[PATCH] D123933: X86: Remove UsesMSVCFloatingPoint from MachineModuleInfo

Matt Arsenault via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Apr 18 12:08:07 PDT 2022


arsenm added a comment.

In D123933#3457324 <https://reviews.llvm.org/D123933#3457324>, @rnk wrote:

> This is sort of reasonable, but I feel like people might disagree so I won't go ahead and stamp it.
>
> I feel like the existing code adheres to the philosophy that the IR will not be available after codegen, and that ISel is responsible for looking at the IR and saving all these analysis results. There's something undesirable about looking back at the IR once we reach the end of assembly emission.

Right, but we're nowhere near that goal. There are tons of backlinks to the IR and targets looking directly at the underlying IR. There are a good number of MachineFunction predicates that look at attributes on the IR for example. Ironically I think this helps move towards the direction of cutting out the IR by making MIR serialization more obviously reliable


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123933/new/

https://reviews.llvm.org/D123933



More information about the llvm-commits mailing list