On Tue, Apr 21, 2026 at 11:10:22PM +0300, Dan Carpenter wrote: > > How did you test this? >
I built the test using:
make -C tools/testing/selftests/kvm ARCH=x86_64 x86/hyperv_tlb_flush
And ran it 100 times to verify it passes consistently
results: 100/100 passed, 0 failed
average time: 466ms
For comparison, the original NOP loop averaged ~537ms on the same machine.

