This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: [PATCH v2 2.6.38-rc8-tip 0/20] 0: Inode based uprobes
- From: Srikar Dronamraju <srikar at linux dot vnet dot ibm dot com>
- To: Andrew Morton <akpm at linux-foundation dot org>
- Cc: Peter Zijlstra <peterz at infradead dot org>, Ingo Molnar <mingo at elte dot hu>, Steven Rostedt <rostedt at goodmis dot org>, Linux-mm <linux-mm at kvack dot org>, Arnaldo Carvalho de Melo <acme at infradead dot org>, Linus Torvalds <torvalds at linux-foundation dot org>, Masami Hiramatsu <masami dot hiramatsu dot pt at hitachi dot com>, Christoph Hellwig <hch at infradead dot org>, Ananth N Mavinakayanahalli <ananth at in dot ibm dot com>, Andi Kleen <andi at firstfloor dot org>, Oleg Nesterov <oleg at redhat dot com>, Jim Keniston <jkenisto at linux dot vnet dot ibm dot com>, Roland McGrath <roland at hack dot frob dot com>, SystemTap <systemtap at sources dot redhat dot com>, LKML <linux-kernel at vger dot kernel dot org>, "Paul E. McKenney" <paulmck at linux dot vnet dot ibm dot com>
- Date: Tue, 15 Mar 2011 07:27:40 +0530
- Subject: Re: [PATCH v2 2.6.38-rc8-tip 0/20] 0: Inode based uprobes
- References: <20110314133403.27435.7901.sendpatchset@localhost6.localdomain6> <20110314163028.a05cec49.akpm@linux-foundation.org>
- Reply-to: Srikar Dronamraju <srikar at linux dot vnet dot ibm dot com>
> >
> > 4. Corelating events from kernels and userspace.
> > Uprobes could be used with other tools like kprobes, tracepoints or as
> > part of higher level tools like perf to give a consolidated set of
> > events from kernel and userspace. In future we could look at a single
> > backtrace showing application, library and kernel calls.
>
> How do you envisage these features actually get used? For example,
> will gdb be modified? Will other debuggers be modified or written?
>
> IOW, I'm trying to get an understanding of how you expect this feature
> will actually become useful to end users - the kernel patch is only
> part of the story.
>
Right, So I am looking at having perf probe for uprobes and also at
having a syscall so that perf probe and other ptrace users can use this
infrastructure. Ingo has already asked for a syscall for the same in
one of his replies to my previous patchset. From whatever
discussions I had with ptrace users, they have shown interest in
using this breakpoint infrastructure.
I am still not sure if this feature should be exposed thro a new
operation to ptrace syscall (something like SET_BP) or a completely new
syscall or both. I would be happy if people here could give inputs on
which route to go forward.
We had perf probe for uprobes implemented in few of the previous
patchset. When the underlying implementation changed from a
pid:vaddr to a file:offset, the way we invoke perf probe changed.
Previously we would do
"perf probe -p 3692 zfree@zsh"
Now we would be doing
"perf probe -u zfree@zsh"
The perf probe for uprobes is still WIP. Will post the patches when it
is in usable fashion. This should be pretty soon.
If you have suggestions on how this infrastructure could be used
above perf probe and syscall, then please do let me know.
--
Thanks and Regards
Srikar