rt_gccstream/gcc/ada/sem_ch13.adb

4766 lines
168 KiB
Ada

------------------------------------------------------------------------------
-- --
-- GNAT COMPILER COMPONENTS --
-- --
-- S E M _ C H 1 3 --
-- --
-- B o d y --
-- --
-- Copyright (C) 1992-2009, 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- --
-- ware Foundation; either version 3, or (at your option) any later ver- --
-- sion. GNAT is distributed in the hope that it will be useful, but WITH- --
-- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY --
-- or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License --
-- for more details. You should have received a copy of the GNU General --
-- Public License distributed with GNAT; see file COPYING3. If not, go to --
-- http://www.gnu.org/licenses for a complete copy of the license. --
-- --
-- GNAT was originally developed by the GNAT team at New York University. --
-- Extensive contributions were provided by Ada Core Technologies Inc. --
-- --
------------------------------------------------------------------------------
with Atree; use Atree;
with Checks; use Checks;
with Einfo; use Einfo;
with Errout; use Errout;
with Exp_Tss; use Exp_Tss;
with Exp_Util; use Exp_Util;
with Lib; use Lib;
with Lib.Xref; use Lib.Xref;
with Namet; use Namet;
with Nlists; use Nlists;
with Nmake; use Nmake;
with Opt; use Opt;
with Restrict; use Restrict;
with Rident; use Rident;
with Rtsfind; use Rtsfind;
with Sem; use Sem;
with Sem_Aux; use Sem_Aux;
with Sem_Ch3; use Sem_Ch3;
with Sem_Ch8; use Sem_Ch8;
with Sem_Eval; use Sem_Eval;
with Sem_Res; use Sem_Res;
with Sem_Type; use Sem_Type;
with Sem_Util; use Sem_Util;
with Sem_Warn; use Sem_Warn;
with Snames; use Snames;
with Stand; use Stand;
with Sinfo; use Sinfo;
with Table;
with Targparm; use Targparm;
with Ttypes; use Ttypes;
with Tbuild; use Tbuild;
with Urealp; use Urealp;
with GNAT.Heap_Sort_G;
package body Sem_Ch13 is
SSU : constant Pos := System_Storage_Unit;
-- Convenient short hand for commonly used constant
-----------------------
-- Local Subprograms --
-----------------------
procedure Alignment_Check_For_Esize_Change (Typ : Entity_Id);
-- This routine is called after setting the Esize of type entity Typ.
-- The purpose is to deal with the situation where an alignment has been
-- inherited from a derived type that is no longer appropriate for the
-- new Esize value. In this case, we reset the Alignment to unknown.
procedure Check_Component_Overlap (C1_Ent, C2_Ent : Entity_Id);
-- Given two entities for record components or discriminants, checks
-- if they have overlapping component clauses and issues errors if so.
function Get_Alignment_Value (Expr : Node_Id) return Uint;
-- Given the expression for an alignment value, returns the corresponding
-- Uint value. If the value is inappropriate, then error messages are
-- posted as required, and a value of No_Uint is returned.
function Is_Operational_Item (N : Node_Id) return Boolean;
-- A specification for a stream attribute is allowed before the full
-- type is declared, as explained in AI-00137 and the corrigendum.
-- Attributes that do not specify a representation characteristic are
-- operational attributes.
procedure New_Stream_Subprogram
(N : Node_Id;
Ent : Entity_Id;
Subp : Entity_Id;
Nam : TSS_Name_Type);
-- Create a subprogram renaming of a given stream attribute to the
-- designated subprogram and then in the tagged case, provide this as a
-- primitive operation, or in the non-tagged case make an appropriate TSS
-- entry. This is more properly an expansion activity than just semantics,
-- but the presence of user-defined stream functions for limited types is a
-- legality check, which is why this takes place here rather than in
-- exp_ch13, where it was previously. Nam indicates the name of the TSS
-- function to be generated.
--
-- To avoid elaboration anomalies with freeze nodes, for untagged types
-- we generate both a subprogram declaration and a subprogram renaming
-- declaration, so that the attribute specification is handled as a
-- renaming_as_body. For tagged types, the specification is one of the
-- primitive specs.
----------------------------------------------
-- Table for Validate_Unchecked_Conversions --
----------------------------------------------
-- The following table collects unchecked conversions for validation.
-- Entries are made by Validate_Unchecked_Conversion and then the
-- call to Validate_Unchecked_Conversions does the actual error
-- checking and posting of warnings. The reason for this delayed
-- processing is to take advantage of back-annotations of size and
-- alignment values performed by the back end.
