On Tue, Nov 15, 2022, Colton Lewis wrote: > Allocate additional space for latency samples. This has been separated > out to call attention to the additional VM memory allocation. A blurb in the changelog is sufficient, no need to split allocation and use into two patches. I would actually collapse all three into one. The changes aren't so big that errors will be difficult to bisect, and without the final printing, the other changes are useless for all intents and purposes, i.e. if for some reason we want to revert the sampling, it will be all or nothing. I do think it makes sense to separate the system counter stuff to a separate patch (and land it in generic code), e.g. add helpers to read the system counter from the guest and convert the result to nanoseconds in a separate patch.