LTTng 0.73 for Linux 2.6.28

LTTng 0.73 for Linux 2.6.28

[Mathieu Desnoyers]

Hi everyone,

I just ported LTTng to 2.6.28. Enjoy !

You may also want to upgrade to lttv 0.12.5 which puts back the syscall
and softirq names in the event list.

For newcomers, instructions to install LTTng can be found at :
http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART

(I just updated the QUICKSTART guide. Please feel free to try it and
report inconsistencies.)

Mathieu


-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints)

[Mathieu Desnoyers]

Hi,

LTTng 0.74 fixes the probe Makefile to build the net-trace probe module.
It also adds jbd2 and ext4 tracepoints, and jbd2 probe module.

A new version including ext4 probe module should be available shortly.

Installation QUICKSTART :
http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART

git url :

http://git.kernel.org/?p=linux/kernel/git/compudj/linux-2.6-lttng.git;a=shortlog;h=2.6.28-lttng-0.74

tarball : 

http://ltt.polymtl.ca/lttng/patch-2.6.28-lttng-0.74.tar.gz

jbd2/ext4 patches within the tarball :

ext4-instrumentation-add-tracepoints.patch
jbd2-instrumentation-move-to-tracepoints.patch
lttng-instrumentation-jbd2-probes.patch

Mathieu

* Mathieu Desnoyers (compudj@krystal.dyndns.org) wrote:
> Hi everyone,
> 
> I just ported LTTng to 2.6.28. Enjoy !
> 
> You may also want to upgrade to lttv 0.12.5 which puts back the syscall
> and softirq names in the event list.
> 
> For newcomers, instructions to install LTTng can be found at :
> http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART
> 
> (I just updated the QUICKSTART guide. Please feel free to try it and
> report inconsistencies.)
> 
> Mathieu
> 
> 
> -- 
> Mathieu Desnoyers
> OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
> 
> _______________________________________________
> ltt-dev mailing list
> ltt-dev@lists.casi.polymtl.ca
> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
> 

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints)

[KOSAKI Motohiro]

Hi Mathieu

2009/1/11 Mathieu Desnoyers <compudj@krystal.dyndns.org>:
> Hi,
>
> LTTng 0.74 fixes the probe Makefile to build the net-trace probe module.
> It also adds jbd2 and ext4 tracepoints, and jbd2 probe module.

-ECANTBUILD

fs/jbd2/commit.c:27:24: error: trace/jbd2.h: No such file or directory
fs/jbd2/commit.c:29: warning: data definition has no type or storage class
fs/jbd2/commit.c:29: warning: type defaults to 'int' in declaration of
'DEFINE_TRACE'
fs/jbd2/commit.c:29: warning: parameter names (without types) in
function declaration
fs/jbd2/commit.c:30: warning: data definition has no type or storage class
fs/jbd2/commit.c:30: warning: type defaults to 'int' in declaration of
'DEFINE_TRACE'
fs/jbd2/commit.c:30: warning: parameter names (without types) in
function declaration
fs/jbd2/commit.c: In function 'jbd2_journal_commit_transaction':
fs/jbd2/commit.c:374: error: implicit declaration of function
'trace_jbd2_start_commit'
fs/jbd2/commit.c:1003: error: implicit declaration of function
'trace_jbd2_end_commit'
make[2]: *** [fs/jbd2/commit.o] Error 1


because

jbd2-instrumentation-move-to-tracepoints patch only change following two files.
it didn't create trace/jbd2.h

