[PATCH] D121905: [AArch64][SVE] Fix lowering of "fcmp ueq/one" when using SVE

[Paul Walker via Phabricator via llvm-commits]

paulwalker-arm added inline comments.


================
Comment at: llvm/test/CodeGen/AArch64/sve-select.ll:550-552
+; CHECK-NEXT:    fcmlt p1.s, p0/z, z0.s, #0.0
+; CHECK-NEXT:    fcmgt p0.s, p0/z, z0.s, #0.0
+; CHECK-NEXT:    sel p0.b, p0, p0.b, p1.b
----------------
Not the fault of this patch just observational but this looks weird now you've fixed the code generation.  This comes from D119424 whereby the original `oeq` is inverted.  That patch uses `getSetCCInverse` which says `Return the operation corresponding to !(X op Y)`.

Perhaps I've got my logic wrong but I think  `!(X oeq Y)` => `X une Y` whereas this output clearly shows we're getting `X one Y`. What do you think @david-arm is my logic silly? or is there a potential bug to investigate?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D121905/new/

https://reviews.llvm.org/D121905