-- Note: the reason we store a Source_Ptr value instead of a Node_Id
-- is that by the time Validate_Unchecked_Conversions is called, Sprint
-- will already have modified all Sloc values if the -gnatD option is set.
type UC_Entry is record
Eloc : Source_Ptr; -- node used for posting warnings
Source : Entity_Id; -- source type for unchecked conversion
Target : Entity_Id; -- target type for unchecked conversion
end record;
package Unchecked_Conversions is new Table.Table (
Table_Component_Type => UC_Entry,
Table_Index_Type => Int,
Table_Low_Bound => 1,
Table_Initial => 50,
Table_Increment => 200,
Table_Name => "Unchecked_Conversions");
----------------------------------------
-- Table for Validate_Address_Clauses --
----------------------------------------
-- If an address clause has the form
-- for X'Address use Expr
-- where Expr is of the form Y'Address or recursively is a reference
-- to a constant of either of these forms, and X and Y are entities of
-- objects, then if Y has a smaller alignment than X, that merits a
-- warning about possible bad alignment. The following table collects
-- address clauses of this kind. We put these in a table so that they
-- can be checked after the back end has completed annotation of the
-- alignments of objects, since we can catch more cases that way.
type Address_Clause_Check_Record is record
N : Node_Id;
-- The address clause
X : Entity_Id;
-- The entity of the object overlaying Y
Y : Entity_Id;
-- The entity of the object being overlaid
Off : Boolean;
-- Whether the address is offseted within Y
end record;
package Address_Clause_Checks is new Table.Table (
Table_Component_Type => Address_Clause_Check_Record,
Table_Index_Type => Int,
Table_Low_Bound => 1,
Table_Initial => 20,
Table_Increment => 200,
Table_Name => "Address_Clause_Checks");
-----------------------------------------
-- Adjust_Record_For_Reverse_Bit_Order --
-----------------------------------------
procedure Adjust_Record_For_Reverse_Bit_Order (R : Entity_Id) is
Max_Machine_Scalar_Size : constant Uint :=
UI_From_Int
(Standard_Long_Long_Integer_Size);
-- We use this as the maximum machine scalar size in the sense of AI-133
Num_CC : Natural;
Comp : Entity_Id;
SSU : constant Uint := UI_From_Int (System_Storage_Unit);
begin
-- This first loop through components does two things. First it deals
-- with the case of components with component clauses whose length is
-- greater than the maximum machine scalar size (either accepting them
-- or rejecting as needed). Second, it counts the number of components
-- with component clauses whose length does not exceed this maximum for
-- later processing.
Num_CC := 0;
Comp := First_Component_Or_Discriminant (R);
while Present (Comp) loop
declare
CC : constant Node_Id := Component_Clause (Comp);
begin
if Present (CC) then
declare
Fbit : constant Uint := Static_Integer (First_Bit (CC));
begin
-- Case of component with size > max machine scalar
if Esize (Comp) > Max_Machine_Scalar_Size then
-- Must begin on byte boundary
if Fbit mod SSU /= 0 then
Error_Msg_N
("illegal first bit value for reverse bit order",
First_Bit (CC));
Error_Msg_Uint_1 := SSU;
Error_Msg_Uint_2 := Max_Machine_Scalar_Size;
Error_Msg_N
("\must be a multiple of ^ if size greater than ^",
First_Bit (CC));
-- Must end on byte boundary
elsif Esize (Comp) mod SSU /= 0 then
Error_Msg_N
("illegal last bit value for reverse bit order",
Last_Bit (CC));
Error_Msg_Uint_1 := SSU;
Error_Msg_Uint_2 := Max_Machine_Scalar_Size;
Error_Msg_N
("\must be a multiple of ^ if size greater than ^",
Last_Bit (CC));
-- OK, give warning if enabled
elsif Warn_On_Reverse_Bit_Order then
Error_Msg_N
("multi-byte field specified with non-standard"
& " Bit_Order?", CC);
if Bytes_Big_Endian then
Error_Msg_N
("\bytes are not reversed "
& "(component is big-endian)?", CC);
else
Error_Msg_N
("\bytes are not reversed "
& "(component is little-endian)?", CC);
end if;
end if;
-- Case where size is not greater than max machine
-- scalar. For now, we just count these.
else
Num_CC := Num_CC + 1;
end if;
end;
end if;
end;
Next_Component_Or_Discriminant (Comp);
end loop;
-- We need to sort the component clauses on the basis of the Position
-- values in the clause, so we can group clauses with the same Position.
