[PATCH] D46241: [CodeGen] Recognize more cases of zero initialization

[Serge Pavlov via Phabricator via cfe-commits]

sepavloff marked an inline comment as done.
sepavloff added inline comments.


================
Comment at: lib/CodeGen/CGExprConstant.cpp:1403
+  if (auto *IL = dyn_cast_or_null<InitListExpr>(Init)) {
+    if (InitTy->isConstantArrayType()) {
+      for (auto I : IL->inits())
----------------
rjmccall wrote:
> Do you actually care if it's an array initialization instead of a struct/enum initialization?
If this code is enabled for for records too, some tests start to fail. For instance, the code:
```
union { int i; double f; } u2 = { };
```
produces output:
```
%union.anon = type { double }
@u2 = global %union.anon zeroinitializer, align 4
```
while previously it produced:
```
@u2 = global { i32, [4 x i8] } { i32 0, [4 x i8] undef }, align 4
```
The latter looks more correct.


================
Comment at: lib/CodeGen/CGExprConstant.cpp:1413
+  } else if (!Init->isEvaluatable(CE.CGM.getContext())) {
+    return false;
+  } else if (InitTy->hasPointerRepresentation()) {
----------------
rjmccall wrote:
> Aren't the null-pointer and integer-constant-expression checks below already checking this?  Also, `isEvaluatable` actually computes the full value internally (as an `APValue`), so if you're worried about the memory and compile-time effects of producing such a value, you really shouldn't call it.
> 
> You could reasonably move this entire function to be a method on `Expr` that takes an `ASTContext`.
Comment for `EvaluateAsRValue` says that it tries calculate expression agressively. Indeed, for the code:
```
  decltype(nullptr) null();
  int *p = null();
```
compiler ignores potential side effect of `null()` and removes the call, leaving only zero initialization. `isNullPointerConstant` behaves similarly.


================
Comment at: lib/CodeGen/CGExprConstant.cpp:1417
+    if (Init->EvaluateAsRValue(ResVal, CE.CGM.getContext()))
+      return ResVal.Val.isLValue() && ResVal.Val.isNullPointer();
+  } else {
----------------
rjmccall wrote:
> There's a `isNullPointerConstant` method (you should use `NPC_NeverValueDependent`).
It make code more readable. Thank you!


Repository:
  rC Clang

https://reviews.llvm.org/D46241