[llvm] [X86] Enable masked interleaved access (PR #224907)

[Simon Pilgrim via llvm-commits]

================
@@ -1,5 +1,5 @@
 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --filter "LV: Found an estimated cost of [0-9]+(.[0-9]+)? for VF 1 For instruction:\s*%i[2,4] = load i16, ptr %[a-zA-Z0-7]+, align 2" --filter "Cost of [0-9]+(.[0-9]+)? for VF [0-9]+: (REPLICATE ir<%i[24]> = load|INTERLEAVE-GROUP with factor [0-9]+)" --filter "^  ir<.* = load from index"
-; RUN: opt -passes=loop-vectorize -enable-interleaved-mem-accesses -tail-folding-policy=must-fold-tail -S -mcpu=skx --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=DISABLED_MASKED_STRIDED
+; RUN: opt -passes=loop-vectorize -enable-interleaved-mem-accesses -enable-masked-interleaved-mem-accesses=0 -tail-folding-policy=must-fold-tail -S -mcpu=skx --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=DISABLED_MASKED_STRIDED
----------------
RKSimon wrote:

If you're altering a test RUN like this I'd normally suggest keeping the original RUN as well for comparison

https://github.com/llvm/llvm-project/pull/224907