On Tue, Apr 29, 2025 at 07:05:35AM -0700, Sean Christopherson wrote: > On Tue, Apr 29, 2025, Peter Zijlstra wrote: > > On Mon, Apr 28, 2025 at 10:13:31AM -0700, Sean Christopherson wrote: > > > On Sat, Apr 26, 2025, Peter Zijlstra wrote: > > > > On Wed, Apr 16, 2025 at 10:38:59AM +0200, Peter Zijlstra wrote: > > > > > > > > > Yeah, I finally got there. I'll go cook up something else. > > > > > > > > Sean, Paolo, can I once again ask how best to test this fastop crud? > > > > > > Apply the below, build KVM selftests, > > > > Patch applied, my own hackery applied, host kernel built and booted, > > foce_emulation_prefix set, but now I'm stuck at this seemingly simple > > step.. > > > > $ cd tools/testing/selftests/kvm/ > > $ make > > ... metric ton of fail ... > > > > Clearly I'm doing something wrong :/ > > Did you install headers in the top level directory? I.e. make > headers_install.
No, of course not :-) I don't use the top directory to build anything, ever. All my builds are into build directories, using make O=foo. This allows me to do parallel builds for multiple architectures etc. Also, much easier to wipe a complete build directory than it is to clean out the top level dir. > The selftests build system was change a while back to require users to > manually > install headers (I forget why, but it is indeed annoying). Bah, I remember NAK-ing that. Clearly the selftest people don't want selftests to be usable :-( Anyway, mingo build me a copy of the fastop selftest, and aside from a few stupid mistakes, I seem to now pass it \o/ I'll attempt a Changelog and update the hyper-v patches and then post the lot.
