[llvm] [RISCV] Support postRA vsetvl insertion pass (PR #70549)

[Philip Reames via llvm-commits]

preames wrote:

Following up to my prior comment, thank you @BeMg and @lukel97 for the analysis on each of the cases I flagged.  For my purposes, the understanding demonstrated on the one remaining regression is sufficient.  Knowing that it's an artifact of scheduling combined with an existing optimization gives me confidence we'll be able to address this.

Given that, LGTM!  

I want to complement both of you for working through a long and challenging processing of splitting this up into reviewable pieces, landing and rebasing many many times, and driving this to conclusion.  Thank you!


https://github.com/llvm/llvm-project/pull/70549