On Thu, Jun 15, 2017 at 08:03:28PM +0300, Roman Kagan wrote: > On Thu, Jun 15, 2017 at 06:45:28PM +0200, Andrew Jones wrote: > > On Thu, Jun 15, 2017 at 06:08:12PM +0300, Roman Kagan wrote: > > > Add a test for Hyper-V message and event connections. > > > > > > It requires QEMU with the extended test device supporting message end > > > event connection test modes (recently posted on qemu-devel). On older > > > QEMU versions the test reports a single SKIP. > > > > > > Signed-off-by: Roman Kagan <rkagan@xxxxxxxxxxxxx> > > > --- > > > v1 -> v2: > > > - dropped already applied patches > > > - add detection of old QEMU per Paolo's suggestion and skip the test if QEMU > > > doesn't support connectionid > > > - adjusted test group in unittests.cfg > > > - fixed indentation to use tabs as it's a new file > > > - note: *not* switched to on_cpus() because different arguments are passed to > > > each cpu > > > > Not a big deal, but you probably could have used on_cpus() for setup_cpu > > and teardown_cpu. > > Indeed. I didn't notice because those are run synchronously in the > current code, but on_cpus() would be fine for them, too. Turns out it isn't, because they use alloc_page/free_page which are not mp-safe. So I'll respin with style fixes and a check for the return value from alloc_page (to save me several hours debugging sporadic test failures and crashes next time I decide to experiment with this code ;) Thanks, Roman.
