[PATCH] D110745: Redefine deref(N) attribute family as point-in-time semantics (aka deref-at-point)

[Philip Reames via Phabricator via llvm-commits]

reames added a comment.
Herald added a subscriber: jeroen.dobbelaere.

@nikic ping on previous question.  It's been a month, and this has been LGTMed.  Without response, I plan to land this.

In D110745#3038848 <https://reviews.llvm.org/D110745#3038848>, @xbolva00 wrote:

> This really needs to be properly benchmarked.

This has been benchmarked on every workload I care about, and shows no interesting regressions.   Unfortunately, those are all non-public Java workloads,

On the C/C++ side, I don't have a ready environment in which to run anything representative.  From the semantic change, I wouldn't expect C++ to show much difference, and besides, this is fixing a long standing fairly major correctness issue.  If you have particular suites you care about, please run them and share results.

At this point, I strongly lean towards committing and letting regressions be reported.  We might revert, or we might simply fix forward depending on what comes up


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D110745/new/

https://reviews.llvm.org/D110745