On Mon, 2015-08-03 at 11:58 +0200, Andreas Beckmann wrote: > On 2015-08-03 02:21, Luca Boccassi wrote: > > Testing the build of nvidia-kernel-dkms in an armhf chroot (via qemu static > > binaries) results in build failure > > for kernels 3.16 (jessie) and 4.0 (sid). Cannot test 4.1 as the armhf build > > of linux-kbuild-4.1 is missing. > > Does any other of the armhf linux-headers-* work? Oh wait, there are not > that many any more :-)
yep, the only other one is armmp-lpae, but that only adds a couple more configs, the rest are all meta-packages. > > I already addressed 2 problems and committed to the SVN repo (patch > > nv-linux.h and enable dma_ops for arm in > > conftest.h, but the final problem is that the armhf kernel ships with > > CONFIG_XEN enabled [1], without CONFIG_PARAVIRT, > > and this is not supported. > > Sounds like a question for the kernel maintainers ... > > > what about ignoring CONFIG_XEN on armhf in nv-linux.h: > > -#if defined(CONFIG_XEN) && !defined(CONFIG_PARAVIRT) > +#if defined(CONFIG_XEN) && !defined(CONFIG_PARAVIRT) && > !defined(CONFIG_whatever_indicates_armhf) Thought about that, but that would probably cause havok in case the driver is actually ran in a VM on top of Xen, wouldn't it (however unlikely that might be)? It won't fail to load anymore since we don't support it, and probably die later. Nonetheless, I'll give it a shot and see how it behaves later tonight or tomorrow. Kind regards, Luca Boccassi
signature.asc
Description: This is a digitally signed message part