Re: [PATCH v2 1/4] trace-cmd: Add parse error checking target

[Steven Rostedt <rostedt@goodmis.org>]

On Fri, 2011-07-29 at 12:07 -0700, Vaibhav Nagarnaik wrote:
> On Fri, Jul 29, 2011 at 10:53 AM, Steven Rostedt <rostedt@goodmis.org> wrote:

> Our intention with this patch was to reuse the format string parsing
> functionality in trace-cmd to look for any errors in new trace events
> added by various kernel developers. So this was targeted to test for
> event format errors and report them during a new kernel release.

Note, I plan on working on the libperf.so soon and this will allow
plugins to be added for both perf and trace-cmd, and will make depending
on a working print-fmt less.

> 
> A trace.dat file has already recorded all the data from the machine and
> when 'report' is used on it, the format string errors will be reported
> regardless. I don't see how using 'check-events' on a trace.dat file
> would be useful. Thoughts?

Just to see if there's issues in a trace.dat file. But perhaps we could
just add a --check to report. I just figured that it shouldn't be to
hard to add, and may prove useful in the future.

Actually, since plugins may not be on all boxes, and a trace.dat file
can come from anywhere. It could be useful to know if the box you are
reading the trace.dat file has the necessary plugins for it or not.

-- Steve