Gregory Haskins wrote: > Avi Kivity wrote: > >> >> One day we'll have lockless injection and we'll want to drop this. I >> guess if we create the fd ourselves we can make it work, but I don't >> see how we can do this with eventfd. >> >> > > Hmm...this is a good point. There probably is no way to use eventfd > "off the shelf" in a way that doesn't cause this callback to be in a > critical section. Should we just worry about switching away from > eventfd when this occurs, or should I implement a custom anon-fd now? >
Something else to consider: eventfd supports calling eventfd_signal() from interrupt context, so even if the callback were not invoked from a preempt/irq off CS due to the wqh->lock, the context may still be a CS anyway. Now, userspace and vbus based injection do not need to worry about this, but I wonder if this is a desirable attribute for some other source (device-assignment perhaps)? If so, is there a way to design the lockless-injection code such that its still friendly to irq-context, or is this a mode that we would never want to support anyway? I think this would be a good thing to have at least to maintain compatibility with the existing eventfd interface, which has its own advantages. -Greg
signature.asc
Description: OpenPGP digital signature
