[LV] Fix versioning-for-unit-stide of loops with small trip count

[Ayal]

Summary

This patch fixes D81345 and PR46652.

If a loop with a small trip count is compiled w/o -Os/-Oz, Loop Access Analysis still generates runtime checks for unit strides that will version the loop.

In such cases, the loop vectorizer should either re-run the analysis or bail-out from vectorizing the loop, as done prior to D81345. The latter is chosen for now as the former requires refactoring.

[fhahn]

LGTM, thanks!

In such cases, the loop vectorizer should either re-run the analysis or bail-out from vectorizing the loop, as done prior to D81345. The latter is chosen for now as the former requires refactoring.

As already discussed in D81345, ideally LV would have more flexibility to drive LAA, but this requires non-trivial refactoring. Which we should do, but until then the patch looks like a reasonable fix to the crash.

llvm/test/Transforms/LoopVectorize/optsize.ll

239

Better to use a non-undef constant/value?

[Ayal]

In D83470#2142057, @fhahn wrote:

LGTM, thanks!

In such cases, the loop vectorizer should either re-run the analysis or bail-out from vectorizing the loop, as done prior to D81345. The latter is chosen for now as the former requires refactoring.

As already discussed in D81345, ideally LV would have more flexibility to drive LAA, but this requires non-trivial refactoring. Which we should do, but until then the patch looks like a reasonable fix to the crash.

Indeed, seems like LV would need to re-run analyzeLoop(), which would affect other users.

llvm/test/Transforms/LoopVectorize/optsize.ll

239

Sure, done, thanks.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

if (!PSE.getUnionPredicate().getPredicates().empty()) {
     reportVectorizationFailure("Runtime SCEV check is required with -Os/-Oz",
         "runtime SCEV checks needed. Enable vectorization of this "
         "loop with '#pragma clang loop vectorize(enable)' when "
         "compiling with -Os/-Oz",
         "CantVersionLoopWithOptForSize", ORE, TheLoop);
     return true;
   }
 
-  assert(Legal->getLAI()->getSymbolicStrides().empty() &&
-         "Specializing for stride == 1 under -Os/-Oz");
+  // FIXME: Avoid specializing for stride==1 instead of bailing out.
+  if (!Legal->getLAI()->getSymbolicStrides().empty()) {
+    reportVectorizationFailure("Runtime stride check for small trip count",
+        "runtime stride == 1 checks needed. Enable vectorization of "
+        "this loop without such check by compiling with -Os/-Oz",
+        "CantVersionLoopWithOptForSize", ORE, TheLoop);
+    return true;
+  }
 
   return false;
 }
 
 Optional<unsigned> LoopVectorizationCostModel::computeMaxVF(unsigned UserVF,
                                                             unsigned UserIC) {
   if (Legal->getRuntimePointerChecking()->Need && TTI.hasBranchDivergence()) {
     // TODO: It may by useful to do since it's still likely to be dynamically

llvm/test/Transforms/LoopVectorize/optsize.ll

for.body:
   %iv.next = add nuw nsw i32 %iv, 1
   %exitcond = icmp eq i32 %iv.next, 1025
   br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !15
 
 for.end:
   ret void
 }
 
+; PR46652: Check that the need for stride==1 check prevents vectorizing a loop
+; having tiny trip count, when compiling w/o -Os/-Oz.
+; CHECK-LABEL: @pr46652
+; CHECK-NOT: vector.scevcheck
+; CHECK-NOT: vector.body
+; CHECK-LABEL: for.body
+
+@g = external global [1 x i16], align 1
+
+define void @pr46652(i16 %stride) {
+entry:
+  br label %for.body
+
+for.body:                                        ; preds = %for.body, %entry
+  %l1.02 = phi i16 [ 1, %entry ], [ %inc9, %for.body ]
+  %mul = mul nsw i16 %l1.02, %stride
+  %arrayidx6 = getelementptr inbounds [1 x i16], [1 x i16]* @g, i16 0, i16 %mul
+  %0 = load i16, i16* %arrayidx6, align 1
+  %inc9 = add nuw nsw i16 %l1.02, 1
+  %exitcond.not = icmp eq i16 %inc9, 16
+  br i1 %exitcond.not, label %for.end, label %for.body
+
+for.end:                                        ; preds = %for.body
+  ret void
+}
+
 !llvm.module.flags = !{!0}
 !0 = !{i32 1, !"ProfileSummary", !1}
 !1 = !{!2, !3, !4, !5, !6, !7, !8, !9}
 !2 = !{!"ProfileFormat", !"InstrProf"}
 !3 = !{!"TotalCount", i64 10000}
 !4 = !{!"MaxCount", i64 10}
 !5 = !{!"MaxInternalCount", i64 1}
 !6 = !{!"MaxFunctionCount", i64 1000}