Re: LTTng 0.10.0-pre7 port to 2.6.23.1

Mathieu Desnoyers <[email protected]> Wed, 17 Oct 2007 11:07:34 -0400
Newsgroups gmane.linux.kernel.tracing
Message-ID <20071017150734.GA2409@Krystal>
Hi,

I just made a port to 2.6.23.1 available on the website.

Mathieu

* Mathieu Desnoyers ([email protected]) wrote:
> * Mathieu Desnoyers ([email protected]) wrote:
> > Hello!
> > 
> > I had some time to finally get lttv in sync with lttng last week, so I
> > release things in their current state. Many thanks to Pierre-Marc
> > Fournier for digging into this with me.
> > 
> > Most interesting changes:
> > 
> > The main feature is a much simplified event addition: add a trace_mark() in the
> > kernel code and you are all set. It now uses ltt-armall and ltt-disarmall to
> > enable/disable all markers. You can look at these scripts to see how to
> > granunarily enable/disable markers. You don't have to create any probe, and no
> > more XML event definition is needed. The event data types and the field names
> > are all taken from the marker format string.
> > 
> > It also supports crash dump trace extraction. See
> > http://sourceforge.net/projects/crosscrash/ for more information about
> > the user-space crash extraction app. I suspect it does not support
> > 2.6.23-mm1 though.
> > 
> > A new --edebug option has been added: it dumps the events header and 
> > variable length data in hexedit-style on the standard output before the
> > events are used by lttv. This can be very useful for debugging lttv.
> > 
> > The lttng patchset has been completely refactored to be easier to get
> > into Linux kernel mainline. Its core is even more architecture independant than
> > before, relying on a logical clock fallback when no precise timestamp counter
> > can be found. The only architecture dependent part left is the architecture
> > specific instrumentation (principally trap entry/exit, ipc, syscall entry/exit,
> > kthread creation).
> > 
> > The newest versions are :
> > LTTng 0.10.0-pre5
> 
> Please use LTTng 0.10.0-pre6, which includes a compilation fix (forgot
> to refresh a patch before the release).
> 
> Mathieu
> 
> > LTTV 0.10.0-pre1
> > ltt-control 0.45
> > 
> > You will find them on the http://ltt.polymtl.ca website. Testing is very
> > welcome.
> > 
> > What is left to be done:
> > 
> > - Create a dictionnary file infrastructure in lttv that will map
> >   integers to strings (enumerations). I.e., syscalls are currently
> >   printed as "syscall X", irq as "irq X"...
> > - User space tracing
> > - Kernel and user space stack dump (sequences are not re-implemented)
> > - Update and reactivate gui disperformance and tutorial plugins.
> > - LTTng : ia64 and s390 build will break because the 8 bit fields used
> >   for thread flags cannot hold the kernel trace thread flag.
> > 
> > There are probably some features I am missing in this TODO list.
> > 
> > Mathieu
> > 
> > -- 
> > Mathieu Desnoyers
> > Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal
> > OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
> > _______________________________________________
> > 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
> _______________________________________________
> 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