Re: [PATCH] c++: more mce_false folding from cp_fully_fold_init [PR108243]

[Jason Merrill <jason@redhat.com>]

On 2/21/23 14:10, Patrick Palka wrote:
> We should also fold the overall initializer passed to cp_fully_fold_init
> with mce_false, which enables folding of the copy-initialization of
> 'a1' in the below testcase (the initializer here is an AGGR_INIT_EXPR).
> 
> Unfortunately this doesn't help with direct- or default-initialization
> because we don't call cp_fully_fold_init in that case, and even if we
> did the initializer in that case is expressed as a bare CALL_EXPR
> instead of an AGGR_INIT_EXPR, which cp_fully_fold_init can't really
> fold.
> 
> Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK for
> trunk?

OK.

> 	PR c++/108243
> 
> gcc/cp/ChangeLog:
> 
> 	* cp-gimplify.cc (cp_fully_fold): Add an internal overload that
> 	additionally takes and propagate an mce_value parameter, and
> 	define the existing public overload in terms of it.
> 	(cp_fully_fold_init): Pass mce_false to cp_fully_fold.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* g++.dg/opt/is_constant_evaluated3.C: New test.
> ---
>   gcc/cp/cp-gimplify.cc                         | 14 +++++++----
>   .../g++.dg/opt/is_constant_evaluated3.C       | 23 +++++++++++++++++++
>   2 files changed, 33 insertions(+), 4 deletions(-)
>   create mode 100644 gcc/testsuite/g++.dg/opt/is_constant_evaluated3.C
> 
> diff --git a/gcc/cp/cp-gimplify.cc b/gcc/cp/cp-gimplify.cc
> index 32fe53521cc..5d5c6efb856 100644
> --- a/gcc/cp/cp-gimplify.cc
> +++ b/gcc/cp/cp-gimplify.cc
> @@ -2447,8 +2447,8 @@ cp_fold_rvalue (tree x)
>   
>   /* Perform folding on expression X.  */
>   
> -tree
> -cp_fully_fold (tree x)
> +static tree
> +cp_fully_fold (tree x, mce_value manifestly_const_eval)
>   {
>     if (processing_template_decl)
>       return x;
> @@ -2456,7 +2456,7 @@ cp_fully_fold (tree x)
>        have to call both.  */
>     if (cxx_dialect >= cxx11)
>       {
> -      x = maybe_constant_value (x);
> +      x = maybe_constant_value (x, /*decl=*/NULL_TREE, manifestly_const_eval);
>         /* Sometimes we are given a CONSTRUCTOR but the call above wraps it into
>   	 a TARGET_EXPR; undo that here.  */
>         if (TREE_CODE (x) == TARGET_EXPR)
> @@ -2469,6 +2469,12 @@ cp_fully_fold (tree x)
>     return cp_fold_rvalue (x);
>   }
>   
> +tree
> +cp_fully_fold (tree x)
> +{
> +  return cp_fully_fold (x, mce_unknown);
> +}
> +
>   /* Likewise, but also fold recursively, which cp_fully_fold doesn't perform
>      in some cases.  */
>   
> @@ -2477,7 +2483,7 @@ cp_fully_fold_init (tree x)
>   {
>     if (processing_template_decl)
>       return x;
> -  x = cp_fully_fold (x);
> +  x = cp_fully_fold (x, mce_false);
>     cp_fold_data data (ff_mce_false);
>     cp_walk_tree (&x, cp_fold_r, &data, NULL);
>     return x;
> diff --git a/gcc/testsuite/g++.dg/opt/is_constant_evaluated3.C b/gcc/testsuite/g++.dg/opt/is_constant_evaluated3.C
> new file mode 100644
> index 00000000000..0a1e46e5638
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/opt/is_constant_evaluated3.C
> @@ -0,0 +1,23 @@
> +// PR c++/108243
> +// { dg-do compile { target c++11 } }
> +// { dg-additional-options "-O -fdump-tree-original" }
> +
> +struct A {
> +  constexpr A(int n) : n(n), m(__builtin_is_constant_evaluated()) { }
> +  constexpr A() : A(42) { }
> +  int n, m;
> +};
> +
> +int main() {
> +  A a1 = {42};
> +  A a2{42};
> +  A a3(42);
> +  A a4;
> +  A a5{};
> +}
> +
> +// { dg-final { scan-tree-dump "a1 = {\\.n=42, \\.m=0}" "original" } }
> +// { dg-final { scan-tree-dump "a2 = {\\.n=42, \\.m=0}" "original" { xfail *-*-* } } }
> +// { dg-final { scan-tree-dump "a3 = {\\.n=42, \\.m=0}" "original" { xfail *-*-* } } }
> +// { dg-final { scan-tree-dump "a4 = {\\.n=42, \\.m=0}" "original" { xfail *-*-* } } }
> +// { dg-final { scan-tree-dump "a5 = {\\.n=42, \\.m=0}" "original" { xfail *-*-* } } }