This is a brief summary of the changes that are sitting in the sh queue
for 2.6.25.
The main points to note are as follows:
- sh64->sh integration.
- A handful of new CPUs (SH7721, SH7763, SH7203, SH7263).
- SH-2A FPU support.
- Board support updates (R2D, R7785RP).
The sh64->sh integration is basically the only thing that's really
interesting, and so it's worth summarizing that a bit.
The sh64 and sh ports had both areas of considerable divergence, and
areas of considerable overlap. In the past consolidation has not been
possible since there was no clean way to abstract the differences with
a common implementation (the obvious case is that the SH-5 and the other
parts use a totally different instruction and register set). With the
work in the nommu area, we've already had to abstract most of the
exception handling code, disjoint syscall ABIs, incompatible instruction
sets, and so on. As a result of that work, the SH-5 integration was
finally at a point where it could be done with minimal pain. The fact
that the sh64 port itself was bitrotting was also a motivator for just
getting the integration done and over with.
There is still more work to do on unifying the _32/_64 splits, especially
as we have to start supporting new CPUs that sit somewhere between the
SH-4A and the SH-5 architecturally. The integration work is an ongoing
effort, and there will likely still be a bit of churn in this area
throughout 2.6.25 and in to 2.6.26.
These changes have basically been in -mm for a few iterations, and so
nothing here should be much of a surprise. We do manage to kill off quite
a bit of code in the process, and this is obviously a number that will go
up considerably as more _32/_64 split unification is done going in to
2.6.25 proper. Most of this is just a reorganization of existing in-tree
code, so there's very little in the way of new code or surprises here.
The tree in question can be found at:
git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6.git
...
