On Thu, Feb 01, 2024, Shaoqin Huang wrote: > > > /* > > > * We reserve page table for 2 times of extra dirty mem which > > > * will definitely cover the original (1G+) test range. Here > > > @@ -825,6 +832,13 @@ static void run_test(enum vm_guest_mode mode, void *arg) > > > sync_global_to_guest(vm, iteration); > > > } > > > + /* > > > + * > > > + * Before we set the host_quit, let the vcpu has time to run, to make > > > + * sure we consume the sem_vcpu_stop and the vcpu consume the > > > + * sem_vcpu_cont, to keep the semaphore balance. > > > + */ > > > + usleep(p->interval * 1000); > > > > Please no. "Wait for a while" is never a complete solution for fixing races. > > In rare cases, adding a delay might be the only sane workaround, but I doubt that's > > the case here. > > If that's the case. I guess I should keep the current solution. Except you > have any better solution, please let me know. Unfortunately I don't have a better solution, and I don't have cycles to stare at this deeply to figure out what how to make the synchronization rock solid. Sorry :-/
