[PATCH] D149768: [ArgumentPromotion] Bail if any callers are minsize

[Shoaib Meenai via Phabricator via llvm-commits]

smeenai added a comment.

Do you have data showing size improvements on your end? We're observing size regressions at `-Oz` across the board instead for Meta's Android apps (we build with LTO, but I haven't tested non-LTO builds). It'll take me a while to come up with a standalone example, but some of the largest regressions are in folly::Optional and std::make_shared, which matches ArgumentPromotion's description of helping with templated code.

If you don't data to the contrary, could we revert this? If you do have cases where this shows overall improvements, what can I provide to help tune this to not be a substantial regression for us?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D149768/new/

https://reviews.llvm.org/D149768