Hi Keir. I suppose it wouldn't very difficult. I'll try it and see what it goes. But daemons live in details...
thanks On Fri, Jan 16, 2009 at 03:58:09PM +0000, Keir Fraser wrote: > Hi Isaku, and other IA64 people, > > In xen-unstable c/s 19054 I get rid of the statically-partitioned Xen heap > for x86_64. Now the domain and Xen heaps are unified, and the Xen heap > 'semantics' are a thin wrapper around the unified heap. > > It would be nice to do the same for IA64 -- like x86/64 it has 1:1 mapping > of all memory, so the static partitioning of Xen heap is not necessary. Also > I can then make the statically-partitioned Xen heap a 32-bit x86 special > case and push some of the code under arch/x86/x86_32 as is appropriate. > > What do you think? If you look, I think you'll see it's not hard to do. You > could keep using init_xenheap_pages() for now and really the hardest bit is > getting rid of xenheap_phys_start/end variables and move over to a > PGC_xen_heap page_info flag instead (needed since Xen heap pages have > different deallocation semantics). > > I suppose this might affect kdump since it won't dump Xen heap objects any > more, but I didn't want to keep the tiny Xen heap partition only for that > reason! > > -- Keir > > > -- yamahata _______________________________________________ Xen-ia64-devel mailing list Xen-ia64-devel@lists.xensource.com http://lists.xensource.com/xen-ia64-devel
