> On Jun 19, 2017, at 9:58 AM, Joe Groff <jgr...@apple.com> wrote: > >> On Jun 15, 2017, at 8:30 PM, Greg Parker via swift-dev <swift-dev@swift.org> >> wrote: >> >> swift_once is not the most efficient solution here - we don't need separate >> once token storage in this case - but it's fine for now. If necessary we can >> get rid of the once token requirement in a backwards-compatible way. >> >> (You can do without a separate token if (1) your storage is two pointers or >> less in size, (2) your storage is sufficiently well-aligned that it does not >> cross cache line boundaries, (3) your uninitialized value is zero, and (4) >> your initialized value is not zero. If all of these are true then you can >> perform the same trick that swift-once and dispatch-once use on the once >> token, but apply it directly to the data instead. And if your data is a >> single pointer you can do it faster than dispatch-once on weakly-ordered >> platforms that are not Alpha.) > > Does the transition have to be strictly zero -> nonzero? I thought it just > had to be "original value at the time page was mapped into process" -> > "something else". For metadata initialization on 64-bit Apple, we could > perhaps go from relative offset of metadata accessor function (which should > always be <2GB in Mach-O) to absolute metadata address (which should always > be >4GB on 64-bit Darwin).
That ought to work too. -- Greg Parker gpar...@apple.com <mailto:gpar...@apple.com> Runtime Wrangler
_______________________________________________ swift-dev mailing list swift-dev@swift.org https://lists.swift.org/mailman/listinfo/swift-dev
