Skip to content

R
riscv-gcc-1
Commit 540d8610 by Ed Schonberg Committed by Arnaud Charlet
Options

sem_ch6.adb: Move Build_Body_To_Inline... 

2014-07-29  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch6.adb: Move Build_Body_To_Inline,
	Check_And_Buid_Body_To_Inline, and Cannot_Inline to package Inline.
	* exp_ch6.adb: Mode Expand_Inlined_Body to package Inline.
	* inline.ads, inline.adb: Package now contains subprograms that
	implement front-end inlining.  No functional changes, no test
	needed.

From-SVN: r213179
parent b973629e
Showing with 87 additions and 36 deletions
gcc/ada/ChangeLog
2014-07-29 Ed Schonberg <schonberg@adacore.com>
* sem_ch6.adb: Move Build_Body_To_Inline,
Check_And_Buid_Body_To_Inline, and Cannot_Inline to package Inline.
* exp_ch6.adb: Mode Expand_Inlined_Body to package Inline.
* inline.ads, inline.adb: Package now contains subprograms that
implement front-end inlining. No functional changes, no test
needed.
2014-07-29 Robert Dewar <dewar@adacore.com>
* exp_dbug.adb, g-expect.adb, sem_elab.adb: Minor typo fix.
......
gcc/ada/inline.adb
This source diff could not be displayed because it is too large. You can view the blob instead.
gcc/ada/inline.ads
......@@ -6,7 +6,7 @@
-- --
-- S p e c --
-- --
-- Copyright (C) 1992-2013, Free Software Foundation, Inc. --
-- Copyright (C) 1992-2014, Free Software Foundation, Inc. --
-- --
-- GNAT is free software; you can redistribute it and/or modify it under --
-- terms of the GNU General Public License as published by the Free Soft- --
......@@ -23,7 +23,7 @@
-- --
------------------------------------------------------------------------------
-- This module handles two kinds of inlining activity:
-- This module handles three kinds of inlining activity:
-- a) Instantiation of generic bodies. This is done unconditionally, after
-- analysis and expansion of the main unit.
......@@ -35,6 +35,13 @@
-- of them uses a workpile algorithm, but they are called independently from
-- Frontend, and thus are not mutually recursive.
-- Front-end inlining for subprograms marked Inline_Always. This is primarily
-- an expansion activity that is performed for performance reasons, and when
-- the target does not use the gcc backend. Inline_Always can also be used
-- in the context of GNATprove, to perform source transformations to simplify
-- proof obligations. The machinery used in both cases is similar, but there
-- are fewer restrictions on the source of subprograms in the latter case.
with Alloc;
with Opt; use Opt;
with Sem; use Sem;
......@@ -122,7 +129,11 @@ package Inline is
Table_Increment => Alloc.Pending_Instantiations_Increment,
Table_Name => "Pending_Descriptor");
-----------------
Inlined_Calls : Elist_Id := No_Elist;
Backend_Calls : Elist_Id := No_Elist;
-- List of frontend inlined calls and inline calls passed to the backend
-----------------
-- Subprograms --
-----------------
......@@ -147,12 +158,76 @@ package Inline is
-- At end of compilation, analyze the bodies of all units that contain
-- inlined subprograms that are actually called.
procedure Build_Body_To_Inline (N : Node_Id; Subp : Entity_Id);
-- If a subprogram has pragma Inline and inlining is active, use generic
-- machinery to build an unexpanded body for the subprogram. This body is
-- subsequently used for inline expansions at call sites. If subprogram can
-- be inlined (depending on size and nature of local declarations) this
-- function returns true. Otherwise subprogram body is treated normally.
-- If proper warnings are enabled and the subprogram contains a construct
-- that cannot be inlined, the offending construct is flagged accordingly.
procedure Cannot_Inline
(Msg : String;
N : Node_Id;
Subp : Entity_Id;
Is_Serious : Boolean := False);
-- This procedure is called if the node N, an instance of a call to
-- subprogram Subp, cannot be inlined. Msg is the message to be issued,
-- which ends with ? (it does not end with ?p?, this routine takes care of
-- the need to change ? to ?p?). Temporarily the behavior of this routine
-- depends on the value of -gnatd.k:
--
-- * If -gnatd.k is not set (ie. old inlining model) then if Subp has
-- a pragma Always_Inlined, then an error message is issued (by
