On Thu, Nov 15, 2012 at 10:10 AM, Jon Masters <jcm(a)redhat.com> wrote:
On 11/15/2012 03:59 AM, Peter Robinson wrote:
> Looking at post 3.7 I'm already starting to poke around and prep
> things. Similar to x86-32 I'm aiming to basically have an armv7 and an
> armv7-PAE kernel. The PAE kernel will support LPAE and the various
> virt things that are starting to land in the kernel now so it will be
> solely for A15 chipsets. I presume the Exynos5 supports those
Nitpick: Let's call it LPAE just because it's not Intel's PAE. The
That was actually the plan.... I'm not after the colour of the bikeshed here!
primary A15 target today should be Exynos IMO, while we can build a
Calxeda "highbank" (they're still calling it that upstream, but I
believe they've announced the actual name, though I won't quote it just
in case that isn't the case) A15 kernel once that's upstream. I looked
at the patches Rob posted and it seemed pretty straightforward.
It will be what ever is supported in the mainline kernel. Hence the
reason I'm looking at 3.8 as I believe that is when the HW will start
to land. There's already OMAP5 support in 3.7 to some degree so it
might be that one first. Ultimately it will only ever be a unified
kernel as I refuse to go backwards here so SoC specific vendor is
irrelevant in the discussion :-)
As you know, I'll have A15 hardware soon to poke also.
Don't rub it in, you need to get me the HW too!