Piotr Zarycki <[email protected]> writes: > Replace the open-coded NOP loop with udelay() which was added to KVM > selftests in commit 6b878cbb87bf ("KVM: selftests: Add guest udelay() > utility for x86"). The NOP loop is CPU speed dependent while udelay() > provides a deterministic delay regardless of host CPU frequency. > > Signed-off-by: Piotr Zarycki <[email protected]> > --- > tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c | 8 +------- > 1 file changed, 1 insertion(+), 7 deletions(-) > > diff --git a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c > b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c > index c542cc4762b1..68ebd790ff41 100644 > --- a/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c > +++ b/tools/testing/selftests/kvm/x86/hyperv_tlb_flush.c > @@ -141,15 +141,9 @@ static void swap_two_test_pages(vm_paddr_t pte_gva1, > vm_paddr_t pte_gva2) > *(uint64_t *)pte_gva2 = tmp; > } > > -/* > - * TODO: replace the silly NOP loop with a proper udelay() implementation. > - */ > static inline void do_delay(void) > { > - int i; > - > - for (i = 0; i < 1000000; i++) > - asm volatile("nop"); > + udelay(100); > } > > /*
Oh, I completely forgot I added the TODO :-) Thanks for the patch! I also gave it a spin and it looks like rdtsc() from udelay() doesn't hurt. Personally, I belive we should now just drop the whole do_delay() function and use udelay(100) instead. We have only two call sites and there's no strict requirement that the delay time should match. In any case, Reviewed-by: Vitaly Kuznetsov <[email protected]> -- Vitaly
