Linux 5.2 rc6

Uhh-uh.

I really was hoping that we'd continue to have an increasingly quiet
and shrinking rc series. But that was not to be.

rc6 is the biggest rc in number of commits we've had so far for this
5.2 cycle (obviously ignoring the merge window itself and rc1). And
it's not just because of trivial patches (although admittedly we have
those too), but we obviously had the TCP SACK/fragmentation/mss fixes
in there, and they in turn required some fixes too.

Happily we did pick up on the problem quickly - largely thanks to the
patches making it into distro kernels quickly and then causing
problems for the steam client of all things - but it's still something
that doesn't exactly make me get the warm and fuzzies at this point in
the release cycle.

I'm also doing this rc on a Saturday, because I am going to spend all
of tomorrow on a plane once again. So I'm traveling first for a
conference and then for some R&R on a liveaboard, so I'm going to have
spotty access to email for a few days, and then for a week I'll be
entirely incommunicado. So rc7 will be delayed.

I was thinking that I timed it all really well in what should be the
quietest period of the release cycle for me, and now I obviously hope
that last week really was a fluke.

Anyway, if something happens when I'm offline, Greg can presumably
step up, although he'll have the same conference travel (but
presumably at least the reverse jetlag 😉

With all that out of the way, I'm still reasonably optimistic that
we're on track for a calm final part of the release, and I don't think
there is anything particularly bad on the horizon.

And while we did have some excitement this week, _most_ of it by far
was the usual small fixes. Including the by now expected SPDX updates,
so the diffstat looks a bit messy again.

Anyway, ignoring the SPDX updates (and you should, even if they
dominate the diffstat), about a quarter of the rc6 update is
networking (the TCP fixes being a fairly small part of it - the bulk
is still network driver and other networking fixes, including bpf).
Another quarter is selftests (mostly bpf) and documentation.

The rest other driver updates (gpu, rdma, thunderbolt, usb..) arch
updates (x86, risc-v and arm[64]), and misc other updates (overlayfs
etc).

But honestly, most of it really is pretty small (again - ignoring the
SPDX noise), so despite my misgivings I don't think we're really in
trouble.

Shortlog appended for the brave souls who want to look at details,

                 Linus