Minutes of 9/1/05 meeting

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

Minutes of SystemTap Conference Call, Sept. 1, 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 (kevinrs@us.ibm.com -- Beaverton)
	Tom Zanussi (trz@us.ibm.com -- Chicago)
	Prasanna Panchamukhi (prasanna@in.ibm.com -- Bangalore)
	Maneesh Soni (maneesh@in.ibm.com -- Bangalore)
Red Hat:
	Elena Zannoni (ezannoni@redhat.com -- Boston)
	Will Cohen (wcohen@redhat.com -- Raleigh)
	Martin Hunt (hunt@redhat.com -- Seattle)
	Frank Eigler (fche@redhat.com -- Toronto)
	Graydon Hoare (graydon@redhat.com -- Vancouver)
Intel:
	Anil KeshavaMurthy (anil.s.keshavamurthy@intel.com -- Hillsboro)
	Josh Stone

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

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.  Jim will prototype an improvement
whereby the saved IP is updated before the handler is called.]

>From 6/9/05:
26. Rusty/Brad: Forward email address of Intel test person to Jay
Turner.  [Update: As of 7/7/05, Brad hasn't done this.  Did Rusty?]

>From 6/23/05:
30. Rusty: Review Will's rendition of Rusty's test that exercises
probes on a wide variety of instruction types.  [Update: As of 7/7/05,
Will still needs to hear back from Rusty.]

>From 7/7/05:
39. Ananth: Summarize the status of the ppc64 runtime port for Martin.
[Update 8/17/05: Jose Santos of the IBM pSeries performance team will
complete the port.]

>From 7/28/05:
41. Anil: Post a summary of known djprobes implementation issues.

>From 8/4/05:
43. Prasanna: Post work on user-space probes.  [Update: Design is
under review at IBM.  This is deferred until after current SystemTap
test/debug push.]

>From 8/11/05:
44. Tom: Automate relayfs tests and post them to mailing list.
[Update: In progress]

>From 8/18/05:
48. Elena, Vara: Manage planning of next face-to-face meeting.

50. Jim: Characterize the problem(s) we've seen when probing
schedule().  [DONE: #1229 is fixed by Ananth and Anil.]

53. Anil: Test RCU-based SMP scalability feature on ia64.

>From 8/25/05:
54. Vara: Report on availability of tech writer(s) at IBM to help
with user documentation.

55. Elena: Investigate approaches to making the full SystemTap bug
list available to SystemTap developers outside Red Hat.

The translator system has about all the features it's going to for
a while.  Graydon and Frank are fixing bugs as needed.

Kevin, Anil, and Josh have been exercising SystemTap, mostly
by using it to develop desired instrumentation.  Tom, Prasanna,
and Hien are piling on this week.  Please add tests to the suite
to reflect what you've tested -- regression tests for any bugs,
and functional tests (or tapsets) for stuff that works.

The runtime system is in good shape.  We need to add register-dump
and stack-trace support for ia64.

Frank has used SystemTap on RHEL4 U2 with a stock (kernel.org) kernel.
It seems to work well, but he's seen at least one unexpected failure.
More investigation is needed.

Tom posted a relayfs bug fix to LKML.  As previously discussed,
the version of relayfs in RHEL4 U2 was frozen about a month ago.
Elena is willing to push relayfs updates for RHEL4 U3.  There was
some discussion about using SystemTap+relayfs with a stock kernel.
By default, the runtime doesn't use relayfs, and there's no stap
command-line option or language feature that enables its use.
Currently, the only way to use relayfs is to hand-adjust a macro in
the translator-generated C (and at the same time set a macro to let
relayfs know that the RHEL4 U2 version of relayfs is being used).

The two high-priority kprobes bugs (bugzilla #1229 and #1234) are
fixed.  Jim and Anil will attach RHEL4-U2-targeted patches to the
bugzilla entries.  Ananth will push these to U2 Monday.

Any suggestions as to what instrumentation we should be developing,
beyond what's already been discussed?  Check the web for "favorite
Dtrace scripts."  IBM is doing tapsets for system calls and the
VFS layer.