  fs/jbd2/checkpoint.c 		
  fs/jbd2/commit.c 		





>
> A new version including ext4 probe module should be available shortly.
>
> Installation QUICKSTART :
> http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART
>
> git url :
>
> http://git.kernel.org/?p=linux/kernel/git/compudj/linux-2.6-lttng.git;a=shortlog;h=2.6.28-lttng-0.74
>
> tarball :
>
> http://ltt.polymtl.ca/lttng/patch-2.6.28-lttng-0.74.tar.gz
>
> jbd2/ext4 patches within the tarball :
>
> ext4-instrumentation-add-tracepoints.patch
> jbd2-instrumentation-move-to-tracepoints.patch
> lttng-instrumentation-jbd2-probes.patch
>
> Mathieu
>
> * Mathieu Desnoyers (compudj@krystal.dyndns.org) wrote:
>> Hi everyone,
>>
>> I just ported LTTng to 2.6.28. Enjoy !
>>
>> You may also want to upgrade to lttv 0.12.5 which puts back the syscall
>> and softirq names in the event list.
>>
>> For newcomers, instructions to install LTTng can be found at :
>> http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART
>>
>> (I just updated the QUICKSTART guide. Please feel free to try it and
>> report inconsistencies.)
>>
>> Mathieu
>>
>>
>> --
>> Mathieu Desnoyers
>> OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
>>
>> _______________________________________________
>> ltt-dev mailing list
>> ltt-dev@lists.casi.polymtl.ca
>> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
>>
>
> --
> Mathieu Desnoyers
> OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints) (fixed in LTTng 0.78)

[Mathieu Desnoyers]

* KOSAKI Motohiro (kosaki.motohiro@jp.fujitsu.com) wrote:
> Hi Mathieu
> 
> 2009/1/11 Mathieu Desnoyers <compudj@krystal.dyndns.org>:
> > Hi,
> >
> > LTTng 0.74 fixes the probe Makefile to build the net-trace probe module.
> > It also adds jbd2 and ext4 tracepoints, and jbd2 probe module.
> 
> -ECANTBUILD
> 

Hi Kosaki,

Thanks for the report. This should be fixed in LTTng 0.78 now.

Best regards,

Mathieu

> fs/jbd2/commit.c:27:24: error: trace/jbd2.h: No such file or directory
> fs/jbd2/commit.c:29: warning: data definition has no type or storage class
> fs/jbd2/commit.c:29: warning: type defaults to 'int' in declaration of
> 'DEFINE_TRACE'
> fs/jbd2/commit.c:29: warning: parameter names (without types) in
> function declaration
> fs/jbd2/commit.c:30: warning: data definition has no type or storage class
> fs/jbd2/commit.c:30: warning: type defaults to 'int' in declaration of
> 'DEFINE_TRACE'
> fs/jbd2/commit.c:30: warning: parameter names (without types) in
> function declaration
> fs/jbd2/commit.c: In function 'jbd2_journal_commit_transaction':
> fs/jbd2/commit.c:374: error: implicit declaration of function
> 'trace_jbd2_start_commit'
> fs/jbd2/commit.c:1003: error: implicit declaration of function
> 'trace_jbd2_end_commit'
> make[2]: *** [fs/jbd2/commit.o] Error 1
> 
> 
> because
> 
> jbd2-instrumentation-move-to-tracepoints patch only change following two files.
> it didn't create trace/jbd2.h
> 
>   fs/jbd2/checkpoint.c 		
>   fs/jbd2/commit.c 		
> 
> 
> 
> 
> 
> >
> > A new version including ext4 probe module should be available shortly.
> >
> > Installation QUICKSTART :
> > http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART
> >
> > git url :
> >
> > http://git.kernel.org/?p=linux/kernel/git/compudj/linux-2.6-lttng.git;a=shortlog;h=2.6.28-lttng-0.74
> >
> > tarball :
> >
> > http://ltt.polymtl.ca/lttng/patch-2.6.28-lttng-0.74.tar.gz
> >
> > jbd2/ext4 patches within the tarball :
> >
> > ext4-instrumentation-add-tracepoints.patch
> > jbd2-instrumentation-move-to-tracepoints.patch
> > lttng-instrumentation-jbd2-probes.patch
> >
> > Mathieu
> >
> > * Mathieu Desnoyers (compudj@krystal.dyndns.org) wrote:
> >> Hi everyone,
> >>
> >> I just ported LTTng to 2.6.28. Enjoy !
> >>
> >> You may also want to upgrade to lttv 0.12.5 which puts back the syscall
> >> and softirq names in the event list.
> >>
> >> For newcomers, instructions to install LTTng can be found at :
> >> http://ltt.polymtl.ca/svn/trunk/lttv/QUICKSTART
> >>
> >> (I just updated the QUICKSTART guide. Please feel free to try it and
> >> report inconsistencies.)
> >>
> >> Mathieu
> >>
> >>
> >> --
> >> Mathieu Desnoyers
> >> OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
> >>
> >> _______________________________________________
> >> ltt-dev mailing list
> >> ltt-dev@lists.casi.polymtl.ca
> >> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
> >>
> >
> > --
> > Mathieu Desnoyers
> > OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
> >
> 
> _______________________________________________
> ltt-dev mailing list
> ltt-dev@lists.casi.polymtl.ca
> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
> 