-- together to determine the relevant machine scalar size.
declare
Comps : array (0 .. Num_CC) of Entity_Id;
-- Array to collect component and discriminant entities. The data
-- starts at index 1, the 0'th entry is for the sort routine.
function CP_Lt (Op1, Op2 : Natural) return Boolean;
-- Compare routine for Sort
procedure CP_Move (From : Natural; To : Natural);
-- Move routine for Sort
package Sorting is new GNAT.Heap_Sort_G (CP_Move, CP_Lt);
Start : Natural;
Stop : Natural;
-- Start and stop positions in component list of set of components
-- with the same starting position (that constitute components in
-- a single machine scalar).
MaxL : Uint;
-- Maximum last bit value of any component in this set
MSS : Uint;
-- Corresponding machine scalar size
-----------
-- CP_Lt --
-----------
function CP_Lt (Op1, Op2 : Natural) return Boolean is
begin
return Position (Component_Clause (Comps (Op1))) <
Position (Component_Clause (Comps (Op2)));
end CP_Lt;
-------------
-- CP_Move --
-------------
procedure CP_Move (From : Natural; To : Natural) is
begin
Comps (To) := Comps (From);
end CP_Move;
begin
-- Collect the component clauses
Num_CC := 0;
Comp := First_Component_Or_Discriminant (R);
while Present (Comp) loop
if Present (Component_Clause (Comp))
and then Esize (Comp) <= Max_Machine_Scalar_Size
then
Num_CC := Num_CC + 1;
Comps (Num_CC) := Comp;
end if;
Next_Component_Or_Discriminant (Comp);
end loop;
-- Sort by ascending position number
Sorting.Sort (Num_CC);
-- We now have all the components whose size does not exceed the max
-- machine scalar value, sorted by starting position. In this loop
-- we gather groups of clauses starting at the same position, to
-- process them in accordance with Ada 2005 AI-133.
Stop := 0;
while Stop < Num_CC loop
Start := Stop + 1;
Stop := Start;
MaxL :=
Static_Integer (Last_Bit (Component_Clause (Comps (Start))));
while Stop < Num_CC loop
if Static_Integer
(Position (Component_Clause (Comps (Stop + 1)))) =
Static_Integer
(Position (Component_Clause (Comps (Stop))))
then
Stop := Stop + 1;
MaxL :=
UI_Max
(MaxL,
Static_Integer
(Last_Bit (Component_Clause (Comps (Stop)))));
else
exit;
end if;
end loop;
-- Now we have a group of component clauses from Start to Stop
-- whose positions are identical, and MaxL is the maximum last bit
-- value of any of these components.
-- We need to determine the corresponding machine scalar size.
-- This loop assumes that machine scalar sizes are even, and that
-- each possible machine scalar has twice as many bits as the
-- next smaller one.
MSS := Max_Machine_Scalar_Size;
while MSS mod 2 = 0
and then (MSS / 2) >= SSU
and then (MSS / 2) > MaxL
loop
MSS := MSS / 2;
end loop;
-- Here is where we fix up the Component_Bit_Offset value to
-- account for the reverse bit order. Some examples of what needs
-- to be done for the case of a machine scalar size of 8 are:
-- First_Bit .. Last_Bit Component_Bit_Offset
-- old new old new
-- 0 .. 0 7 .. 7 0 7
-- 0 .. 1 6 .. 7 0 6
-- 0 .. 2 5 .. 7 0 5
-- 0 .. 7 0 .. 7 0 4
-- 1 .. 1 6 .. 6 1 6
-- 1 .. 4 3 .. 6 1 3
-- 4 .. 7 0 .. 3 4 0
-- The general rule is that the first bit is obtained by
-- subtracting the old ending bit from machine scalar size - 1.
for C in Start .. Stop loop
declare
Comp : constant Entity_Id := Comps (C);
CC : constant Node_Id := Component_Clause (Comp);
LB : constant Uint := Static_Integer (Last_Bit (CC));
NFB : constant Uint := MSS - Uint_1 - LB;
NLB : constant Uint := NFB + Esize (Comp) - 1;
Pos : constant Uint := Static_Integer (Position (CC));
begin
if Warn_On_Reverse_Bit_Order then
Error_Msg_Uint_1 := MSS;
Error_Msg_N
("info: reverse bit order in machine " &
"scalar of length^?", First_Bit (CC));
Error_Msg_Uint_1 := NFB;
Error_Msg_Uint_2 := NLB;
if Bytes_Big_Endian then
Error_Msg_NE
("?\info: big-endian range for "
& "component & is ^ .. ^",
First_Bit (CC), Comp);
else
Error_Msg_NE
("?\info: little-endian range "
& "for component & is ^ .. ^",
First_Bit (CC), Comp);
end if;
end if;
Set_Component_Bit_Offset (Comp, Pos * SSU + NFB);
Set_Normalized_First_Bit (Comp, NFB mod SSU);
end;
end loop;
end loop;
end;
end Adjust_Record_For_Reverse_Bit_Order;
--------------------------------------
-- Alignment_Check_For_Esize_Change --
--------------------------------------
procedure Alignment_Check_For_Esize_Change (Typ : Entity_Id) is
begin
-- If the alignment is known, and not set by a rep clause, and is
-- inconsistent with the size being set, then reset it to unknown,
-- we assume in this case that the size overrides the inherited
-- alignment, and that the alignment must be recomputed.
