[llvm] [LV] Use wide lane masks as the canonical form when tail-folding & interleaving (PR #209484)

[David Sherwood via llvm-commits]

================
@@ -1,124 +1,88 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^middle.block:" --version 4
-; RUN: opt -S --passes=loop-vectorize -tail-folding-policy=must-fold-tail -enable-wide-lane-mask -force-vector-interleave=1 < %s | FileCheck %s -check-prefix CHECK-UF1
-; RUN: opt -S --passes=loop-vectorize -tail-folding-policy=must-fold-tail -enable-wide-lane-mask -force-vector-interleave=4 < %s | FileCheck %s -check-prefix CHECK-UF4
-; RUN: opt -S --passes=loop-vectorize -enable-wide-lane-mask -tail-folding-policy=must-fold-tail < %s | FileCheck %s -check-prefix CHECK-TF
-; RUN: opt -S --passes=forceattrs,loop-vectorize -enable-wide-lane-mask -tail-folding-policy=must-fold-tail -force-attribute=optsize < %s | FileCheck %s -check-prefix CHECK-UF1
+; RUN: opt -S --passes=loop-vectorize -tail-folding-policy=must-fold-tail -force-vector-interleave=1 < %s | FileCheck %s -check-prefix CHECK-UF1
+; RUN: opt -S --passes=loop-vectorize -tail-folding-policy=must-fold-tail -force-vector-interleave=4 < %s | FileCheck %s -check-prefix CHECK-UF4
+; RUN: opt -S --passes=forceattrs,loop-vectorize -tail-folding-policy=must-fold-tail -force-attribute=optsize < %s | FileCheck %s -check-prefix CHECK-UF1
 
 target triple = "aarch64-unknown-linux"
 
 define void @scalable_wide_active_lane_mask(ptr noalias %dst, ptr readonly %src, i64 %n) #0 {
 ; CHECK-UF1-LABEL: define void @scalable_wide_active_lane_mask(
 ; CHECK-UF1-SAME: ptr noalias [[DST:%.*]], ptr readonly [[SRC:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
 ; CHECK-UF1-NEXT:  entry:
-; CHECK-UF1-NEXT:    br label [[VECTOR_PH1:%.*]]
+; CHECK-UF1-NEXT:    br label [[VECTOR_PH:%.*]]
----------------
david-arm wrote:

Perhaps manually regenerate these CHECK lines in a pre-commit to reduce the diff here?

https://github.com/llvm/llvm-project/pull/209484