On 10/16/2014 10:12 AM, Dr. David Alan Gilbert wrote:
Yes; I suspect it's the vmstate_register_ram that uses TARGET_PAGE_MASK, one of the patches in my postcopy world adds a function in exec.c that returns TARGET_PAGE_BITS so that stuff that needs to know target page size can make a call to find it out; that might solve that.
Yeah, or you can just move vmstate_register_ram to migration/ram.c together with the arch_init.c bits.
Paolo