brainstorming for UDS-N - Performance

[Kees Cook]

On Mon, Oct 04, 2010 at 10:12:45AM -0700, Bryce Harrington wrote:
> On Mon, Oct 04, 2010 at 09:38:25AM -0700, Kees Cook wrote:
> > On Mon, Oct 04, 2010 at 10:28:14AM +0100, Matthew Paul Thomas wrote:
> > > Well then, here's a simple task to start with: We should have a single
> > > graph showing, for every application installed by default in Ubuntu over
> > > the past six months, how long that application took each day to show a
> > > non-splash-screen window after a cold launch.
> > 
> > Is that actually a meaningful measurement? For the desktop team, I would
> > expect day-to-day operational timing to be more interesting. (e.g. firefox
> > start-up happens once, but rendering facebook could happen repeatedly.)
> > 
> > > How *would* someone automate that?
> > 
> > Have a tool polling for x clients? Use kvm-autotest's screen content
> > checker? There a lot of options, I think.
> 
> For gtk apps, gtk provides a routine to do a single iteration through
> the main loop and then return.
> 
> http://www.mail-archive.com/gtg-contributors@lists.launchpad.net/msg00249.html
> http://bazaar.launchpad.net/~gtg/gtg/trunk/revision/858
> 
> Not perfect (no guarantee the app is going to be fully initialized in
> one main loop iteration), but at least gives a crude measure of the
> initial initialization cost.

Wouldn't this require patching each Gtk application that we wanted to
collect measurements from? I wonder if it would be easy to externally
instrument runtime via gdb breakpoints or something to create the same
behavior?

-- 
Kees Cook
Ubuntu Security Team