if Known_Alignment (Typ)
and then not Has_Alignment_Clause (Typ)
and then Esize (Typ) mod (Alignment (Typ) * SSU) /= 0
then
Init_Alignment (Typ);
end if;
end Alignment_Check_For_Esize_Change;
-----------------------
-- Analyze_At_Clause --
-----------------------
-- An at clause is replaced by the corresponding Address attribute
-- definition clause that is the preferred approach in Ada 95.
procedure Analyze_At_Clause (N : Node_Id) is
CS : constant Boolean := Comes_From_Source (N);
begin
-- This is an obsolescent feature
Check_Restriction (No_Obsolescent_Features, N);
if Warn_On_Obsolescent_Feature then
Error_Msg_N
("at clause is an obsolescent feature (RM J.7(2))?", N);
Error_Msg_N
("\use address attribute definition clause instead?", N);
end if;
-- Rewrite as address clause
Rewrite (N,
Make_Attribute_Definition_Clause (Sloc (N),
Name => Identifier (N),
Chars => Name_Address,
Expression => Expression (N)));
-- We preserve Comes_From_Source, since logically the clause still
-- comes from the source program even though it is changed in form.
Set_Comes_From_Source (N, CS);
-- Analyze rewritten clause
Analyze_Attribute_Definition_Clause (N);
end Analyze_At_Clause;
-----------------------------------------
-- Analyze_Attribute_Definition_Clause --
-----------------------------------------
procedure Analyze_Attribute_Definition_Clause (N : Node_Id) is
Loc : constant Source_Ptr := Sloc (N);
Nam : constant Node_Id := Name (N);
Attr : constant Name_Id := Chars (N);
Expr : constant Node_Id := Expression (N);
Id : constant Attribute_Id := Get_Attribute_Id (Attr);
Ent : Entity_Id;
U_Ent : Entity_Id;
FOnly : Boolean := False;
-- Reset to True for subtype specific attribute (Alignment, Size)
-- and for stream attributes, i.e. those cases where in the call
-- to Rep_Item_Too_Late, FOnly is set True so that only the freezing
-- rules are checked. Note that the case of stream attributes is not
-- clear from the RM, but see AI95-00137. Also, the RM seems to
-- disallow Storage_Size for derived task types, but that is also
-- clearly unintentional.
procedure Analyze_Stream_TSS_Definition (TSS_Nam : TSS_Name_Type);
-- Common processing for 'Read, 'Write, 'Input and 'Output attribute
-- definition clauses.
-----------------------------------
-- Analyze_Stream_TSS_Definition --
-----------------------------------
procedure Analyze_Stream_TSS_Definition (TSS_Nam : TSS_Name_Type) is
Subp : Entity_Id := Empty;
I : Interp_Index;
It : Interp;
Pnam : Entity_Id;
Is_Read : constant Boolean := (TSS_Nam = TSS_Stream_Read);
function Has_Good_Profile (Subp : Entity_Id) return Boolean;
-- Return true if the entity is a subprogram with an appropriate
-- profile for the attribute being defined.
