[PATCH] D92706: [coroutine] should disable inline before calling coro split

Yifeng Dong via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Sun Dec 13 23:03:47 PST 2020


dongAxis1944 added inline comments.


================
Comment at: llvm/test/Transforms/Coroutines/coro-inline.ll:1
+; RUN: opt < %s -always-inline -barrier -coro-early -barrier -coro-split -S | FileCheck %s
+; RUN: opt < %s -enable-new-pm  -always-inline -coro-early -coro-split  -S | FileCheck %s
----------------
aeubanks wrote:
> dongAxis1944 wrote:
> > lxfind wrote:
> > > aeubanks wrote:
> > > > Are these `-barrier`s necessary? The test passes without them. And no other test in all of LLVM uses it.
> > > @dongAxis1944, do you know the context on why this was initially added?
> > this is because we want to run always_inline, then coro-early, split at last. using barrier, will keep this sequence well.
> Does it matter if pass A runs on all SCCs, then pass B runs on all SCCs, or if for each SCC, pass A runs then pass B runs before going to the next SCC?
the actual actions is  that llvm will do always inline  for all scc, before doing early and split, so we add barrier.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D92706/new/

https://reviews.llvm.org/D92706



More information about the llvm-commits mailing list