[PATCH] D80510: Update BFI when handling inlined landing / eh pad

Haibo Huang via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Jun 1 16:16:55 PDT 2020


hhb added a comment.

In D80510#2067572 <https://reviews.llvm.org/D80510#2067572>, @mtrofin wrote:

> In D80510#2067484 <https://reviews.llvm.org/D80510#2067484>, @hhb wrote:
>
> > In D80510#2067446 <https://reviews.llvm.org/D80510#2067446>, @hhb wrote:
> >
> > > D75341 <https://reviews.llvm.org/D75341> and comments there seem highly related.
> >
> >
> > Actually that change supposes to fix this kind of non-deterministic. Not sure why it does not work as intended.
> >
> > We may still want to copy BFI when split BBs. But other concerns should have been solved by CallbackVH...
>
>
> Huh.. added davidxl for more background, too, he remembered something along the lines of this non-determinism being addressed somehow (probably via the CallbackVH)


Aha it is simply because BFICallbackVH is new enough to not include in Android tree (that's what I test against). I will try again with latest llvm.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D80510/new/

https://reviews.llvm.org/D80510





More information about the llvm-commits mailing list