----------------------
-- Has_Good_Profile --
----------------------
function Has_Good_Profile (Subp : Entity_Id) return Boolean is
F : Entity_Id;
Is_Function : constant Boolean := (TSS_Nam = TSS_Stream_Input);
Expected_Ekind : constant array (Boolean) of Entity_Kind :=
(False => E_Procedure, True => E_Function);
Typ : Entity_Id;
begin
if Ekind (Subp) /= Expected_Ekind (Is_Function) then
return False;
end if;
F := First_Formal (Subp);
if No (F)
or else Ekind (Etype (F)) /= E_Anonymous_Access_Type
or else Designated_Type (Etype (F)) /=
Class_Wide_Type (RTE (RE_Root_Stream_Type))
then
return False;
end if;
if not Is_Function then
Next_Formal (F);
declare
Expected_Mode : constant array (Boolean) of Entity_Kind :=
(False => E_In_Parameter,
True => E_Out_Parameter);
begin
if Parameter_Mode (F) /= Expected_Mode (Is_Read) then
return False;
end if;
end;
Typ := Etype (F);
else
Typ := Etype (Subp);
end if;
return Base_Type (Typ) = Base_Type (Ent)
and then No (Next_Formal (F));
end Has_Good_Profile;
-- Start of processing for Analyze_Stream_TSS_Definition
begin
FOnly := True;
if not Is_Type (U_Ent) then
Error_Msg_N ("local name must be a subtype", Nam);
return;
end if;
Pnam := TSS (Base_Type (U_Ent), TSS_Nam);
-- If Pnam is present, it can be either inherited from an ancestor
-- type (in which case it is legal to redefine it for this type), or
-- be a previous definition of the attribute for the same type (in
-- which case it is illegal).
-- In the first case, it will have been analyzed already, and we
-- can check that its profile does not match the expected profile
-- for a stream attribute of U_Ent. In the second case, either Pnam
-- has been analyzed (and has the expected profile), or it has not
-- been analyzed yet (case of a type that has not been frozen yet
-- and for which the stream attribute has been set using Set_TSS).
if Present (Pnam)
and then (No (First_Entity (Pnam)) or else Has_Good_Profile (Pnam))
then
Error_Msg_Sloc := Sloc (Pnam);
Error_Msg_Name_1 := Attr;
Error_Msg_N ("% attribute already defined #", Nam);
return;
end if;
Analyze (Expr);
if Is_Entity_Name (Expr) then
if not Is_Overloaded (Expr) then
if Has_Good_Profile (Entity (Expr)) then
Subp := Entity (Expr);
end if;
else
Get_First_Interp (Expr, I, It);
while Present (It.Nam) loop
if Has_Good_Profile (It.Nam) then
Subp := It.Nam;
exit;
end if;
Get_Next_Interp (I, It);
end loop;
end if;
end if;
if Present (Subp) then
if Is_Abstract_Subprogram (Subp) then
Error_Msg_N ("stream subprogram must not be abstract", Expr);
return;
end if;
Set_Entity (Expr, Subp);
Set_Etype (Expr, Etype (Subp));
New_Stream_Subprogram (N, U_Ent, Subp, TSS_Nam);
else
Error_Msg_Name_1 := Attr;
Error_Msg_N ("incorrect expression for% attribute", Expr);
end if;
end Analyze_Stream_TSS_Definition;
-- Start of processing for Analyze_Attribute_Definition_Clause
begin
-- Process Ignore_Rep_Clauses option
if Ignore_Rep_Clauses then
case Id is
-- The following should be ignored. They do not affect legality
-- and may be target dependent. The basic idea of -gnatI is to
-- ignore any rep clauses that may be target dependent but do not
-- affect legality (except possibly to be rejected because they
-- are incompatible with the compilation target).
when Attribute_Alignment |
Attribute_Bit_Order |
Attribute_Component_Size |
Attribute_Machine_Radix |
Attribute_Object_Size |
Attribute_Size |
Attribute_Small |
Attribute_Stream_Size |
Attribute_Value_Size =>
Rewrite (N, Make_Null_Statement (Sloc (N)));
return;
-- The following should not be ignored, because in the first place
-- they are reasonably portable, and should not cause problems in
-- compiling code from another target, and also they do affect
-- legality, e.g. failing to provide a stream attribute for a
-- type may make a program illegal.
when Attribute_External_Tag |
Attribute_Input |
Attribute_Output |
Attribute_Read |
Attribute_Storage_Pool |
Attribute_Storage_Size |
Attribute_Write =>
null;
-- Other cases are errors, which will be caught below
when others =>
null;
end case;
end if;
Analyze (Nam);
Ent := Entity (Nam);
if Rep_Item_Too_Early (Ent, N) then
return;
end if;
-- Rep clause applies to full view of incomplete type or private type if
-- we have one (if not, this is a premature use of the type). However,
-- certain semantic checks need to be done on the specified entity (i.e.
-- the private view), so we save it in Ent.
if Is_Private_Type (Ent)
and then Is_Derived_Type (Ent)
and then not Is_Tagged_Type (Ent)
and then No (Full_View (Ent))
then
-- If this is a private type whose completion is a derivation from
-- another private type, there is no full view, and the attribute
-- belongs to the type itself, not its underlying parent.