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints) (fixed in LTTng 0.78)

[KOSAKI Motohiro]

> * KOSAKI Motohiro (kosaki.motohiro@jp.fujitsu.com) wrote:
> > Hi Mathieu
> > 
> > 2009/1/11 Mathieu Desnoyers <compudj@krystal.dyndns.org>:
> > > Hi,
> > >
> > > LTTng 0.74 fixes the probe Makefile to build the net-trace probe module.
> > > It also adds jbd2 and ext4 tracepoints, and jbd2 probe module.
> > 
> > -ECANTBUILD
> > 
> 
> Hi Kosaki,
> 
> Thanks for the report. This should be fixed in LTTng 0.78 now.
> 
> Best regards,

Unfortunately, lockdep part also can't build.

==
Subject: [PATCH] fix lockdep tracing build error

commit 6afe40b4dace385d7ba2faf24b352f066f3b71bf rename argument a0 to ip.
it cause following build error to lttng tree.

  CC      kernel/lockdep.o
kernel/lockdep.c: In function 'trace_hardirqs_on_caller':
kernel/lockdep.c:2186: error: 'a0' undeclared (first use in this function)
kernel/lockdep.c:2186: error: (Each undeclared identifier is reported only once
kernel/lockdep.c:2186: error: for each function it appears in.)
kernel/lockdep.c: In function 'trace_hardirqs_off_caller':
kernel/lockdep.c:2241: error: 'a0' undeclared (first use in this function)
make[1]: *** [kernel/lockdep.o] Error 1


therefore we also need change trace argument.


Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
---
 kernel/lockdep.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/lockdep.c b/kernel/lockdep.c
index 2ec98a3..21ef2ad 100644
--- a/kernel/lockdep.c
+++ b/kernel/lockdep.c
@@ -2183,7 +2183,7 @@ void trace_hardirqs_on_caller(unsigned long ip)
 
 	time_hardirqs_on(CALLER_ADDR0, ip);
 
-	_trace_lockdep_hardirqs_on(a0);
+	_trace_lockdep_hardirqs_on(ip);
 
 	if (unlikely(!debug_locks || current->lockdep_recursion))
 		return;
@@ -2238,7 +2238,7 @@ void trace_hardirqs_off_caller(unsigned long ip)
 
 	time_hardirqs_off(CALLER_ADDR0, ip);
 
-	_trace_lockdep_hardirqs_off(a0);
+	_trace_lockdep_hardirqs_off(ip);
 
 	if (unlikely(!debug_locks || current->lockdep_recursion))
 		return;
-- 
1.6.0.6

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints) (fixed in LTTng 0.78)

[Theodore Tso]

On Tue, Jan 13, 2009 at 05:58:27PM -0500, Mathieu Desnoyers wrote:
> > 
> > -ECANTBUILD
> > 
> Thanks for the report. This should be fixed in LTTng 0.78 now.

Out of curiosity, do you maintain a branch of LTTNG versus any of the
-rc kernels, or do you only track the stable kernels?  I tried pulling
LTTNG 0.78 against the most recent Linus mainline and got a whole pile
of conflicts.

