsystem in your care is: ftrace

2) useful patch description

explains what the patch does. Which bug/problem it fixes or which
enhancement it provides

Look at the git log. Each commit has an (more or lesss) usefull
description, but you get the idea. If you see a confusing or
cryptic description you need way more time to figure out what the
patch does or what it is supposed to do.

3) Your Signed-off-by

Signed-off-by is the certificate of origin as described in

Documentation/SubmittingPatches

You can only add your own Signed-off-by. You never can add the
Signed-off-by of some one else.

> I just appended below "singned-off" contents for convenience.
> Sorry again.
>
> Signed-off-by: GeunSik Lim <[email protected]>
> Signed-off-by: Steven Rostedt <[email protected]>
> Signed-off-by: Ingo Molnar <[email protected]>

That's how it might look later on. It would say:

GeunSik Lim wrote the patch
Steven Rostedt picked it up from your mail and sent it to Ingo
Ingo Molnar received it from Steven's mail and applied it

There is no way for you to forsee how your patch will be picked up and
propagated. It might go via Andrew Morton or directly to Linus. Ok ?

So you only add your own Signed-off-by.

Hope that helps. Thanks,

tglx

2009-04-07 00:37:52

by Steven Rostedt

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace


On Tue, 2009-04-07 at 01:32 +0200, Thomas Gleixner wrote:
> GeunSik,
>
> On Tue, 7 Apr 2009, GeunSik Lim wrote:
>
> > Sorry,
> > I mistake because I don't understand how to submit patch exactly.
>
> To submit a patch you need to have:
>
> 1) useful subject line
>
> [PATCH] subsystem: what does the patch do
>
> subsystem in your care is: ftrace
>
> 2) useful patch description
>
> explains what the patch does. Which bug/problem it fixes or which
> enhancement it provides
>
> Look at the git log. Each commit has an (more or lesss) usefull
> description, but you get the idea. If you see a confusing or
> cryptic description you need way more time to figure out what the
> patch does or what it is supposed to do.
>
> 3) Your Signed-off-by
>
> Signed-off-by is the certificate of origin as described in
>
> Documentation/SubmittingPatches
>
> You can only add your own Signed-off-by. You never can add the
> Signed-off-by of some one else.
>
> > I just appended below "singned-off" contents for convenience.
> > Sorry again.
> >
> > Signed-off-by: GeunSik Lim <[email protected]>
> > Signed-off-by: Steven Rostedt <[email protected]>
> > Signed-off-by: Ingo Molnar <[email protected]>
>
> That's how it might look later on. It would say:
>
> GeunSik Lim wrote the patch
> Steven Rostedt picked it up from your mail and sent it to Ingo

Please send to my [email protected] account. From now on I'm using
that as my Signed-off-by and keep Red Hat as the author. I'm about to
travel and I do not read my Red Hat email while I'm out.

I'll pick this patch up now before I leave.

Thanks,

-- Steve

> Ingo Molnar received it from Steven's mail and applied it
>
> There is no way for you to forsee how your patch will be picked up and
> propagated. It might go via Andrew Morton or directly to Linus. Ok ?
>
> So you only add your own Signed-off-by.
>
> Hope that helps. Thanks,
>
> tglx

2009-04-07 01:25:41

by Minchan Kim

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace

One more thing.

Please, Not attach your patch but inline one.
Many people want to add their comments to line-by-line.

Tip -
If you use gmail with web, your patch is mangled by web server.
That will hurt maintainers.

Documentation/email-clients.txt can help you. :)

