On 29 July 2015 at 10:45, Pavel Fedin <p.fe...@samsung.com> wrote: >> > because i discovered problems with 32-bit guests. They simply truncate >> > high word and end up in attempt to put PCI at 0x00000000 - 0xFFFFFFFF, >> > fail, >> > and do not work of course. So, my next version will have high MMIO >> > only for 64-bit guests. >> >> That sounds like a guest bug to me. If the device tree says that >> addresses are two-cells wide then the code reading it ought to >> read both cells, not just ignore the top 32 bits... > > May be. It's old kernel, v3.19, but still it works as it works.
It matters because you can run 32 bit guests in 64-bit-capable CPUs. I don't really want to have two different address layouts just to work around a kernel bug if we could fix the kernel bug instead... -- PMM