On 7/17/2026 11:51 AM, Michael Tokarev wrote: > On 3/13/26 01:44, Pierrick Bouvier wrote: >> Code concerned is under a kvm_enabled() guard. >> >> Signed-off-by: Pierrick Bouvier <[email protected]> >> --- >> hw/vfio/listener.c | 4 ---- >> 1 file changed, 4 deletions(-) >> >> diff --git a/hw/vfio/listener.c b/hw/vfio/listener.c >> index 960da9e0a93..31c3113f8fb 100644 >> --- a/hw/vfio/listener.c >> +++ b/hw/vfio/listener.c >> @@ -20,9 +20,7 @@ >> #include "qemu/osdep.h" >> #include <sys/ioctl.h> >> -#ifdef CONFIG_KVM >> #include <linux/kvm.h> >> -#endif >> #include <linux/vfio.h> > > This now fails to build on sparc64: > > In file included from ../../hw/vfio/listener.c:23: > /build/reproducible-path/qemu-11.1.0~rc0+ds/linux-headers/linux/ > kvm.h:16:10: fatal error: asm/kvm.h: No such file or directory > 16 | #include <asm/kvm.h> > | ^~~~~~~~~~~ >
It seems like it was unnoticed because we don't test cross builds for this platform. We can't guess something not tested is broken. Seems like it should be added to CI if we want to keep it working on these hosts in the future. > I wonder if this is a sparc problem or qemu problem. > QEMU issue. > I don't think kvm has been enabled on sparc, ever.> Right. If I'm correct, it should imply that vfio is not supported as well, which means that hw/vfio should not be built at all on this host. meson.build correctly adds files only when CONFIG_VFIO is set. So you can a look at why CONFIG_VFIO is set on sparc64 host first. By fixing this, and it will fix the build compilation by not including those files. For more safety in the future, feel free to add a sparc64 cross container (debian-sparc64-cross.docker) + a CI build job using it. If you're stuck, feel free to ping for help :). > Thanks, > > /mjt > Regards, Pierrick
