[LAA] Update early-exit test to cover last valid & first invalid access.

Make sure tests cover loops accessing the last valid and the first
invalid memory location. Note that the test
`@all_exits_dominate_latch_countable_exits_at_most_1000_iterations_kno`
has been modified to execute at most 1000 iterations; it previously
executed 1001 iterations.

Author: fhahn

llvm/test/Analysis/LoopAccessAnalysis/early-exit-runtime-checks.ll

@@ -59,6 +59,65 @@ e.2:
   ret void
 }
 
+define void @all_exits_dominate_latch_countable_exits_at_most_501_iterations_known_deref(ptr dereferenceable(2000) %A, ptr dereferenceable(2000) %B) {
+; CHECK-LABEL: 'all_exits_dominate_latch_countable_exits_at_most_501_iterations_known_deref'
+; CHECK-NEXT:    loop.header:
+; CHECK-NEXT:      Memory dependences are safe with run-time checks
+; CHECK-NEXT:      Dependences:
+; CHECK-NEXT:      Run-time memory checks:
+; CHECK-NEXT:      Check 0:
+; CHECK-NEXT:        Comparing group GRP0:
+; CHECK-NEXT:          %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
+; CHECK-NEXT:        Against group GRP1:
+; CHECK-NEXT:          %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
+; CHECK-NEXT:      Grouped accesses:
+; CHECK-NEXT:        Group GRP0:
+; CHECK-NEXT:          (Low: %B High: (2004 + %B))
+; CHECK-NEXT:            Member: {%B,+,4}<nuw><%loop.header>
+; CHECK-NEXT:        Group GRP1:
+; CHECK-NEXT:          (Low: %A High: (2004 + %A))
+; CHECK-NEXT:            Member: {%A,+,4}<nuw><%loop.header>
+; CHECK-EMPTY:
+; CHECK-NEXT:      Non vectorizable stores to invariant address were not found in loop.
+; CHECK-NEXT:      SCEV assumptions:
+; CHECK-EMPTY:
+; CHECK-NEXT:      Expressions re-written:
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
+  %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
+  %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
+  %l = load i32, ptr %gep.A, align 4
+  store i32 0, ptr %gep.B, align 4
+  %cntable.c.1 = icmp ult i64 %iv, 1000
+  %iv.next = add nuw nsw i64 %iv, 1
+  br i1 %cntable.c.1, label %b2, label %e.1
+
+b2:
+  %uncntable.c.0 = icmp eq i32 %l, 0
+  br i1 %uncntable.c.0, label %e.2, label %b3
+
+b3:
+  %cntable.c.2 = icmp eq i64 %iv.next, 501
+  br i1 %cntable.c.2, label %cleanup4, label %latch
+
+latch:
+  br label %loop.header
+
+cleanup4:
+  ret void
+
+e.1:
+  ret void
+
+e.2:
+  ret void
+}
+
+
 define void @all_exits_dominate_latch_countable_exits_at_most_500_iterations_not_known_deref(ptr %A, ptr %B) {
 ; CHECK-LABEL: 'all_exits_dominate_latch_countable_exits_at_most_500_iterations_not_known_deref'
 ; CHECK-NEXT:    loop.header:
@@ -130,6 +189,64 @@ define i32 @all_exits_dominate_latch_countable_exits_at_most_1000_iterations_kno
 ; CHECK-NEXT:          %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
 ; CHECK-NEXT:      Grouped accesses:
 ; CHECK-NEXT:        Group GRP0:
+; CHECK-NEXT:          (Low: %B High: (4000 + %B)<nuw>)
+; CHECK-NEXT:            Member: {%B,+,4}<nuw><%loop.header>
+; CHECK-NEXT:        Group GRP1:
+; CHECK-NEXT:          (Low: %A High: (4000 + %A)<nuw>)
+; CHECK-NEXT:            Member: {%A,+,4}<nuw><%loop.header>
+; CHECK-EMPTY:
+; CHECK-NEXT:      Non vectorizable stores to invariant address were not found in loop.
+; CHECK-NEXT:      SCEV assumptions:
+; CHECK-EMPTY:
+; CHECK-NEXT:      Expressions re-written:
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %latch ]
+  %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
+  %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
+  %l = load i32, ptr %gep.A, align 4
+  store i32 0, ptr %gep.B, align 4
+  %cntable.c.1 = icmp ult i64 %iv, 999
+  br i1 %cntable.c.1, label %b2, label %e.1
+
+b2:
+  %uncntable.c.0 = icmp eq i32 %l, 0
+  br i1 %uncntable.c.0, label %e.2, label %b3
+
+b3:
+  %iv.next = add nuw nsw i64 %iv, 1
+  %cntable.c.2 = icmp eq i64 %iv.next, 2000
+  br i1 %cntable.c.2, label %e.0, label %latch
+
+latch:
+  br label %loop.header
+
+e.0:
+  ret i32 0
+
+e.1:
+  ret i32 1
+
+e.2:
+  ret i32 2
+}
+
+define i32 @all_exits_dominate_latch_countable_exits_at_most_1001_iterations_known_deref(ptr dereferenceable(4000) %A, ptr dereferenceable(4000) %B) {
+; CHECK-LABEL: 'all_exits_dominate_latch_countable_exits_at_most_1001_iterations_known_deref'
+; CHECK-NEXT:    loop.header:
+; CHECK-NEXT:      Memory dependences are safe with run-time checks
+; CHECK-NEXT:      Dependences:
+; CHECK-NEXT:      Run-time memory checks:
+; CHECK-NEXT:      Check 0:
+; CHECK-NEXT:        Comparing group GRP0:
+; CHECK-NEXT:          %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
+; CHECK-NEXT:        Against group GRP1:
+; CHECK-NEXT:          %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
+; CHECK-NEXT:      Grouped accesses:
+; CHECK-NEXT:        Group GRP0:
 ; CHECK-NEXT:          (Low: %B High: (4004 + %B))
 ; CHECK-NEXT:            Member: {%B,+,4}<nuw><%loop.header>
 ; CHECK-NEXT:        Group GRP1:
@@ -188,10 +305,10 @@ define i32 @all_exits_dominate_latch_countable_exits_at_most_1000_iterations_not
 ; CHECK-NEXT:          %gep.A = getelementptr inbounds i32, ptr %A, i64 %iv
 ; CHECK-NEXT:      Grouped accesses:
 ; CHECK-NEXT:        Group GRP0:
-; CHECK-NEXT:          (Low: %B High: (4004 + %B))
+; CHECK-NEXT:          (Low: %B High: (4000 + %B))
 ; CHECK-NEXT:            Member: {%B,+,4}<nuw><%loop.header>
 ; CHECK-NEXT:        Group GRP1:
-; CHECK-NEXT:          (Low: %A High: (4004 + %A))
+; CHECK-NEXT:          (Low: %A High: (4000 + %A))
 ; CHECK-NEXT:            Member: {%A,+,4}<nuw><%loop.header>
 ; CHECK-EMPTY:
 ; CHECK-NEXT:      Non vectorizable stores to invariant address were not found in loop.
@@ -208,7 +325,7 @@ loop.header:
   %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
   %l = load i32, ptr %gep.A, align 4
   store i32 0, ptr %gep.B, align 4
-  %cntable.c.1 = icmp ult i64 %iv, 1000
+  %cntable.c.1 = icmp ult i64 %iv, 999
   br i1 %cntable.c.1, label %b2, label %e.1
 
 b2: