[PATCH] D46008: [X86][AArch64][NFC] Add tests for vector masked merge unfolding

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri added inline comments.


================
Comment at: test/CodeGen/X86/machine-cp.ll:13
 ; CHECK-NEXT:  ## %bb.2: ## %while.body.preheader
-; CHECK-NEXT:  movl %esi, %edx
+; CHECK-NEXT:    movl %esi, %edx
 ; CHECK-NEXT:    .p2align 4, 0x90
----------------
spatel wrote:
> Unrelated white-space diff? This patch is NFC already, so I think it's ok to keep this here or separate if I'm missing something.
Yep. This function is affected by the dependent revision, so why not fix this space right away?
¯\_(ツ)_/¯


================
Comment at: test/CodeGen/X86/unfold-masked-merge-vector-variablemask.ll:2
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=-sse,-sse2 < %s | FileCheck %s --check-prefix=CHECK-BASELINE
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,-sse2 < %s | FileCheck %s --check-prefix=CHECK-SSE
----------------
spatel wrote:
> Why is this config interesting? IMO, it just distracts from the cases that we do care about, but I may not be seeing it.
I think we do want to check that we don't do anything stupid in `andn`-less case,
much like without `BMI` in scalar case. 


================
Comment at: test/CodeGen/X86/unfold-masked-merge-vector-variablemask.ll:731-733
+; ============================================================================ ;
+; 128-bit vector width
+; ============================================================================ ;
----------------
spatel wrote:
> I still doubt the value of everything above here (illegal vector types), but if you think it shows/proves something...
I do think it does, note that these illegal vectors **improve** in D46528.



Repository:
  rL LLVM

https://reviews.llvm.org/D46008