On Thu, Apr 2, 2020 at 9:38 AM <studyinglian...@gmail.com> wrote: > > Hi, I just read the commit from > https://go-review.googlesource.com/c/go/+/166960 and it said the reason to > use lock-free is because when using mutex "Clearing must zero out the > pointer to every object in every Pool to prevent a concurrent slow path from > causing all objects to be retained. Since this happens during STW, this has > an O(# objects in Pools) effect on STW time." > I just can't understand why concurrent Get and clearPool will cause all > objects to be retained. Thank in advance for any help!
Note that that comment is referring to the old code that that series of changes is replacing. See the implementation of poolCleanup in 1.12 or earlier. In that older implementation, when pinSlow and poolCleanup run simultaneously, it's possible for pinSlow to return an entry to a pool that poolCleanup has already set to nil in the allPools slice. That would retain that pool pointer for as long as the pointer returned by pinSlow was alive. To avoid keeping everything alive, poolCleanup clears the pointers. Ian -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/CAOyqgcXMesC3MT45hrT42VOEykrb_goF_xFq-yBQ0oaPbbnoUQ%40mail.gmail.com.
