Fix bug in addAffineParallelOpDomain upper bound constraint

[bondhugula]

Summary

Fix upper bound constraint addition in addAffineParallelOpDomain; it was
off by one in the case of constants.

[bondhugula]

Adjust test case CHECK.

[bondhugula]

Update test case.

mlir/lib/Dialect/Affine/Analysis/AffineStructures.cpp

for (Value iv : parallelOp.getIVs()) {
     if (lowerBound.isConstant())
       addBound(BoundType::LB, pos, lowerBound.getSingleConstantResult());
     else if (failed(addBound(BoundType::LB, pos, lowerBound,
                              parallelOp.getLowerBoundsOperands())))
       return failure();
 
     auto upperBound = parallelOp.getUpperBoundMap(ivPos);
     if (upperBound.isConstant())
-      addBound(BoundType::UB, pos, upperBound.getSingleConstantResult());
+      addBound(BoundType::UB, pos, upperBound.getSingleConstantResult() - 1);
     else if (failed(addBound(BoundType::UB, pos, upperBound,
                              parallelOp.getUpperBoundsOperands())))
       return failure();
     ++ivPos;
   }
   return success();
 }
 

mlir/test/Transforms/memref-dependence-check.mlir

affine.if affine_set<(d0, d1, d2, d3) : ((d0 + 1) mod 8 >= 0, d0 * -8 >= 0)>(%c1, %c1, %c1, %c1){
     // expected-remark@above {{dependence from 1 to 0 at depth 1 = false}}
     // expected-remark@above {{dependence from 1 to 1 at depth 1 = false}}
   }
   return
 }
 
 // -----
 
-// CHECK-LABEL: func @affine_parallel_dep_check
-func.func @affine_parallel_dep_check() {
+func.func @affine_parallel_dep_check_1() {
   %memref_23 = memref.alloc() : memref<1x130xf32>
   %memref_25 = memref.alloc() : memref<1x130x130xf32>
   %cst = arith.constant 0.0 : f32
   affine.parallel (%arg4, %arg5) = (0, 0) to (1, 130) {
     affine.store %cst, %memref_25[%arg4, %arg5, 129] : memref<1x130x130xf32>
     // expected-remark@above {{dependence from 0 to 0 at depth 1 = false}}
     // expected-remark@above {{dependence from 0 to 0 at depth 2 = false}}
     // expected-remark@above {{dependence from 0 to 0 at depth 3 = false}}
     // expected-remark@above {{dependence from 0 to 1 at depth 1 = true}}
   }
   %memref_27 = memref.alloc() : memref<1x128x128xf32>
   affine.parallel (%arg4) = (0) to (130) {
     affine.parallel (%arg5, %arg6) = (0, 0) to (1, 130) {
       affine.load %memref_25[0, %arg4 + %arg5, %arg6] : memref<1x130x130xf32>
       // expected-remark@above {{dependence from 1 to 1 at depth 1 = false}}
       // expected-remark@above {{dependence from 1 to 1 at depth 2 = false}}
       // expected-remark@above {{dependence from 1 to 1 at depth 3 = false}}
       // expected-remark@above {{dependence from 1 to 1 at depth 4 = false}}
       // expected-remark@above {{dependence from 1 to 0 at depth 1 = false}}
     }
   }
   return
 }
+
+// -----
+
+func.func @affine_parallel_dep_check_2() {
+  %m = memref.alloc() : memref<128xf32>
+  %cst = arith.constant 0.0 : f32
+  affine.parallel (%arg4) = (0) to (127) {
+    affine.store %cst, %m[%arg4] : memref<128xf32>
+    // expected-remark@above {{dependence from 0 to 0 at depth 1 = false}}
+    // expected-remark@above {{dependence from 0 to 0 at depth 2 = false}}
+    // expected-remark@above {{dependence from 0 to 1 at depth 1 = false}}
+  }
+  affine.load %m[127]: memref<128xf32>
+  // expected-remark@above {{dependence from 1 to 1 at depth 1 = false}}
+  // expected-remark@above {{dependence from 1 to 0 at depth 1 = false}}
+  return
+}