public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "Marc Poulhiès" <poulhies@adacore.com>
To: gcc-patches@gcc.gnu.org
Cc: Eric Botcazou <ebotcazou@adacore.com>
Subject: [COMMITTED] ada: Elide temporary for aliased array with unconstrained nominal subtype
Date: Tue,  7 Nov 2023 10:20:33 +0100	[thread overview]
Message-ID: <20231107092034.3906691-1-poulhies@adacore.com> (raw)

From: Eric Botcazou <ebotcazou@adacore.com>

When the array is initialized with the result of a call to a function whose
result type is unconstrained, then the result is allocated with its bounds,
so the array can be rewritten as a renaming of the result in this case too.

gcc/ada/

	* exp_ch3.adb (Expand_N_Object_Declaration): Fold initialization
	expression of Nominal_Subtype_Is_Constrained_Array constant into
	the computation of Rewrite_As_Renaming and remove the constant.
	Set it to True for an aliased array with unconstrained nominal
	subtype if the subtype of the expression is also unconstrained.

Tested on x86_64-pc-linux-gnu, committed on master.

---
 gcc/ada/exp_ch3.adb | 47 +++++++++++++++++++++++----------------------
 1 file changed, 24 insertions(+), 23 deletions(-)

diff --git a/gcc/ada/exp_ch3.adb b/gcc/ada/exp_ch3.adb
index 511d4c09b22..f88ac7e6542 100644
--- a/gcc/ada/exp_ch3.adb
+++ b/gcc/ada/exp_ch3.adb
@@ -7348,13 +7348,6 @@ package body Exp_Ch3 is
       Rewrite_As_Renaming : Boolean := False;
       --  Whether to turn the declaration into a renaming at the end
 
-      Nominal_Subtype_Is_Constrained_Array : constant Boolean :=
-        Comes_From_Source (Obj_Def)
-        and then Is_Array_Type (Typ) and then Is_Constrained (Typ);
-      --  Used to avoid rewriting as a renaming for constrained arrays,
-      --  which is only a problem for source arrays; others have the
-      --  correct bounds (see below).
-
    --  Start of processing for Expand_N_Object_Declaration
 
    begin
@@ -8098,10 +8091,25 @@ package body Exp_Ch3 is
 
               Is_Entity_Name (Original_Node (Obj_Def))
 
-                --  Nor if it is effectively an unconstrained declaration
+                --  If we have "X : S := ...;", and S is a constrained array
+                --  subtype, then we cannot rename, because renamings ignore
+                --  the constraints of S, so that would change the semantics
+                --  (sliding would not occur on the initial value). This is
+                --  only a problem for source objects though, the others have
+                --  the correct bounds.
+
+                and then not (Comes_From_Source (Obj_Def)
+                               and then Is_Array_Type (Typ)
+                               and then Is_Constrained (Typ))
+
+                --  Moreover, if we have "X : aliased S := "...;" and S is an
+                --  unconstrained array type, then we can rename only if the
+                --  initialization expression has an unconstrained subtype too,
+                --  because the bounds must be present within X.
 
                 and then not (Is_Array_Type (Typ)
-                               and then Is_Constr_Subt_For_UN_Aliased (Typ))
+                               and then Is_Constr_Subt_For_UN_Aliased (Typ)
+                               and then Is_Constrained (Etype (Expr_Q)))
 
                 --  We may use a renaming if the initialization expression is a
                 --  captured function call that meets a few conditions.
@@ -8109,23 +8117,16 @@ package body Exp_Ch3 is
                 and then
                   (Is_Renamable_Function_Call (Expr_Q)
 
-                   --  Or else if it is a variable with OK_To_Rename set
+                    --  Or else if it is a variable with OK_To_Rename set
 
-                   or else (OK_To_Rename_Ref (Expr_Q)
-                             and then not Special_Ret_Obj)
+                    or else (OK_To_Rename_Ref (Expr_Q)
+                              and then not Special_Ret_Obj)
 
-                   --  Or else if it is a slice of such a variable
-
-                   or else (Nkind (Expr_Q) = N_Slice
-                             and then OK_To_Rename_Ref (Prefix (Expr_Q))
-                             and then not Special_Ret_Obj))
-
-                --  If we have "X : S := ...;", and S is a constrained array
-                --  subtype, then we cannot rename, because renamings ignore
-                --  the constraints of S, so that would change the semantics
-                --  (sliding would not occur on the initial value).
+                    --  Or else if it is a slice of such a variable
 
-                and then not Nominal_Subtype_Is_Constrained_Array;
+                    or else (Nkind (Expr_Q) = N_Slice
+                              and then OK_To_Rename_Ref (Prefix (Expr_Q))
+                              and then not Special_Ret_Obj));
 
             --  If the type needs finalization and is not inherently limited,
             --  then the target is adjusted after the copy and attached to the
-- 
2.42.0


                 reply	other threads:[~2023-11-07  9:20 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20231107092034.3906691-1-poulhies@adacore.com \
    --to=poulhies@adacore.com \
    --cc=ebotcazou@adacore.com \
    --cc=gcc-patches@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).