U_Ent := Ent;
elsif Ekind (Ent) = E_Incomplete_Type then
-- The attribute applies to the full view, set the entity of the
-- attribute definition accordingly.
Ent := Underlying_Type (Ent);
U_Ent := Ent;
Set_Entity (Nam, Ent);
else
U_Ent := Underlying_Type (Ent);
end if;
-- Complete other routine error checks
if Etype (Nam) = Any_Type then
return;
elsif Scope (Ent) /= Current_Scope then
Error_Msg_N ("entity must be declared in this scope", Nam);
return;
elsif No (U_Ent) then
U_Ent := Ent;
elsif Is_Type (U_Ent)
and then not Is_First_Subtype (U_Ent)
and then Id /= Attribute_Object_Size
and then Id /= Attribute_Value_Size
and then not From_At_Mod (N)
then
Error_Msg_N ("cannot specify attribute for subtype", Nam);
return;
end if;
-- Switch on particular attribute
case Id is
-------------
-- Address --
-------------
-- Address attribute definition clause
when Attribute_Address => Address : begin
-- A little error check, catch for X'Address use X'Address;
if Nkind (Nam) = N_Identifier
and then Nkind (Expr) = N_Attribute_Reference
and then Attribute_Name (Expr) = Name_Address
and then Nkind (Prefix (Expr)) = N_Identifier
and then Chars (Nam) = Chars (Prefix (Expr))
then
Error_Msg_NE
("address for & is self-referencing", Prefix (Expr), Ent);
return;
end if;
-- Not that special case, carry on with analysis of expression
Analyze_And_Resolve (Expr, RTE (RE_Address));
-- Even when ignoring rep clauses we need to indicate that the
-- entity has an address clause and thus it is legal to declare
-- it imported.
if Ignore_Rep_Clauses then
if Ekind (U_Ent) = E_Variable
or else Ekind (U_Ent) = E_Constant
then
Record_Rep_Item (U_Ent, N);
end if;
return;
end if;
if Present (Address_Clause (U_Ent)) then
Error_Msg_N ("address already given for &", Nam);
-- Case of address clause for subprogram
elsif Is_Subprogram (U_Ent) then
if Has_Homonym (U_Ent) then
Error_Msg_N
("address clause cannot be given " &
"for overloaded subprogram",
Nam);
return;
end if;
-- For subprograms, all address clauses are permitted, and we
-- mark the subprogram as having a deferred freeze so that Gigi
-- will not elaborate it too soon.
-- Above needs more comments, what is too soon about???
Set_Has_Delayed_Freeze (U_Ent);
-- Case of address clause for entry
elsif Ekind (U_Ent) = E_Entry then
if Nkind (Parent (N)) = N_Task_Body then
Error_Msg_N
("entry address must be specified in task spec", Nam);
return;
end if;
-- For entries, we require a constant address
Check_Constant_Address_Clause (Expr, U_Ent);
-- Special checks for task types
if Is_Task_Type (Scope (U_Ent))
and then Comes_From_Source (Scope (U_Ent))
then
Error_Msg_N
("?entry address declared for entry in task type", N);
Error_Msg_N
("\?only one task can be declared of this type", N);
end if;
-- Entry address clauses are obsolescent
Check_Restriction (No_Obsolescent_Features, N);
if Warn_On_Obsolescent_Feature then
Error_Msg_N
("attaching interrupt to task entry is an " &
"obsolescent feature (RM J.7.1)?", N);
Error_Msg_N
("\use interrupt procedure instead?", N);
end if;
-- Case of an address clause for a controlled object which we
-- consider to be erroneous.
elsif Is_Controlled (Etype (U_Ent))
or else Has_Controlled_Component (Etype (U_Ent))
then
Error_Msg_NE
("?controlled object& must not be overlaid", Nam, U_Ent);
Error_Msg_N
("\?Program_Error will be raised at run time", Nam);
Insert_Action (Declaration_Node (U_Ent),
Make_Raise_Program_Error (Loc,
Reason => PE_Overlaid_Controlled_Object));
return;
-- Case of address clause for a (non-controlled) object
elsif
Ekind (U_Ent) = E_Variable
or else
Ekind (U_Ent) = E_Constant
then
declare
Expr : constant Node_Id := Expression (N);
O_Ent : Entity_Id;
Off : Boolean;
begin
-- Exported variables cannot have an address clause, because
-- this cancels the effect of the pragma Export.
