Reduce unroll of constant bounds loop with TripCount that is not modulo of unroll factor.

[evstupac]

Summary

Allow the type of unroll only when AllowRuntime is true.
Limit Count to DefaultRuntimeCount.

[escha]

LGTM

[zansari]

I'd prefer we used a different way to control this, like perhaps another option.

The way it is now overloads the documented meaning of "-unroll-runtime", which may be confusing and undesirable/restricting.

For example, if I disable -unroll-runtime to stop unrolling loops that don't have compile time constants (as it's documented to do), with this patch, I'll also stop some loops that DO have compile time constants from being unrolled.

We need to either re-purpose that option, or introduce a new one. I'd prefer the latter for minimal impact.

Zia.

lib/Transforms/Scalar/LoopUnrollPass.cpp

if (!AllowPartial && !CountSetExplicitly) {
       return false;
     }
     if (UP.PartialThreshold != NoThreshold &&
         UnrolledSize > UP.PartialThreshold) {
       // Reduce unroll count to be modulo of TripCount for partial unrolling.
       Count = (std::max(UP.PartialThreshold, 3u) - 2) / (LoopSize - 2);
       while (Count != 0 && TripCount % Count != 0)
         Count--;
-      if (Count <= 1) {
+      if (AllowRuntime && Count <= 1) {
         // If there is no Count that is modulo of TripCount, set Count to
         // largest power-of-two factor that satisfies the threshold limit.
-        Count = (std::max(UP.PartialThreshold, 3u) - 2) / (LoopSize - 2);
+        // As we'll create fixup loop, do the type of unrolling only if
+        // runtime unrolling is allowed.
+        Count = DefaultUnrollRuntimeCount;
         UnrolledSize = (LoopSize - 2) * Count + 2;
         while (Count != 0 && UnrolledSize > UP.PartialThreshold) {
           Count >>= 1;
           UnrolledSize = (LoopSize - 2) * Count + 2;
         }
       }
     }
   } else if (Unrolling == Runtime) {

test/Transforms/LoopUnroll/partial-unroll-const-bounds.ll

-; RUN: opt < %s -S -unroll-threshold=20 -loop-unroll -unroll-allow-partial | FileCheck %s
+; RUN: opt < %s -S -unroll-threshold=20 -loop-unroll -unroll-allow-partial -unroll-runtime | FileCheck %s
 
 ; The Loop TripCount is 9. However unroll factors 3 or 9 exceed given threshold.
 ; The test checks that we choose a smaller, power-of-two, unroll count and do not give up on unrolling.
 
 ; CHECK: for.body:
 ; CHECK: store
 ; CHECK: for.body.1:
 ; CHECK: store