On Tue, Apr 7, 2009 at 8:32 AM, Thomas Gleixner <[email protected]> wrote:
> GeunSik,
>
> On Tue, 7 Apr 2009, GeunSik Lim wrote:
>
>> Sorry,
>> I mistake because I don't understand how to submit patch exactly.
>
> To submit a patch you need to have:
>
> 1) useful subject line
>
>   [PATCH] subsystem: what does the patch do
>
>   subsystem in your care is: ftrace
>
> 2) useful patch description
>
>   explains what the patch does. Which bug/problem it fixes or which
>   enhancement it provides
>
>   Look at the git log. Each commit has an (more or lesss) usefull
>   description, but you get the idea. If you see a confusing or
>   cryptic description you need way more time to figure out what the
>   patch does or what it is supposed to do.
>
> 3) Your Signed-off-by
>
>   Signed-off-by is the certificate of origin as described in
>
>   Documentation/SubmittingPatches
>
>   You can only add your own Signed-off-by. You never can add the
>   Signed-off-by of some one else.
>
>> I just appended below "singned-off" contents for convenience.
>> Sorry again.
>>
>>    Signed-off-by: GeunSik Lim <[email protected]>
>>    Signed-off-by: Steven Rostedt <[email protected]>
>>    Signed-off-by: Ingo Molnar <[email protected]>
>
> That's how it might look later on. It would say:
>
>       GeunSik Lim       wrote the patch
>       Steven Rostedt    picked it up from your mail and sent it to Ingo
>       Ingo Molnar       received it from Steven's mail and applied it
>
> There is no way for you to forsee how your patch will be picked up and
> propagated. It might go via Andrew Morton or directly to Linus. Ok ?
>
> So you only add your own Signed-off-by.
>
> Hope that helps. Thanks,
>
>     tglx
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>



--
Kinds regards,
Minchan Kim

2009-04-07 01:27:37

by Steven Rostedt

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace


On Mon, 2009-04-06 at 23:17 +0900, GeunSik Lim wrote:
> I attached patch file again because no-operation is changed
> from "none"word to "nop" word.
>
> for example,
> - "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
> + "wakeup preemptirqsoff preemptoff irqsoff function sched_switch nop\n\n"
>
> ./trace_nop.c:struct tracer nop_trace __read_mostly =
> .name = "nop",
> .init = nop_trace_init,
> .reset = nop_trace_reset,
> .selftest = trace_selftest_startup_nop,
> .flags = &nop_flags,
> .set_flag = nop_set_flag
>
>
>
> 2009/4/6 GeunSik Lim <[email protected]>:
> > Dear Steven,
> >
> > The author of ftrace(Internal kernel Function Tracer) feature
> > renamed the ftrace tracer
> > to function to avoid further confusion between the ftrace
> > infrastructure and the
> > function tracer on Oct-20, 2008.
> >
> > But, Update of Mini-HOWTO document are still not working
> > .
> > Most of the developers often utilize "/debug/tracing/README" file
> > as Mini-HOWTO.
> > They will confuse definition and usage about ftrace between
> > README(ftrace) file
> > and available_tracers(function) file after linux-2.6.27 like me.
> > And, I want you to append "chrt -f 5 sleep 1" command as simple
> > example for latency.
> >
> > Signed-off-by: GeunSik Lim <[email protected]>
> > Signed-off-by: Steven Rostedt <[email protected]>
> > Signed-off-by: Ingo Molnar <[email protected]>
> > ---
> > kernel/trace/trace.c | 3 ++-
> > 1 files changed, 2 insertions(+), 1 deletions(-)
> >
> > diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> > index a0174a4..15f0eec 100644
> > --- a/kernel/trace/trace.c
> > +++ b/kernel/trace/trace.c
> > @@ -2358,7 +2358,7 @@ static const char readme_msg[] =
> > "# mkdir /debug\n"
> > "# mount -t debugfs nodev /debug\n\n"
> > "# cat /debug/tracing/available_tracers\n"
> > - "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
> > + "wakeup preemptirqsoff preemptoff irqsoff function
> > sched_switch none\n\n"
> > "# cat /debug/tracing/current_tracer\n"
> > "none\n"
> > "# echo sched_switch > /debug/tracing/current_tracer\n"
> > @@ -2368,6 +2368,7 @@ static const char readme_msg[] =
> > "noprint-parent nosym-offset nosym-addr noverbose\n"
> > "# echo print-parent > /debug/tracing/trace_options\n"
> > "# echo 1 > /debug/tracing/tracing_enabled\n"
> > + "# chrt -f 5 sleep 1\n"


What's the chrt for?

