Skip to contentSkip to conversation

kmemtrace RFC patch series

RFC PATCH In review Eduard - Gabriel Munteanu 3 participants 5 messages Jul 10, 2008 – Jul 11, 2008 Memory management

lore
Share & tools
lore
#0 lore raw

cover kmemtrace RFC patch series

Hi everybody,

I'd like to hear your opinion regarding kmemtrace and SL*B hooks.

This is just a RFC, it's not intended to be merged yet. The userspace app is not included.

BTW, there may be some whitespace errors, but disregard them for now, they will be fixed.

Cheers,
Eduard

Eduard - Gabriel Munteanu (5):
kmemtrace: Core implementation.
Add new GFP flag __GFP_NOTRACE.
kmemtrace: SLAB hooks.
kmemtrace: SLUB hooks.
kmemtrace: SLOB hooks.

 MAINTAINERS               |    6 ++
 include/linux/gfp.h       |    1 +
 include/linux/kmemtrace.h |  110 +++++++++++++++++++++++
 include/linux/slab_def.h  |   16 +++-
 include/linux/slub_def.h  |    9 ++-
 init/main.c               |    2 +
 lib/Kconfig.debug         |    4 +
 mm/Makefile               |    2 +-
 mm/kmemtrace.c            |  213 +++++++++++++++++++++++++++++++++++++++++++++
 mm/slab.c                 |   35 ++++++--
 mm/slob.c                 |   37 +++++++--
 mm/slub.c                 |   49 +++++++++--
 12 files changed, 460 insertions(+), 24 deletions(-)
 create mode 100644 include/linux/kmemtrace.h
 create mode 100644 mm/kmemtrace.c
#1 lore raw

Hi Eduard-Gabriel,

Quoted 4 lines · Eduard - Gabriel Munteanu

I'd like to hear your opinion regarding kmemtrace and SL*B hooks.

This is just a RFC, it's not intended to be merged yet. The userspace app is not included.

It would probably be helpful for anyone reviewing these patches to see what the userspace program is doing. Can you please post an URL to a tarball of the thing or its git repository?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#2 lore raw
Quoted 4 lines · On Fri, 11 Jul 2008 11:25:31 +0300 "Pekka Enberg" <[email protected]>…

Hi Eduard-Gabriel,

On Thu, Jul 10, 2008 at 9:05 PM, Eduard - Gabriel Munteanu <[email protected]> wrote:

Quoted 8 lines
Quoted 4 lines

I'd like to hear your opinion regarding kmemtrace and SL*B hooks.

This is just a RFC, it's not intended to be merged yet. The userspace app is not included.

It would probably be helpful for anyone reviewing these patches to see what the userspace program is doing. Can you please post an URL to a tarball of the thing or its git repository?

Yes, I'll move this to its own git repo and post a link here soon.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#3 lore raw
In reply to Eduard - Gabriel Munteanu

Hi Eduard,

Did you have a look at the new tracepoints infrastructure ? I think it could simplify your patchset a _lot_ !

Basically, it removes the format string from markers and allows to pass complex structure pointers as arguments. It aims at simplifying the life of in-kernel tracers which would want to use the facility. Turning a marker implementation to tracepoints is really straightforward, for an example see :

http://lkml.org/lkml/2008/7/9/569

For the tracepoints patchset :

http://lkml.org/lkml/2008/7/9/199

I think much of include/linux/kmemtrace.h, which is really just wrappers around marker code, could then go away.

Regards,

Mathieu

Quoted 36 lines · * Eduard - Gabriel Munteanu ([email protected]) wrote:

Hi everybody,

I'd like to hear your opinion regarding kmemtrace and SL*B hooks.

This is just a RFC, it's not intended to be merged yet. The userspace app is not included.

BTW, there may be some whitespace errors, but disregard them for now, they will be fixed.

Cheers,
Eduard

Eduard - Gabriel Munteanu (5):
kmemtrace: Core implementation.
Add new GFP flag __GFP_NOTRACE.
kmemtrace: SLAB hooks.
kmemtrace: SLUB hooks.
kmemtrace: SLOB hooks.

 MAINTAINERS               |    6 ++
 include/linux/gfp.h       |    1 +
 include/linux/kmemtrace.h |  110 +++++++++++++++++++++++
 include/linux/slab_def.h  |   16 +++-
 include/linux/slub_def.h  |    9 ++-
 init/main.c               |    2 +
 lib/Kconfig.debug         |    4 +
 mm/Makefile               |    2 +-
 mm/kmemtrace.c            |  213 +++++++++++++++++++++++++++++++++++++++++++++
 mm/slab.c                 |   35 ++++++--
 mm/slob.c                 |   37 +++++++--
 mm/slub.c                 |   49 +++++++++--
 12 files changed, 460 insertions(+), 24 deletions(-)
 create mode 100644 include/linux/kmemtrace.h
 create mode 100644 mm/kmemtrace.c
Signature · 2 lines
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
#4 lore raw
Quoted 4 lines · On Fri, 11 Jul 2008 11:38:41 -0400 Mathieu Desnoyers…

Hi Eduard,

Did you have a look at the new tracepoints infrastructure ? I think it could simplify your patchset a _lot_ !

Hi,

Yes, I did. In fact, I asked you to keep me Cc-ed, which you did.

Quoted 6 lines

Basically, it removes the format string from markers and allows to pass complex structure pointers as arguments. It aims at simplifying the life of in-kernel tracers which would want to use the facility. Turning a marker implementation to tracepoints is really straightforward, for an example see :

Quoted 8 lines

http://lkml.org/lkml/2008/7/9/569

For the tracepoints patchset :

http://lkml.org/lkml/2008/7/9/199

I think much of include/linux/kmemtrace.h, which is really just wrappers around marker code, could then go away.

Basically, I want just as much to get rid of markers. I'm just waiting for tracepoints to get closer to mainline.

Cheers,
Eduard

Quoted 3 lines

Regards,

Mathieu

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>