[llvm] [IR] Add llvm.masked.speculative.load intrinsic (PR #156470)

[Benjamin Maxwell via llvm-commits]

================
@@ -1031,6 +1031,68 @@ static void scalarizeMaskedVectorHistogram(const DataLayout &DL, CallInst *CI,
   ModifiedDT = true;
 }
 
+static void scalarizeMaskedSpeculativeLoad(const DataLayout &DL, CallInst *CI,
+                                           DomTreeUpdater *DTU,
+                                           bool &ModifiedDT) {
+  // For a target without speculative/first-faulting load support, we can't
----------------
MacDue wrote:

I think it'd be good to have an IR test with `-passes=scalarize-masked-mem-intrin`, similar to `llvm/test/Transforms/ScalarizeMaskedMemIntrin/AArch64/streaming-compatible-expand-masked-gather-scatter.ll`.

https://github.com/llvm/llvm-project/pull/156470