if Is_Exported (U_Ent) then
Error_Msg_N
("cannot export object with address clause", Nam);
return;
end if;
Find_Overlaid_Entity (N, O_Ent, Off);
-- Overlaying controlled objects is erroneous
if Present (O_Ent)
and then (Has_Controlled_Component (Etype (O_Ent))
or else Is_Controlled (Etype (O_Ent)))
then
Error_Msg_N
("?cannot overlay with controlled object", Expr);
Error_Msg_N
("\?Program_Error will be raised at run time", Expr);
Insert_Action (Declaration_Node (U_Ent),
Make_Raise_Program_Error (Loc,
Reason => PE_Overlaid_Controlled_Object));
return;
elsif Present (O_Ent)
and then Ekind (U_Ent) = E_Constant
and then not Is_Constant_Object (O_Ent)
then
Error_Msg_N ("constant overlays a variable?", Expr);
elsif Present (Renamed_Object (U_Ent)) then
Error_Msg_N
("address clause not allowed"
& " for a renaming declaration (RM 13.1(6))", Nam);
return;
-- Imported variables can have an address clause, but then
-- the import is pretty meaningless except to suppress
-- initializations, so we do not need such variables to
-- be statically allocated (and in fact it causes trouble
-- if the address clause is a local value).
elsif Is_Imported (U_Ent) then
Set_Is_Statically_Allocated (U_Ent, False);
end if;
-- We mark a possible modification of a variable with an
-- address clause, since it is likely aliasing is occurring.
Note_Possible_Modification (Nam, Sure => False);
-- Here we are checking for explicit overlap of one variable
-- by another, and if we find this then mark the overlapped
-- variable as also being volatile to prevent unwanted
-- optimizations. This is a significant pessimization so
-- avoid it when there is an offset, i.e. when the object
-- is composite; they cannot be optimized easily anyway.
if Present (O_Ent)
and then Is_Object (O_Ent)
and then not Off
then
Set_Treat_As_Volatile (O_Ent);
end if;
-- Legality checks on the address clause for initialized
-- objects is deferred until the freeze point, because
-- a subsequent pragma might indicate that the object is
-- imported and thus not initialized.
Set_Has_Delayed_Freeze (U_Ent);
-- If an initialization call has been generated for this
-- object, it needs to be deferred to after the freeze node
-- we have just now added, otherwise GIGI will see a
-- reference to the variable (as actual to the IP call)
-- before its definition.
declare
Init_Call : constant Node_Id := Find_Init_Call (U_Ent, N);
begin
if Present (Init_Call) then
Remove (Init_Call);
Append_Freeze_Action (U_Ent, Init_Call);
end if;
end;
if Is_Exported (U_Ent) then
Error_Msg_N
("& cannot be exported if an address clause is given",
Nam);
Error_Msg_N
("\define and export a variable " &
"that holds its address instead",
Nam);
end if;
-- Entity has delayed freeze, so we will generate an
-- alignment check at the freeze point unless suppressed.
if not Range_Checks_Suppressed (U_Ent)
and then not Alignment_Checks_Suppressed (U_Ent)
then
Set_Check_Address_Alignment (N);
end if;
-- Kill the size check code, since we are not allocating
-- the variable, it is somewhere else.
Kill_Size_Check_Code (U_Ent);
-- If the address clause is of the form:
-- for Y'Address use X'Address
-- or
-- Const : constant Address := X'Address;
-- ...
-- for Y'Address use Const;
-- then we make an entry in the table for checking the size
-- and alignment of the overlaying variable. We defer this
-- check till after code generation to take full advantage
-- of the annotation done by the back end. This entry is
-- only made if the address clause comes from source.
if Address_Clause_Overlay_Warnings
and then Comes_From_Source (N)
and then Present (O_Ent)
and then Is_Object (O_Ent)
then
Address_Clause_Checks.Append ((N, U_Ent, O_Ent, Off));
-- If variable overlays a constant view, and we are
-- warning on overlays, then mark the variable as
-- overlaying a constant (we will give warnings later
-- if this variable is assigned).
if Is_Constant_Object (O_Ent)
and then Ekind (U_Ent) = E_Variable
then
Set_Overlays_Constant (U_Ent);
end if;
end if;
end;
-- Not a valid entity for an address clause
else
Error_Msg_N ("address cannot be given for &", Nam);
end if;
end Address;
---------------
-- Alignment --
---------------
-- Alignment attribute definition clause
when Attribute_Alignment => Alignment : declare
Align : constant Uint := Get_Alignment_Value (Expr);
begin
FOnly := True;
if not Is_Type (U_Ent)
and then Ekind (U_Ent) /= E_Variable
and then Ekind (U_Ent) /= E_Constant
then
Error_Msg_N ("alignment cannot be given for &", Nam);
elsif Has_Alignment_Clause (U_Ent) then
Error_Msg_Sloc := Sloc (Alignment_Clause (U_Ent));
Error_Msg_N ("alignment clause previously given#", N);
elsif Align /= No_Uint then
Set_Has_Alignment_Clause (U_Ent);
Set_Alignment (U_Ent, Align);
-- For an array type, U_Ent is the first subtype. In that case,
-- also set the alignment of the anonymous base type so that
-- other subtypes (such as the itypes for aggregates of the
-- type) also receive the expected alignment.
