[PATCH] D107677: Prevent machine licm if remattable with a vreg use

[Stanislav Mekhanoshin via Phabricator via llvm-commits]

rampitec added a comment.

In D107677#2936366 <https://reviews.llvm.org/D107677#2936366>, @dmgreen wrote:

> This does effect the ARM backend, apparently at some point it obtained the ability to hoist VCTP instructions which take a register use. I'm not sure if that really fits the definition of trivially rematerializable though, from the code comment on isTriviallyReMaterializable. (But I'm not sure that comment is up to date.)

I am also looking at the D87280 <https://reviews.llvm.org/D87280> which has enabled rematerialization of VCTP. It looks like the tests there want it to be rematerialized, effectively sunk into a loop body, and not hoisted. I am not sure if such hoisting (which will happen less frequently if ever after this patch) considered a good thing. My impression from the D87280 <https://reviews.llvm.org/D87280> and its description that it would not be desired.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D107677/new/

https://reviews.llvm.org/D107677