[LV] Ignore user-specified interleave count when unsafe. (llvm#153009)

When an VF is specified via a loop hint, it will be clamped to a safe
VF or ignored if it is found to be unsafe. This is not the case for
user-specified interleave counts, which can lead to loops such as
the following with a memory dependence being vectorised with
interleaving:

```
#pragma clang loop interleave_count(4)
for (int i = 4; i < LEN; i++)
    b[i] = b[i - 4] + a[i];
```

According to [1], loop hints are ignored if they are not safe to apply.

This patch adds a check to prevent vectorisation with interleaving if
isSafeForAnyVectorWidth() returns false. This is already checked in
selectInterleaveCount().

[1]
https://llvm.org/docs/LangRef.html#llvm-loop-vectorize-and-llvm-loop-interleave

Author: kmclaughlin-arm

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

@@ -9860,6 +9860,8 @@ bool LoopVectorizePass::processLoop(Loop *L) {
   // Get user vectorization factor and interleave count.
   ElementCount UserVF = Hints.getWidth();
   unsigned UserIC = Hints.getInterleave();
+  if (UserIC > 1 && !LVL.isSafeForAnyVectorWidth())
+    UserIC = 1;
 
   // Plan how to best vectorize.
   LVP.plan(UserVF, UserIC);
@@ -9924,7 +9926,15 @@ bool LoopVectorizePass::processLoop(Loop *L) {
     VectorizeLoop = false;
   }
 
-  if (!LVP.hasPlanWithVF(VF.Width) && UserIC > 1) {
+  if (UserIC == 1 && Hints.getInterleave() > 1) {
+    assert(!LVL.isSafeForAnyVectorWidth() &&
+           "UserIC should only be ignored due to unsafe dependencies");
+    LLVM_DEBUG(dbgs() << "LV: Ignoring user-specified interleave count.\n");
+    IntDiagMsg = {"InterleavingUnsafe",
+                  "Ignoring user-specified interleave count due to possibly "
+                  "unsafe dependencies in the loop."};
+    InterleaveLoop = false;
+  } else if (!LVP.hasPlanWithVF(VF.Width) && UserIC > 1) {
     // Tell the user interleaving was avoided up-front, despite being explicitly
     // requested.
     LLVM_DEBUG(dbgs() << "LV: Ignoring UserIC, because vectorization and "

llvm/test/Transforms/LoopVectorize/AArch64/scalable-reductions.ll

@@ -417,21 +417,17 @@ for.end:                                 ; preds = %for.body, %entry
 
 ; Note: This test was added to ensure we always check the legality of reductions (end emit a warning if necessary) before checking for memory dependencies
 ; CHECK-REMARK: Scalable vectorization not supported for the reduction operations found in this loop.
-; CHECK-REMARK: vectorized loop (vectorization width: 4, interleaved count: 2)
+; CHECK-REMARK: Ignoring user-specified interleave count due to possibly unsafe dependencies in the loop.
+; CHECK-REMARK: vectorized loop (vectorization width: 4, interleaved count: 1)
 define i32 @memory_dependence(ptr noalias nocapture %a, ptr noalias nocapture readonly %b, i64 %n) {
 ; CHECK-LABEL: @memory_dependence
 ; CHECK: vector.body:
 ; CHECK: %[[LOAD1:.*]] = load <4 x i32>
 ; CHECK: %[[LOAD2:.*]] = load <4 x i32>
-; CHECK: %[[LOAD3:.*]] = load <4 x i32>
-; CHECK: %[[LOAD4:.*]] = load <4 x i32>
-; CHECK: %[[ADD1:.*]] = add nsw <4 x i32> %[[LOAD3]], %[[LOAD1]]
-; CHECK: %[[ADD2:.*]] = add nsw <4 x i32> %[[LOAD4]], %[[LOAD2]]
-; CHECK: %[[MUL1:.*]] = mul <4 x i32> %[[LOAD3]]
-; CHECK: %[[MUL2:.*]] = mul <4 x i32> %[[LOAD4]]
+; CHECK: %[[ADD1:.*]] = add nsw <4 x i32> %[[LOAD2]], %[[LOAD1]]
+; CHECK: %[[MUL1:.*]] = mul <4 x i32> %[[LOAD2]]
 ; CHECK: middle.block:
-; CHECK: %[[RDX:.*]] = mul <4 x i32> %[[MUL2]], %[[MUL1]]
-; CHECK: call i32 @llvm.vector.reduce.mul.v4i32(<4 x i32> %[[RDX]])
+; CHECK: call i32 @llvm.vector.reduce.mul.v4i32(<4 x i32> %[[MUL1]])
 entry:
   br label %for.body
 

llvm/test/Transforms/LoopVectorize/unsafe-ic-hint-remark.ll

@@ -0,0 +1,30 @@
+; RUN: opt -passes=loop-vectorize -pass-remarks-analysis=loop-vectorize -S < %s 2>&1 | FileCheck %s
+
+; Make sure the unsafe user specified interleave count is ignored.
+
+; CHECK: remark: <unknown>:0:0: Ignoring user-specified interleave count due to possibly unsafe dependencies in the loop.
+; CHECK-LABEL: @loop_distance_4
+define void @loop_distance_4(ptr %a, ptr %b) {
+entry:
+  br label %loop
+
+loop:
+  %iv = phi i64 [ 4, %entry ], [ %iv.next, %loop ]
+  %0 = getelementptr i32, ptr %b, i64 %iv
+  %arrayidx = getelementptr i8, ptr %0, i64 -16
+  %1 = load i32, ptr %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds nuw i32, ptr %a, i64 %iv
+  %2 = load i32, ptr %arrayidx2, align 4
+  %add = add nsw i32 %2, %1
+  store i32 %add, ptr %0, align 4
+  %iv.next = add nuw nsw i64 %iv, 1
+  %exitcond.not = icmp eq i64 %iv.next, 64
+  br i1 %exitcond.not, label %for.end, label %loop, !llvm.loop !1
+
+for.end:
+  ret void
+}
+
+!1 = !{!1, !2, !3}
+!2 = !{!"llvm.loop.interleave.count", i32 4}
+!3 = !{!"llvm.loop.vectorize.width", i32 4}