Semantics of kernel ABI-nums

[Andy Whitcroft]

On Fri, Nov 11, 2011 at 06:19:09AM +0800, Eric Miao wrote:
> kernel-team Cc'ed
> 
> On Fri, Nov 11, 2011 at 6:12 AM, Steve Magoun
> <steve.magoun at canonical.com> wrote:
> > Hi,
> >
> > Can someone help me answer this question from HP?
> >
> > Thanks,
> > Steve
> >
> >
> > -------- Original Message --------
> > Subject:        Semantics of kernel ABI-nums
> >
> >
> > In the ARM world, it appears that the kernel for each different SoC or board
> > is considered to be a build of a different kernel “flavour”. Are the kernel
> > ABI-nums meant to be assigned globally across all flavours of kernels, or
> > are they per kernel-flavour? If the latter, then can you explain why
> > debian.n900/etc/getabis references other kernel flavours, namely
> > linaro-mx51, linaro-omap, and linaro-xexpress? Is there documentation on how
> > the kernel flavour scheme has been applied to the ARM SoC world?

Each tree has its own ABI sequence number.  This is a community branch
(perhaps from ogra) and likely they have simmply not configured etc/getabis
correctly.

-apw