I then tried building LTTng 0.78 and got build failures:

  CC      arch/x86/kernel/init_task.o
distcc[27501] ERROR: compile /var/cache/ccache/init_task.tmp.closure.27497.i on localhost failed
/usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: unknown field ‘user_markers_mutex’ specified in initializer
/usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: warning: braces around scalar initializer
/usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: warning: (near initialization for ‘init_task.latency_record_count’)
/usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: field name not in record or union initializer
/usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: (near initialization for ‘init_task.latency_record_count’)

Looks like CONFIG_MARKERS_USERSPAC is busted....

						- Ted

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints) (fixed in LTTng 0.78)

[Mathieu Desnoyers]

* KOSAKI Motohiro (kosaki.motohiro@jp.fujitsu.com) wrote:
> > * KOSAKI Motohiro (kosaki.motohiro@jp.fujitsu.com) wrote:
> > > Hi Mathieu
> > > 
> > > 2009/1/11 Mathieu Desnoyers <compudj@krystal.dyndns.org>:
> > > > Hi,
> > > >
> > > > LTTng 0.74 fixes the probe Makefile to build the net-trace probe module.
> > > > It also adds jbd2 and ext4 tracepoints, and jbd2 probe module.
> > > 
> > > -ECANTBUILD
> > > 
> > 
> > Hi Kosaki,
> > 
> > Thanks for the report. This should be fixed in LTTng 0.78 now.
> > 
> > Best regards,
> 
> Unfortunately, lockdep part also can't build.
> 

I may be missing something, but I'm pretty sure the
lttng-instrumentation-lockdep.patch in 0.78 also fixes this. (it's been
updated)

Mathieu

> ==
> Subject: [PATCH] fix lockdep tracing build error
> 
> commit 6afe40b4dace385d7ba2faf24b352f066f3b71bf rename argument a0 to ip.
> it cause following build error to lttng tree.
> 
>   CC      kernel/lockdep.o
> kernel/lockdep.c: In function 'trace_hardirqs_on_caller':
> kernel/lockdep.c:2186: error: 'a0' undeclared (first use in this function)
> kernel/lockdep.c:2186: error: (Each undeclared identifier is reported only once
> kernel/lockdep.c:2186: error: for each function it appears in.)
> kernel/lockdep.c: In function 'trace_hardirqs_off_caller':
> kernel/lockdep.c:2241: error: 'a0' undeclared (first use in this function)
> make[1]: *** [kernel/lockdep.o] Error 1
> 
> 
> therefore we also need change trace argument.
> 
> 
> Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
> ---
>  kernel/lockdep.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/lockdep.c b/kernel/lockdep.c
> index 2ec98a3..21ef2ad 100644
> --- a/kernel/lockdep.c
> +++ b/kernel/lockdep.c
> @@ -2183,7 +2183,7 @@ void trace_hardirqs_on_caller(unsigned long ip)
>  
>  	time_hardirqs_on(CALLER_ADDR0, ip);
>  
> -	_trace_lockdep_hardirqs_on(a0);
> +	_trace_lockdep_hardirqs_on(ip);
>  
>  	if (unlikely(!debug_locks || current->lockdep_recursion))
>  		return;
> @@ -2238,7 +2238,7 @@ void trace_hardirqs_off_caller(unsigned long ip)
>  
>  	time_hardirqs_off(CALLER_ADDR0, ip);
>  
> -	_trace_lockdep_hardirqs_off(a0);
> +	_trace_lockdep_hardirqs_off(ip);
>  
>  	if (unlikely(!debug_locks || current->lockdep_recursion))
>  		return;
> -- 
> 1.6.0.6
> 
> 
> 
> 
> 
> 
> 

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints) (fixed in LTTng 0.78)

[Mathieu Desnoyers]

