On Tue, 2024-11-05 at 11:36 -0800, Christoph Lameter (Ampere) wrote: > CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe. > > > > On Tue, 5 Nov 2024, Haris Okanovic wrote: > > > +/** > > + * smp_vcond_load_relaxed() - (Spin) wait until an expected value at address > > + * with no ordering guarantees. Spins until `(*addr & mask) == val` or > > + * `nsecs` elapse, and returns the last observed `*addr` value. > > + * > > + * @nsecs: timeout in nanoseconds > > Please use an absolute time in nsecs instead of a timeout. I went with relative time because it clock agnostic. I agree deadline is nicer because it can propagate down layers of functions, but it pins the caller to single time base. > You do not know > what will happen to your execution thread until the local_clock_noinstr() > is run. Not sure what you mean. Could you perhaps give an example where it would break? > > One alternative is to do timekeeping with delay() in all cases, to decouple from sched/clock.
