On Mon, 2011-05-30 at 12:29 +0300, Pekka Enberg wrote: > On Mon, May 30, 2011 at 11:43 AM, Ingo Molnar <[email protected]> wrote: > > Testing idea: for example 'make test locking' could do the bzImage > > test-bootup, with rwlocks built instead of the brlocksj? > > > > Pekka might have more ideas about how a good locking test-suite > > should be done, as JATO has one, right? > > The current users of brlocks won't actually cause the guest to be > paused under load. That's the part I worry about breaking. So again, > the test case can be a simple as firing up 100-1000 threads where most > of them are taking the read lock but few of them racing to take the > write lock.
It would mean we need that many VCPUs: current br_read_lock() doesn't really do anything, which means that running these tests with dummy threads won't work. > It's usually pretty easy to make buggy suspend/locking implementation > deadlock when run with more than one physical cores. > > Pekka -- Sasha. -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