* Theodore Tso (tytso@mit.edu) wrote:
> On Tue, Jan 13, 2009 at 05:58:27PM -0500, Mathieu Desnoyers wrote:
> > > 
> > > -ECANTBUILD
> > > 
> > Thanks for the report. This should be fixed in LTTng 0.78 now.
> 
> Out of curiosity, do you maintain a branch of LTTNG versus any of the
> -rc kernels, or do you only track the stable kernels?  I tried pulling
> LTTNG 0.78 against the most recent Linus mainline and got a whole pile
> of conflicts.
> 
> I then tried building LTTng 0.78 and got build failures:
> 
>   CC      arch/x86/kernel/init_task.o
> distcc[27501] ERROR: compile /var/cache/ccache/init_task.tmp.closure.27497.i on localhost failed
> /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: unknown field ‘user_markers_mutex’ specified in initializer
> /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: warning: braces around scalar initializer
> /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: warning: (near initialization for ‘init_task.latency_record_count’)
> /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: field name not in record or union initializer
> /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: (near initialization for ‘init_task.latency_record_count’)
> 
> Looks like CONFIG_MARKERS_USERSPAC is busted....
> 

Hi Ted,

I'm currently integrating Lai's work to the 2.6.28 lttng branch (ascii
output !) :) and the next step will be to move to 2.6.29-rc1. I'll keep
you posted. Note that you can just leave the 

#New revamped userspace markers

markers-userspace.patch
markers-userspace-x86.patch
markers-userspace-x86_64.patch
sparc64-support-userspace-tracing.patch

Out of the patchset and it should not hurt the kernel tracing part at
all. Note that these 4 patches are planned to be heavily reworked before
lkml submission.

Mathieu

> 						- Ted
> 

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

Re: [ltt-dev] LTTng 0.74 for Linux 2.6.28 (ext4 tracepoints) (fixed in LTTng 0.78)

[Mathieu Desnoyers]

* Mathieu Desnoyers (compudj@krystal.dyndns.org) wrote:
> * Theodore Tso (tytso@mit.edu) wrote:
> > On Tue, Jan 13, 2009 at 05:58:27PM -0500, Mathieu Desnoyers wrote:
> > > > 
> > > > -ECANTBUILD
> > > > 
> > > Thanks for the report. This should be fixed in LTTng 0.78 now.
> > 
> > Out of curiosity, do you maintain a branch of LTTNG versus any of the
> > -rc kernels, or do you only track the stable kernels?  I tried pulling
> > LTTNG 0.78 against the most recent Linus mainline and got a whole pile
> > of conflicts.
> > 
> > I then tried building LTTng 0.78 and got build failures:
> > 
> >   CC      arch/x86/kernel/init_task.o
> > distcc[27501] ERROR: compile /var/cache/ccache/init_task.tmp.closure.27497.i on localhost failed
> > /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: unknown field ‘user_markers_mutex’ specified in initializer
> > /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: warning: braces around scalar initializer
> > /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: warning: (near initialization for ‘init_task.latency_record_count’)
> > /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: field name not in record or union initializer
> > /usr/projects/linux/ext4/arch/x86/kernel/init_task.c:35: error: (near initialization for ‘init_task.latency_record_count’)
> > 
> > Looks like CONFIG_MARKERS_USERSPAC is busted....
> > 
> 
> Hi Ted,
> 
> I'm currently integrating Lai's work to the 2.6.28 lttng branch (ascii
> output !) :) and the next step will be to move to 2.6.29-rc1. I'll keep
> you posted. Note that you can just leave the 
> 
> #New revamped userspace markers
> 
> markers-userspace.patch
> markers-userspace-x86.patch
> markers-userspace-x86_64.patch
> sparc64-support-userspace-tracing.patch
> 
> Out of the patchset and it should not hurt the kernel tracing part at
> all. Note that these 4 patches are planned to be heavily reworked before
> lkml submission.
>

Actually, the x86_32 2.6.28 build was broken because of a bad patch
hunk which caused this problem. I fixed it in lttng 0.81. The port to
2.6.29-rc1 should some in a few days.

Thanks for the report,

Mathieu

> Mathieu
> 
> > 						- Ted
> > 
> 
> -- 
> Mathieu Desnoyers
> OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
> 
> _______________________________________________
> ltt-dev mailing list
> ltt-dev@lists.casi.polymtl.ca
> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68