On Tue, Nov 21, 2017 at 05:58:12PM +0100, Igor Mammedov wrote: > anyways, if you prefer having mixed endianness here, I won't mind much, > as impact of test case is much smaller so we'll see down the road > if it became any better.
I agree here. > (as far as we continue going in direction > of using host byteorder within main qemu acpi code, which we were > doing by gradually converting acpi structures to build_append_foo API) So APIs are absolutely a good alternative. If we get rid of packed structures in main qemu code, we could convert tests to whatever makes sense for tests. -- MST
