So nothing all that exciting happened, and while I was tempted a couple of times to do an rc8, there really wasn't any reason for it.
Just as an example, Sasha Levin used KASan and found an interesting bug in paravirtualized spinlocks, but realistically it's been around forever, and it's not even clear that it can really ever trigger in practice. We'll get it fixed, and mark it for stable, and tempting as it was, it wasn't really a reason to delay 3.19. And the actual fixes that went in (see appended shortlog) were all fairly small, with the exception of some medium-sized infiniband changes that were all reverting code that just wasn't ready. So it's out there - go and get it. And as a result, the merge window for 3.20 is obviously also now open. Linus