This is the mail archive of the
mailing list for the systemtap project.
stap meeting 20060817
- From: Elena Zannoni <ezannoni at redhat dot com>
- To: systemtap at sources dot redhat dot com
- Date: Thu, 17 Aug 2006 12:56:54 -0400
- Subject: stap meeting 20060817
Will, David, Elena, Josh, Anil, Ananth, Jim.
[I am sure there were more people]
IBM: userspace probes. Still brainstorming. uprobes over utrace not
done and not doable in the next couple of weeks.
Did anybody do any tests of time/overhead of ptrace/utrace? elena: to
post. there is a minimal comparison on the Frysk mailing list. See
mail on August 16 from IBM on http://sources.redhat.com/ml/frysk.
request from IBM about when/how the kabi for rhel5 is oign to be frozen:
elena to find out.
Will: update the test coverage info on wcohen people.redhat.com page.
Lookup of variable in the translator: need to understand that in order
to understand the register that the counter is using. Similalry to
$return, $perfcntr (or similar) will need to get the information about
the register content. needs checks that the architecture is the same
(compiled on == run on), will be relevant when we do the cross
David: Will post something tomorrow. [sorry i completely missed this one]
Josh: bug 3079: deref macro cannot deal with 64 bit values on 32 bit
platform. Needs to be solved to safely deref a pointer/catch when a
pointer is bogus. There should be another bug open about this in the
kprobes side 1303. Roland rejected some ideas but we need an
alternative. Dprobes handles this. Somebody needs to follow up on the
kprobes team: Jim.
Anil: utrace main part is done. Running gdb tests. Close to old
ptrace. Idea to run frysk tests in 32 bit more on ia64 with ia32el
(wow). Should be in rawhide very soon now.
Jeff: GUI. Please submit scripts. There aren't that many available and
the gui needs them. Any script that generates periodic output, do not
worry about the graphical display of the output, the GUI team will
figure out what to do there.