[PATCH] D138602: [WIP] Alwaysinliner time explosion with new pass manager

[Arthur Eubanks via Phabricator via llvm-commits]

aeubanks added a comment.

if alwaysinline is causing issues with recursive code, can you not mark the recursive code with alwaysinline (if I'm understanding the issue correctly)? that seems bad in general, we should respect alwaysinline as much as possible

when you say that the exponential cost model "fix" doesn't apply to alwaysinline, do you mean that the alwaysinliner (a separate pass) doesn't do the exponent cost modelling (fixable)? or that alwaysinline always trumps the cost multiplier (what I mentioned above)?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138602/new/

https://reviews.llvm.org/D138602