[PATCH] D94644: [Inliner] Inline alwaysinline calls first
Mircea Trofin via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Wed Jan 13 22:12:50 PST 2021
mtrofin added a comment.
In D94644#2497180 <https://reviews.llvm.org/D94644#2497180>, @aeubanks wrote:
> An alternative is to run the mandatory inliner in the same CGSCC pipeline as everything else, but the way InlineAdvisorAnalysis is setup made it hard to implement
I don't remember there being a particular challenge with the advisor - it was more that there was the other value of having a more accurate 'view' of the overall shape of the functions, if we want to include more context in inlining decision making. My preference would be to have a robust solution that addresses both this and the previous problem.
So I make sure I understand it correctly: it sounds like in PR48734, the callee wasn't first optimized through the function simplification pipeline, is that correct?
Let's try the mandatory inliner in the same CGSCC - happy to do that. Is PR48734 an emergency (I can take a look at this tomorrow morning PST)
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D94644/new/
https://reviews.llvm.org/D94644
More information about the cfe-commits
mailing list