Juan Quintela wrote: [Fri Jun 02 2006, 12:14:25PM EDT] > > - run the xen-mkbuildtree-pre hook following applying the xen > > patch. This touches a couple files on ia64 but doesn't > > interfere with any patching down the line. (I and others have > > been submitting patches to xen-ia64-devel and xen-devel to > > further reduce its function, but it should be harmless as-is.) > > I think that this one is not acceptable :( > We compile from the same source on _all_ architectures. If the source > changes from architecture to architecture, we are broken :( > > Notice also that I do my compile testing as something like:i > > cd tree > for arch in $CONFIGS; do > make O=../$ARCH all > done Ok, we can make this work, no problem :-) Here is the current xen-mkbuildtree-pre in xen-unstable: http://xenbits.xensource.com/xen-unstable.hg?f=a2d784a51f25;file=linux-2.6-xen-sparse/arch/ia64/xen-mkbuildtree-pre Instead of running it only on ia64, we can run it on all architectures. Since the recent patches, the last remaining files created by that script are: include/asm-ia64/xen/asm-xsi-offsets.h drivers/xen/core/xenia64_init.c Neither of those affect any architecture other than ia64. I'm hoping to move the second one to a better home next week. The first one is less trivial since it's auto-generated in the hypervisor build. > > - update numerous hardcoded references to /boot to usetu > > %{image_install_path} instead. But make an exception for the > > xenU kernels which never need to be available for system > > boot. > > Uh, oh. If everything happens as I hope, after this weekend, xenU > kernel is history. xen0 & xenU kernels are going to be the same > kernel :p You need another solution here. I haven't studied the ia64 > efi thing to be able to suggest anything, sorry. No problem. I'm very happy to switch to a unified xen kernel. > > - build the hypervisor with %{?_smp_mflags} (it works fine) > > > > %if %{buildup} > > -BuildKernel %make_target %kernel_image > > +BuildKernel %{make_target} %{kernel_image} > > %endif > > Any reason why you do this changes? in space separated cases, it > shouldn't be needed the use of {} Nope, it just seemed more consistent with, for example, the %if conditional preceding it :-) Aron
Attachment:
pgp5nMSG83Rfc.pgp
Description: PGP signature
