[llvm] [LoopVectorize] Teach LoopVectorizationLegality about more early exits (PR #107004)

[Paul Walker via llvm-commits]

================
@@ -2445,8 +2445,11 @@ bool LoopAccessInfo::analyzeLoop(AAResults *AA, const LoopInfo *LI,
       // vectorize a loop if it contains known function calls that don't set
       // the flag. Therefore, it is safe to ignore this read from memory.
       auto *Call = dyn_cast<CallInst>(&I);
-      if (Call && getVectorIntrinsicIDForCall(Call, TLI))
-        continue;
+      if (Call) {
+        NumCalls++;
----------------
paulwalker-arm wrote:

Based on the use, is this too broad? For example, would it prevent loops that call sqrt intrinsics from being vectorisable?  What about keeping a count of non-store instructions that can modify memory? This can be added to the code below.

https://github.com/llvm/llvm-project/pull/107004