Re: [RFC PATCH 4/5] kmemtrace: SLUB hooks.

[Eduard - Gabriel Munteanu]

On Sat, 12 Jul 2008 16:28:36 +0300
Eduard - Gabriel Munteanu <eduard.munteanu@xxxxxxxxxxx> wrote:

> On Fri, 11 Jul 2008 09:48:03 -0500
> Christoph Lameter <cl@xxxxxxxxxxxxxxxxxxxx> wrote:
> 
> > Pekka Enberg wrote:
> >  
> > > Christoph, can you please take a look at this?
> > 
> > Yeah. I saw it. Is there some high level description as to how this
> > is going to be used?
> 
> Here is the userspace application's git tree:
> http://repo.or.cz/w/kmemtrace-user.git

Basically, you go like this:
1. Boot the kernel, preferably into 'single' runlevel.
2. Mount debugfs and whatever other filesystems.
3. Run kmemtraced, wait a few seconds and stop it.
4*. Check /debug/kmemtrace/total_overruns (debugfs is mounted on /debug)
to see if there were any buffer overruns.
5*. Run kmemtrace-check on all CPU cpu*.out to see if there are any
erroneous events.
6. Run kmemtrace-report with no parameters to get a short summary of
how the allocator performs.

* - you can optionally skip these steps.

To build the userspace app, launch './configure' when running on that
particular kmemtrace-enabled kernel. It should correctly detect the
headers directory. Otherwise, add KERNEL_SOURCES variable as an
argument to configure. Then run 'make'.

BTW, that repo may change, so don't rely on those commits to be
consistent for a long time. I could delete them all and restructure the
project.


	Eduard
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/