-- removing the last character of Msg). If Subp is not Always_Inlined,
-- then a warning is issued if the flag Ineffective_Inline_Warnings
-- is set, adding ?p to the msg, and if not, the call has no effect.
--
-- * If -gnatd.k is set (ie. new inlining model) then:
-- - If Is_Serious is true, then an error is reported (by removing the
-- last character of Msg);
--
-- - otherwise:
--
-- * Compiling without optimizations if Subp has a pragma
-- Always_Inlined, then an error message is issued; if Subp is
-- not Always_Inlined, then a warning is issued if the flag
-- Ineffective_Inline_Warnings is set (adding p?), and if not,
-- the call has no effect.
--
-- * Compiling with optimizations then a warning is issued if the
-- flag Ineffective_Inline_Warnings is set (adding p?); otherwise
-- no effect since inlining may be performed by the backend.
procedure Check_And_Build_Body_To_Inline
(N : Node_Id;
Spec_Id : Entity_Id;
Body_Id : Entity_Id);
-- Spec_Id and Body_Id are the entities of the specification and body of
-- the subprogram body N. If N can be inlined by the frontend (supported
-- cases documented in Check_Body_To_Inline) then build the body-to-inline
-- associated with N and attach it to the declaration node of Spec_Id.
procedure Check_Body_For_Inlining (N : Node_Id; P : Entity_Id);
-- If front-end inlining is enabled and a package declaration contains
-- inlined subprograms, load and compile the package body to collect the
-- bodies of these subprograms, so they are available to inline calls.
-- N is the compilation unit for the package.
procedure Expand_Inlined_Call
(N : Node_Id;
Subp : Entity_Id;
Orig_Subp : Entity_Id);
-- If called subprogram can be inlined by the front-end, retrieve the
-- analyzed body, replace formals with actuals and expand call in place.
-- Generate thunks for actuals that are expressions, and insert the
-- corresponding constant declarations before the call. If the original
-- call is to a derived operation, the return type is the one of the
-- derived operation, but the body is that of the original, so return
-- expressions in the body must be converted to the desired type (which
-- is simply not noted in the tree without inline expansion).
procedure Remove_Dead_Instance (N : Node_Id);
-- If an instantiation appears in unreachable code, delete the pending
-- body instance.
......
gcc/ada/sem_ch6.ads
......@@ -68,39 +68,6 @@ package Sem_Ch6 is
-- and body declarations. Returns the defining entity for the
-- specification N.
procedure Cannot_Inline
(Msg : String;
N : Node_Id;
Subp : Entity_Id;
Is_Serious : Boolean := False);
-- This procedure is called if the node N, an instance of a call to
-- subprogram Subp, cannot be inlined. Msg is the message to be issued,
-- which ends with ? (it does not end with ?p?, this routine takes care of
-- the need to change ? to ?p?). Temporarily the behavior of this routine
-- depends on the value of -gnatd.k:
--
-- * If -gnatd.k is not set (ie. old inlining model) then if Subp has
-- a pragma Always_Inlined, then an error message is issued (by
-- removing the last character of Msg). If Subp is not Always_Inlined,
-- then a warning is issued if the flag Ineffective_Inline_Warnings
-- is set, adding ?p to the msg, and if not, the call has no effect.
--
-- * If -gnatd.k is set (ie. new inlining model) then:
-- - If Is_Serious is true, then an error is reported (by removing the
-- last character of Msg);
--
-- - otherwise:
--
-- * Compiling without optimizations if Subp has a pragma
-- Always_Inlined, then an error message is issued; if Subp is
-- not Always_Inlined, then a warning is issued if the flag
-- Ineffective_Inline_Warnings is set (adding p?), and if not,
-- the call has no effect.
--
-- * Compiling with optimizations then a warning is issued if the
-- flag Ineffective_Inline_Warnings is set (adding p?); otherwise
-- no effect since inlining may be performed by the backend.
procedure Check_Conventions (Typ : Entity_Id);
-- Ada 2005 (AI-430): Check that the conventions of all inherited and
-- overridden dispatching operations of type Typ are consistent with their
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment