Reland [SimplifyCFG] FoldBranchToCommonDest: lift use-restriction on bonus…

Authored by lebedev.ri on Nov 26 2020, 11:21 PM.


Reland [SimplifyCFG] FoldBranchToCommonDest: lift use-restriction on bonus instructions

This was orginally committed in 2245fb8aaa1c1f85f53f7b19a1ee3ac69b1a1dfe.
but was immediately reverted in f3abd54958ab90ba7c100d3fa936a3ce0dd2ad04
because of a PHI handling issue.

Original commit message:

  1. It doesn't make sense to enforce that the bonus instruction is only used once in it's basic block. What matters is whether those user instructions fit within our budget, sure, but that is another question.
  2. It doesn't make sense to enforce that said bonus instructions are only used within their basic block. Perhaps the branch condition isn't using the value computed by said bonus instruction, and said bonus instruction is simply being calculated to be used in successors?

So iff we can clone bonus instructions, to lift these restrictions,
we just need to carefully update their external uses
to use the new cloned instructions.

Notably, this transform (even without this change) appears to be
poison-unsafe as per alive2, but is otherwise (including the patch) legal.

We don't introduce any new PHI nodes, but only "move" the instructions
around, i'm not really seeing much potential for extra cost modelling
for the transform, especially since now we allow at most one such
bonus instruction by default.

This causes the fold to fire +11.4% more (13216 -> 14725)
as of vanilla llvm test-suite + RawSpeed.

The motivational pattern is IEEE-754-2008 Binary16->Binary32
extension code:
^ that should be a switch, but it is not now: https://godbolt.org/z/bvja5v
That being said, even thought this seemed like this would fix it: https://godbolt.org/z/xGq3TM
apparently that fold is happening somewhere else afterall,
so something else also has a similar 'artificial' restriction.


lebedev.riNov 27 2020, 1:47 AM
rG40188063296f: [NFC][SimplifyCFG] FoldBranchToCommonDest: add one more test with PHI