[PATCH] D28489: [CodeGen] Generalize MacroFusion for any instructions pair

[Matthias Braun via Phabricator via llvm-commits]

MatzeB added a comment.

In https://reviews.llvm.org/D28489#640628, @atrick wrote:

> Would it make sense to have a TII.mayFuseWithPrecedingInstr() to avoid testing all DAG edges? The DAG is quadratic, but this a rare opportunity.


Or let targets subclass or write their own scheduledag mutation with an apropriate search strategy instead of the TII callback?


Repository:
  rL LLVM

https://reviews.llvm.org/D28489