Re: Current state of LTTng?
Mathieu Desnoyers <[email protected]> Tue, 11 Sep 2007 18:06:08 -0400
| Newsgroups | gmane.linux.kernel.tracing |
|---|---|
| Message-ID | <20070911220608.GA32515@Krystal> |
* Deepak Saxena ([email protected]) wrote: > > What is the current state of LTTng? The git tree has not been updated > in months and I'm not sure where all the non-lttng patches in the snapshots > director are from (-mm?). Are there any plans to update the git tree? > Hi Deepak, I currently work on top on -mm to make integration of my stuff into mainline easier. The most recent snapshots are at http://ltt.polymtl.ca/lttng/snapshot-* The reason why I stopped using git to manage my tree is that I currently use quilt to manage a patchset which I send to Andrew Morton chunk by chunk (as you probably know, that's a lengthy process). The current state is: LTTng is fully reimplemented on top of the markers, dynamic marker actication can be done through /proc/ltt, marker per marker. The "facility" concept has been dropped, so now the information is exported as "marker load" and "marker format" events. The event types are parsed from the format string of the marker. As you might see, it will make the process of adding new markers "trivial", which is the goal. However, in the current implementation state, the stack dumps are disabled because they use sequences (not reimplemented yet), I plan to drop some composite types which are not absolutely necessary. Also, I am currently porting LTTV to this new facility-free trace format. Last thing: user-space tracing is temporarily removed from the series file because it's the one that causes the most rejects (system call numbers must be changed every new release). In order to keep to speed with -mm, I just don't deal with those for now. They will have to be changed quite a bit to user the markers (yes, I plan to use markers for user-space tracing too). As you can see, things are not standing still, but the change being quite big, it takes some time before I can cut a release. Mathieu > ~Deepak > > -- > Deepak Saxena - [email protected] - http://www.plexity.net > > "Ever bigger machines, entailing ever bigger concentrations of economic > power and exerting ever greater violence against the environment, do not > represent progress: they are a denial of wisdom. Wisdom demands a new > orientation of science and technology towards the organic, the gentle, the > non-violent, the elegant and beautiful." - Small is Beautiful > > _______________________________________________ > Ltt-dev mailing list > [email protected] > http://listserv.shafik.org/mailman/listinfo/ltt-dev > -- Mathieu Desnoyers Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68