On Monday 04 January 2016 14:50:34 Andre Przywara wrote: > This new config enables KVM (and thus LPAE) on top of > multi_v7_defconfig. This creates a kernel that can _host_ KVM guests, > for guests a multi_v7_defconfig with virtio drivers is sufficient. > Since LPAE kernels are not compatible with non-LPAE capable hardware, > this has to be separate from the existing multi_v7_defconfig. > > The purpose of this new defconfig is both to simplify creation of KVM > host kernels as well as to give the LPAE option some better testing > coverage, as the enlarged dma_addr_t and phys_addr_t types create some > headache to some drivers. > > Signed-off-by: Andre Przywara <andre.przywara@xxxxxxx> > I agree about the purpose, but I want something better here: * Use config fragments to just set the few extra options that we can't enable otherwise, and have a shared config file with the normal multi_v7. There is still an ongoing question about big-endian defconfig, which has exactly the same problem, and we should have a common approach for all combinations. We might also want to do this in a way that lets us build v6/v6k/v7/v7ve/v8 kernels in big-endian and little-endian all from the same defconfig file, with the set of platforms getting smaller as we go to higher arch levels. * I really want to see the ARMv7VE dependencies sorted out. Arnd _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
