The rc3 release is bigger than normal, which is obviously never anything I want to see, but at the same time it's early enough in the rc series that it's not something I really worry about. Yet.
And while it's bigger, nothing really unusual stands out. The single biggest patch in there (by far - it's in fact about a third of the whole rc3 patch) is just removal of the mt7621-eth staging driver, which is because the regular mediatek ethernet driver now handles that hardware. In fact, due to the removal of that driver, the full rc3 patch removes more lines than it adds, which is always nice to see (honestly in advertising: looking at the whole release so far rather than just this week's rc, that is very much not true).
But no, that driver removal isn't why I'm saying that rc3 is larger than usual - that's just a single commit. And we simply also have more commits than we normally do at this point. But I'm hopeful that it's just a one-off - the whole "size of rc's" measure tends to be somewhat noisy and depends on whether a particular week saw all the major subsystems merging in their fixes or not. This time we obviously just had a lot of it.
If you ignore that big chunk of the rc3 patch that is the removal of that single driver, the changes really are pretty widely spread out. About a third is drivers in general, another roughly third is tooling and documentation changes, and the final third is just "misc" all over (networking is a big chunk of it, but there are arch updates, core kernel, kvm, you name it).
The appended shortlog gives a flavor of the details, and I don't think there's any fundamental pattern there. Just a lot of small details.
Let's hope the rc's will start shrinking from this point on, but I don't think we have anything particularly bad going on.