Minutes of 6/30/05 meeting

[Jim Keniston <jkenisto at us dot ibm dot com>]

Minutes of SystemTap Conference Call, June 30, 2005

Participants:
IBM:
	Vara Prasad (varap@us.ibm.com -- Beaverton)
	Jim Keniston (jkenisto@us.ibm.com -- Beaverton)
	Hien Nguyen (nguyhien@us.ibm.com -- Beaverton)
	Kevin Stafford (krstaffo@us.ibm.com - Beaverton)
	Tom Zanussi (trz@us.ibm.com -- Chicago)
	Ananth Mavinakayanahalli (amavin@redhat.com, ananth@in.ibm.com --
Boston)
	Prasanna Panchamukhi (prasanna@in.ibm.com -- Bangalore)
	Maneesh Soni (maneesh@in.ibm.com -- Bangalore)
Red Hat:
	Martin Hunt (hunt@redhat.com -- Seattle)
Intel:
	Brad Chen (brad.chen@intel.com -- Santa Clara)
	Rusty Lynch (rusty.lynch@intel.com -- Hillsboro, OR)
	Charles Spirakis (charles.spirakis@intel.com)
	Chris Elford

Actions Required (ARs):
>From 5/19/05:
19. Vara: Collect sample kprobes modules and post them to the
SystemTap web site.

21. Rusty, Will: Look into adding kprobes tests to LTP and the
Red Hat suite, respectively.

22. Rusty: Look into x86_64 limitation re: temporary disarming of
reentrant probes.

23. Jim, Martin: Ensure that the runtime's stack-trace feature works
appropriately when there are return probes in the call stack.
[Update: Martin is working on this.]

>From 6/9/05:
26. Rusty/Brad: Forward email address of Intel test person to Jay
Turner.

>From 6/23/05:
30. Rusty: Review Will's rendition of Rusty's test that exercises
probes on a wide variety of instruction types.

31. Anil: Publish description of problem and solution regarding ia64
probes set on slot 1 of a bundle.

32. Martin/Tom: Benchmark relayfs performance vs. netlink in a
kprobes/SystemTap context.  [Update: in progress]

33. Vara: Publish time slot for tracing/SystemTap BOF at OLS.
[DONE: The BOF will be Friday, 7/22, at 8 pm.]

34. Vara/Satoshi: Discuss possibility of LKST demo at BOF.

35. Prasanna: Contact Andrew M to request that all approved kprobes
patches get pushed to Linus.

36. Jim: Create and submit Documentation/kprobes.txt writeup.
[Update: Doc by Prasanna & Jim posted to SystemTap list 6/28/05]

>From 6/30/05:
37. Vara: Publish C Tapsets Rev 3 after Jim's review.
[Update: Jim's review is done.]

Vara has sent email introducing some performance experts from IBM to
performance experts on the SystemTap team.

Ananth is up to date in his work backporting kprobes features to
RHEL4 U2.

Rusty's enhancements for return probes are in v2.6.13-rc1.  One
minor sparc64-related patch remains.

Charles is helping Stefane Eranian with his updates to perfmon2.
Stefane's goal is a platform-independent infrastructure for managing
and accessing performance counters.  perfmon2 is infrastructure,
not a tool.  (For example, oprofile uses perfmon[2].)  As currently
implemented, perfctr and perfmon can't coexist in the same
kernel, since both assume they own the pmu.  perfmon2 is trying
to include upward compatibility from perfctr.  See Charles's
src/tapsets/profile/profile_tapset.txt document in CVS for more info.

Brad says the validator is going well.  As the translator and
runtime settle down, Brad will update the validator's list of known
"safe" functions from the runtime.  Brad is looking into system-call
instrumentation, and will consult Prasanna on this.

Vara is trying to bring David Gibson into the perf-counter discussion.
Brad expects that David will be interested in perf counters, but
perhaps not in SystemTap in particular.

Prasanna has been working on a variety of projects, including VFS-layer
tracing, Documentation/kprobes.txt, enhancing kprobes to refuse to
probe its own code (or code that it uses), system-call instrumentation,
and something else that I missed.

Jim hacked up Prasanna's Documentation/kprobes.txt and published it
for review by SystemTap folks.  He will incorporate comments and
post it to LKML next week.

Anath has been working on:
- backporting kprobes patches to RHEL4 U2
- porting the runtime to ppc64.  He has checked in his work so far,
but a bug or two remain.  Martin will help as needed.
- read-copy-update locking changes for SMP scalability

Vara will publish Rev 3 of his C Tapsets proposal as soon as
Jim reviews it.  Vara should know the OLS BOF schedule soon.

There's some demand for kprobes (and eventually SystemTap, presumably)
on s390.

Martin has been polishing up the runtime, with work on kernel-user
transport, benchmarking, and error handling.  Next comes integration
with the translator.

Brad again voiced the concern that the concept of a tapset is still
poorly defined.