This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: [RFC 0/3] Kprobes: support for userspace probes
- From: Prasanna S Panchamukhi <prasanna at in dot ibm dot com>
- To: James Dickens <jamesd dot wi at gmail dot com>
- Cc: systemtap at sources dot redhat dot com
- Date: Thu, 15 Sep 2005 18:29:46 +0530
- Subject: Re: [RFC 0/3] Kprobes: support for userspace probes
- References: <20050914063035.GA15955@in.ibm.com> <cd09bdd105091411147ddb8cad@mail.gmail.com>
- Reply-to: prasanna at in dot ibm dot com
Hi James,
On Wed, Sep 14, 2005 at 01:14:15PM -0500, James Dickens wrote:
> On 9/14/05, Prasanna S Panchamukhi <prasanna@in.ibm.com> wrote:
> > Hi All,
> >
> > Please find the details of the userspace probes below.
> > Also I have tried to implement the same to demonistrate
> > this design, in 3 patches for linux kernel version 2.6.13.
> >
> > 1. Provides interface to insert probes when pages are in memory.
> > 2. Adds a feature to insert probes on pages not present in the memory.
>
> in the case where you probe a userland app, and its swapped out, then
> memory pressure gets worse, and the code segment is discarded, thus
> will be reloaded from disk next time. do you have a hook to tell the
> monitoring software than they are no longer being probed? or do you
> have a machanism that protects a block with probes in it from being
> discarded?
>
As Richard mentioned, even here we add hooks at readpage() and readpages().
When the page is not present during registeration time, readpage hooks
will take care of inserting the probes.
Please see this patch for detials.
[RFC 2/3] Kprobes: userspace probes adding readpages hook
Thanks
Prasanna
--
Prasanna S Panchamukhi
Linux Technology Center
India Software Labs, IBM Bangalore
Ph: 91-80-25044636
<prasanna@in.ibm.com>