[LoopInfo] Pointer to stack object may not be loop invariant in a coroutine function

A coroutine function may be split to ramp function and resume function,
and they have different stack frames, so a pointer to stack objects may
have different addresses depending on where it is used, so it's not a
loop invariant.

Author: weiguozhi

llvm/lib/Analysis/LoopInfo.cpp

@@ -59,8 +59,16 @@ static cl::opt<bool, true>
 //
 
 bool Loop::isLoopInvariant(const Value *V) const {
-  if (const Instruction *I = dyn_cast<Instruction>(V))
-    return !contains(I);
+  if (const Instruction *I = dyn_cast<Instruction>(V)) {
+    // If V is a pointer to stack object and F is a coroutine function, then V
+    // may not be loop invariant because the ramp function and resume function
+    // have different stack frames.
+    if (isa<AllocaInst>(I) &&
+        I->getParent()->getParent()->isPresplitCoroutine())
+      return false;
+    else
+      return !contains(I);
+  }
   return true; // All non-instructions are loop invariant
 }
 

llvm/test/Transforms/LICM/licm-coroutine.ll

@@ -0,0 +1,78 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt < %s -passes=licm -S | FileCheck %s
+
+; %fca.0 and %fca.1 should not be hoisted out of the loop because the ramp
+; function and resume function have different stack frames, so %pointer1 and
+; %pointer2 have different values before and after @llvm.coro.suspend.
+
+define ptr @f(i32 %n) presplitcoroutine {
+; CHECK-LABEL: define ptr @f(
+; CHECK-SAME: i32 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK-NEXT:  [[ENTRY:.*]]:
+; CHECK-NEXT:    [[POINTER1:%.*]] = alloca ptr, align 8
+; CHECK-NEXT:    [[POINTER2:%.*]] = alloca ptr, align 8
+; CHECK-NEXT:    [[ID:%.*]] = call token @llvm.coro.id(i32 0, ptr null, ptr null, ptr null)
+; CHECK-NEXT:    [[SIZE:%.*]] = call i32 @llvm.coro.size.i32()
+; CHECK-NEXT:    [[ALLOC:%.*]] = call ptr @malloc(i32 [[SIZE]])
+; CHECK-NEXT:    [[HDL:%.*]] = call noalias ptr @llvm.coro.begin(token [[ID]], ptr [[ALLOC]])
+; CHECK-NEXT:    br label %[[LOOP:.*]]
+; CHECK:       [[LOOP]]:
+; CHECK-NEXT:    [[N_VAL:%.*]] = phi i32 [ [[N]], %[[ENTRY]] ], [ [[INC:%.*]], %[[RESUME:.*]] ]
+; CHECK-NEXT:    [[INC]] = add nsw i32 [[N_VAL]], 1
+; CHECK-NEXT:    call void @print(i32 [[N_VAL]])
+; CHECK-NEXT:    [[TMP0:%.*]] = call i8 @llvm.coro.suspend(token none, i1 false)
+; CHECK-NEXT:    switch i8 [[TMP0]], label %[[SUSPEND_LOOPEXIT:.*]] [
+; CHECK-NEXT:      i8 0, label %[[RESUME]]
+; CHECK-NEXT:      i8 1, label %[[CLEANUP:.*]]
+; CHECK-NEXT:    ]
+; CHECK:       [[RESUME]]:
+; CHECK-NEXT:    [[FCA_0:%.*]] = insertvalue [2 x ptr] poison, ptr [[POINTER1]], 0
+; CHECK-NEXT:    [[FCA_1:%.*]] = insertvalue [2 x ptr] [[FCA_0]], ptr [[POINTER2]], 1
+; CHECK-NEXT:    call void @foo([2 x ptr] [[FCA_1]])
+; CHECK-NEXT:    br label %[[LOOP]]
+; CHECK:       [[CLEANUP]]:
+; CHECK-NEXT:    [[MEM:%.*]] = call ptr @llvm.coro.free(token [[ID]], ptr [[HDL]])
+; CHECK-NEXT:    call void @free(ptr [[MEM]])
+; CHECK-NEXT:    br label %[[SUSPEND:.*]]
+; CHECK:       [[SUSPEND_LOOPEXIT]]:
+; CHECK-NEXT:    br label %[[SUSPEND]]
+; CHECK:       [[SUSPEND]]:
+; CHECK-NEXT:    [[UNUSED:%.*]] = call i1 @llvm.coro.end(ptr [[HDL]], i1 false, token none)
+; CHECK-NEXT:    ret ptr [[HDL]]
+;
+entry:
+  %pointer1 = alloca ptr
+  %pointer2 = alloca ptr
+  %id = call token @llvm.coro.id(i32 0, ptr null, ptr null, ptr null)
+  %size = call i32 @llvm.coro.size.i32()
+  %alloc = call ptr @malloc(i32 %size)
+  %hdl = call noalias ptr @llvm.coro.begin(token %id, ptr %alloc)
+  br label %loop
+
+loop:
+  %n.val = phi i32 [ %n, %entry ], [ %inc, %resume ]
+  %inc = add nsw i32 %n.val, 1
+  call void @print(i32 %n.val)
+  %0 = call i8 @llvm.coro.suspend(token none, i1 false)
+  switch i8 %0, label %suspend [i8 0, label %resume
+  i8 1, label %cleanup]
+
+resume:
+  %fca.0 = insertvalue [2 x ptr] poison, ptr %pointer1, 0
+  %fca.1 = insertvalue [2 x ptr] %fca.0, ptr %pointer2, 1
+  call void @foo([2 x ptr] %fca.1)
+  br label %loop
+
+cleanup:
+  %mem = call ptr @llvm.coro.free(token %id, ptr %hdl)
+  call void @free(ptr %mem)
+  br label %suspend
+suspend:
+  %unused = call i1 @llvm.coro.end(ptr %hdl, i1 false, token none)
+  ret ptr %hdl
+}
+
+declare void @free(ptr)
+declare ptr @malloc(i32)
+declare void @print(i32)
+declare void @foo([2 x ptr])