-- Steve

2009-04-07 01:47:20

by Steven Rostedt

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace


On Mon, 2009-04-06 at 23:02 +0900, GeunSik Lim wrote:
> Dear Steven,
>
> The author of ftrace(Internal kernel Function Tracer) feature
> renamed the ftrace tracer
> to function to avoid further confusion between the ftrace
> infrastructure and the
> function tracer on Oct-20, 2008.
>
> But, Update of Mini-HOWTO document are still not working
> .
> Most of the developers often utilize "/debug/tracing/README" file
> as Mini-HOWTO.
> They will confuse definition and usage about ftrace between
> README(ftrace) file
> and available_tracers(function) file after linux-2.6.27 like me.
> And, I want you to append "chrt -f 5 sleep 1" command as simple
> example for latency.
>
> Signed-off-by: GeunSik Lim <[email protected]>
> Signed-off-by: Steven Rostedt <[email protected]>
> Signed-off-by: Ingo Molnar <[email protected]>
> ---
> kernel/trace/trace.c | 3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index a0174a4..15f0eec 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -2358,7 +2358,7 @@ static const char readme_msg[] =
> "# mkdir /debug\n"
> "# mount -t debugfs nodev /debug\n\n"
> "# cat /debug/tracing/available_tracers\n"
> - "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
> + "wakeup preemptirqsoff preemptoff irqsoff function
> sched_switch none\n\n"
> "# cat /debug/tracing/current_tracer\n"
> "none\n"
> "# echo sched_switch > /debug/tracing/current_tracer\n"
> @@ -2368,6 +2368,7 @@ static const char readme_msg[] =
> "noprint-parent nosym-offset nosym-addr noverbose\n"
> "# echo print-parent > /debug/tracing/trace_options\n"
> "# echo 1 > /debug/tracing/tracing_enabled\n"
> + "# chrt -f 5 sleep 1\n"
> "# cat /debug/tracing/trace > /tmp/trace.txt\n"
> "echo 0 > /debug/tracing/tracing_enabled\n"
> ;
>

Actually Nikanth sent a similar patch a month ago, but somehow it
slipped under the radar. I'll be picking that one up instead.

http://lkml.org/lkml/2009/3/23/24

But thanks anyway.

-- Steve

2009-04-07 04:37:31

by Geunsik Lim

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace

> What's the chrt for?
We often need simple test to get latency result of realtime using
"current_tracers" file.
In this case, We can run easy testcase as realtime task with the chrt command.

Although we usually use function tracer for tracing internal kernel
functions by ftraced.
I think that we need simple example like the chrt for monitoring
latencies and the kernel
functoins using sleep command as realtime task.


2009/4/7 Steven Rostedt <[email protected]>:
>
> On Mon, 2009-04-06 at 23:17 +0900, GeunSik Lim wrote:
>> I attached patch file again because no-operation is changed
>> from "none"word to "nop" word.
>>
>> for example,
>> -       "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
>> +       "wakeup preemptirqsoff preemptoff irqsoff function sched_switch nop\n\n"
>>
>> ./trace_nop.c:struct tracer nop_trace __read_mostly =
>>       .name   = "nop",
>>       .init           = nop_trace_init,
>>       .reset  = nop_trace_reset,
>>       .selftest       = trace_selftest_startup_nop,
>>       .flags  = &nop_flags,
>>       .set_flag       = nop_set_flag
>>
>>
>>
>> 2009/4/6 GeunSik Lim <[email protected]>:
>> >   Dear Steven,
>> >
>> >   The author of ftrace(Internal kernel Function Tracer) feature
>> > renamed the ftrace tracer
>> >    to function to avoid further confusion between the ftrace
>> > infrastructure and the
>> >    function tracer on Oct-20, 2008.
>> >
>> >    But, Update of Mini-HOWTO document are still not working
>> >    .
>> >    Most of the developers often utilize "/debug/tracing/README" file
>> > as Mini-HOWTO.
>> >    They will confuse definition and usage about ftrace  between
>> > README(ftrace) file
>> >    and available_tracers(function) file after linux-2.6.27 like me.
>> >    And, I want you to append "chrt -f 5 sleep 1" command as simple
>> > example for latency.
>> >
>> >    Signed-off-by: GeunSik Lim <[email protected]>
>> >    Signed-off-by: Steven Rostedt <[email protected]>
>> >    Signed-off-by: Ingo Molnar <[email protected]>
>> > ---
>> >  kernel/trace/trace.c |    3 ++-
>> >  1 files changed, 2 insertions(+), 1 deletions(-)
>> >
>> > diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
>> > index a0174a4..15f0eec 100644
>> > --- a/kernel/trace/trace.c
>> > +++ b/kernel/trace/trace.c
>> > @@ -2358,7 +2358,7 @@ static const char readme_msg[] =
>> >        "# mkdir /debug\n"
>> >        "# mount -t debugfs nodev /debug\n\n"
>> >        "# cat /debug/tracing/available_tracers\n"
>> > -       "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
>> > +       "wakeup preemptirqsoff preemptoff irqsoff function
>> > sched_switch none\n\n"
>> >        "# cat /debug/tracing/current_tracer\n"
>> >        "none\n"
>> >        "# echo sched_switch > /debug/tracing/current_tracer\n"
>> > @@ -2368,6 +2368,7 @@ static const char readme_msg[] =
>> >        "noprint-parent nosym-offset nosym-addr noverbose\n"
>> >        "# echo print-parent > /debug/tracing/trace_options\n"
>> >        "# echo 1 > /debug/tracing/tracing_enabled\n"
>> > +       "# chrt -f 5 sleep 1\n"
>
>
> What's the chrt for?
>
> -- Steve
>
>
>



--
Regards,
GeunSik Lim

2009-04-07 05:08:24

by Geunsik Lim

[permalink] [raw]
Subject: Re: [PATCH] Update old Mini-HOWTO for ftrace

Dear steven,

I confirm http://lkml.org/lkml/2009/3/23/24 webpage now.
I am gratified with a similar patch that another people sent this problem.

In fact, I figured out trivial problem of old mini-HOWTO according to
http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_MRG/1.1/html/Realtime_Tuning_Guide/sect-Realtime_Tuning_Guide-Realtime_Specific_Tuning-Using_the_ftrace_Utility_for_Tracing_Latencies.html
web manual.

I want you to change MRG Kernel from ftrace to function for
consistency in private.
Thank you for interesting article

Regards,
GeunSik Lim..

2009/4/7 Steven Rostedt <[email protected]>:
>
> On Mon, 2009-04-06 at 23:02 +0900, GeunSik Lim wrote:
>> Dear Steven,
>>
>>    The author of ftrace(Internal kernel Function Tracer) feature
>> renamed the ftrace tracer
>>     to function to avoid further confusion between the ftrace
>> infrastructure and the
>>     function tracer on Oct-20, 2008.
>>
>>     But, Update of Mini-HOWTO document are still not working
>>     .
>>     Most of the developers often utilize "/debug/tracing/README" file
>> as Mini-HOWTO.
>>     They will confuse definition and usage about ftrace  between
>> README(ftrace) file
>>     and available_tracers(function) file after linux-2.6.27 like me.
>>     And, I want you to append "chrt -f 5 sleep 1" command as simple
>> example for latency.
>>
>>     Signed-off-by: GeunSik Lim <[email protected]>
>>     Signed-off-by: Steven Rostedt <[email protected]>
>>     Signed-off-by: Ingo Molnar <[email protected]>
>> ---
>>  kernel/trace/trace.c |    3 ++-
>>  1 files changed, 2 insertions(+), 1 deletions(-)
>>
>> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
>> index a0174a4..15f0eec 100644
>> --- a/kernel/trace/trace.c
>> +++ b/kernel/trace/trace.c
>> @@ -2358,7 +2358,7 @@ static const char readme_msg[] =
>>         "# mkdir /debug\n"
>>         "# mount -t debugfs nodev /debug\n\n"
>>         "# cat /debug/tracing/available_tracers\n"
>> -       "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n"
>> +       "wakeup preemptirqsoff preemptoff irqsoff function
>> sched_switch none\n\n"
>>         "# cat /debug/tracing/current_tracer\n"
>>         "none\n"
>>         "# echo sched_switch > /debug/tracing/current_tracer\n"
>> @@ -2368,6 +2368,7 @@ static const char readme_msg[] =
>>         "noprint-parent nosym-offset nosym-addr noverbose\n"
>>         "# echo print-parent > /debug/tracing/trace_options\n"
>>         "# echo 1 > /debug/tracing/tracing_enabled\n"
>> +       "# chrt -f 5 sleep 1\n"
>>         "# cat /debug/tracing/trace > /tmp/trace.txt\n"
>>         "echo 0 > /debug/tracing/tracing_enabled\n"
>>  ;
>>
>
> Actually Nikanth sent a similar patch a month ago, but somehow it
> slipped under the radar. I'll be picking that one up instead.
>
> http://lkml.org/lkml/2009/3/23/24
>
> But thanks anyway.
>
> -- Steve
>
>
>



--
Regards,
GeunSik Lim