On Thu, May 12, 2022 at 12:02:30AM +0900, Wonhyuk Yang wrote: > > The original intent of that tracepoint was to trace when pages were > > removed from the buddy list. That would suggest this untested patch on > > top of yours as a simplication; > > > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > > index 0351808322ba..66a70b898130 100644 > > --- a/mm/page_alloc.c > > +++ b/mm/page_alloc.c > > @@ -2476,6 +2476,8 @@ struct page *__rmqueue_smallest(struct zone *zone, unsigned int order, > > del_page_from_free_list(page, zone, current_order); > > expand(zone, page, order, current_order, migratetype); > > set_pcppage_migratetype(page, migratetype); > > + trace_mm_page_alloc_zone_locked(page, order, migratetype, > > + pcp_allowed_order(order) && migratetype < MIGRATE_PCPTYPES); > > return page; > > } > > Interestingly, my first approach was quite similar your suggestion. But I > noticed that there can be a request whose migration type is MOVABLE > and alloc_flags doen't have ALLOC_CMA. In that case, page are marked > as percpu-refill even though it was allocated from buddy-list directly. > Is there no problem if we just ignore this case? > I assume you are referring to the case where CMA allocations are being balanced between regular and CMA areas. I think it's relatively harmless if percpu_refill field is not 100% accurate for that case. There are also cases like the percpu list is too small to hold a THP and it's not a percpu_refill either. If 100% accuracy is an issue, I would prefer renaming it to percpu_eligible or just deleting it instead of adding complexity for a tracepoint. The main value of that tracepoint is determining what percentage of allocations are potentially contending on zone lock at a particular time. -- Mel Gorman SUSE Labs
