https://gcc.gnu.org/g:06728901a03376bdeafc3ff48115113a0ffe169e
commit r16-2411-g06728901a03376bdeafc3ff48115113a0ffe169e
Author: Eric Botcazou <ebotca...@adacore.com>
Date: Tue Jul 8 11:05:19 2025 +0200
ada: Fix assertion failure on aggregate with controlled component
The assertion is:
pragma Assert (Side_Effect_Free (L));
in Make_Tag_Ctrl_Assignment and demonstrates that the sequence:
Remove_Side_Effects (L);
pragma Assert (Side_Effect_Free (L));
does not hold in this case.
What happens is that Remove_Side_Effects uses a renaming to remove the side
effects of L but, at the end, the renamed object is substituted back for the
renamed object in the node by Expand_Renaming, which is invoked because the
Is_Renaming_Of_Object flag is set on the renaming after Evaluate_Name has
been invoked on its Name.
This is a general discrepancy between Evaluate_Name and Side_Effect_Free of
Exp_Util, coming from the call to Safe_Unchecked_Type_Conversion present in
Side_Effect_Free in this case. The long term goal is probably to remove the
call but, in the meantime, this change is sufficient to fix the failure.
gcc/ada/ChangeLog:
* exp_util.adb (Safe_Unchecked_Type_Conversion): Always return True
if the expression is the prefix of an N_Selected_Component.
Diff:
---
gcc/ada/exp_util.adb | 13 ++++---------
1 file changed, 4 insertions(+), 9 deletions(-)
diff --git a/gcc/ada/exp_util.adb b/gcc/ada/exp_util.adb
index 80e9a8101166..ac49c6b38267 100644
--- a/gcc/ada/exp_util.adb
+++ b/gcc/ada/exp_util.adb
@@ -13771,11 +13771,12 @@ package body Exp_Util is
-- The above requirements should be documented in Sinfo ???
function Safe_Unchecked_Type_Conversion (Exp : Node_Id) return Boolean is
+ Pexp : constant Node_Id := Parent (Exp);
+
Otyp : Entity_Id;
Ityp : Entity_Id;
Oalign : Uint;
Ialign : Uint;
- Pexp : constant Node_Id := Parent (Exp);
begin
-- If the expression is the RHS of an assignment or object declaration
@@ -13793,18 +13794,12 @@ package body Exp_Util is
return True;
-- If the expression is the prefix of an N_Selected_Component we should
- -- also be OK because GCC knows to look inside the conversion except if
- -- the type is discriminated. We assume that we are OK anyway if the
- -- type is not set yet or if it is controlled since we can't afford to
- -- introduce a temporary in this case.
+ -- also be OK because GCC knows to look inside the conversion.
elsif Nkind (Pexp) = N_Selected_Component
and then Prefix (Pexp) = Exp
then
- return No (Etype (Pexp))
- or else not Is_Type (Etype (Pexp))
- or else not Has_Discriminants (Etype (Pexp))
- or else Is_Constrained (Etype (Pexp));
+ return True;
end if;
-- Set the output type, this comes from Etype if it is set, otherwise we