if Is_Array_Type (U_Ent) then
Set_Alignment (Base_Type (U_Ent), Align);
end if;
end if;
end Alignment;
---------------
-- Bit_Order --
---------------
-- Bit_Order attribute definition clause
when Attribute_Bit_Order => Bit_Order : declare
begin
if not Is_Record_Type (U_Ent) then
Error_Msg_N
("Bit_Order can only be defined for record type", Nam);
else
Analyze_And_Resolve (Expr, RTE (RE_Bit_Order));
if Etype (Expr) = Any_Type then
return;
elsif not Is_Static_Expression (Expr) then
Flag_Non_Static_Expr
("Bit_Order requires static expression!", Expr);
else
if (Expr_Value (Expr) = 0) /= Bytes_Big_Endian then
Set_Reverse_Bit_Order (U_Ent, True);
end if;
end if;
end if;
end Bit_Order;
--------------------
-- Component_Size --
--------------------
-- Component_Size attribute definition clause
when Attribute_Component_Size => Component_Size_Case : declare
Csize : constant Uint := Static_Integer (Expr);
Btype : Entity_Id;
Biased : Boolean;
New_Ctyp : Entity_Id;
Decl : Node_Id;
begin
if not Is_Array_Type (U_Ent) then
Error_Msg_N ("component size requires array type", Nam);
return;
end if;
Btype := Base_Type (U_Ent);
if Has_Component_Size_Clause (Btype) then
Error_Msg_N
("component size clause for& previously given", Nam);
elsif Csize /= No_Uint then
Check_Size (Expr, Component_Type (Btype), Csize, Biased);
if Has_Aliased_Components (Btype)
and then Csize < 32
and then Csize /= 8
and then Csize /= 16
then
Error_Msg_N
("component size incorrect for aliased components", N);
return;
end if;
-- For the biased case, build a declaration for a subtype
-- that will be used to represent the biased subtype that
-- reflects the biased representation of components. We need
-- this subtype to get proper conversions on referencing
-- elements of the array. Note that component size clauses
-- are ignored in VM mode.
if VM_Target = No_VM then
if Biased then
New_Ctyp :=
Make_Defining_Identifier (Loc,
Chars =>
New_External_Name (Chars (U_Ent), 'C', 0, 'T'));
Decl :=
Make_Subtype_Declaration (Loc,
Defining_Identifier => New_Ctyp,
Subtype_Indication =>
New_Occurrence_Of (Component_Type (Btype), Loc));
Set_Parent (Decl, N);
Analyze (Decl, Suppress => All_Checks);
Set_Has_Delayed_Freeze (New_Ctyp, False);
Set_Esize (New_Ctyp, Csize);
Set_RM_Size (New_Ctyp, Csize);
Init_Alignment (New_Ctyp);
Set_Has_Biased_Representation (New_Ctyp, True);
Set_Is_Itype (New_Ctyp, True);
Set_Associated_Node_For_Itype (New_Ctyp, U_Ent);
Set_Component_Type (Btype, New_Ctyp);
if Warn_On_Biased_Representation then
Error_Msg_N
("?component size clause forces biased "
& "representation", N);
end if;
end if;
Set_Component_Size (Btype, Csize);
-- For VM case, we ignore component size clauses
else
-- Give a warning unless we are in GNAT mode, in which case
-- the warning is suppressed since it is not useful.
if not GNAT_Mode then
Error_Msg_N
("?component size ignored in this configuration", N);
end if;
end if;
Set_Has_Component_Size_Clause (Btype, True);
Set_Has_Non_Standard_Rep (Btype, True);
end if;
end Component_Size_Case;
------------------
-- External_Tag --
------------------
when Attribute_External_Tag => External_Tag :
begin
if not Is_Tagged_Type (U_Ent) then
Error_Msg_N ("should be a tagged type", Nam);
end if;
Analyze_And_Resolve (Expr, Standard_String);
if not Is_Static_Expression (Expr) then
Flag_Non_Static_Expr
("static string required for tag name!", Nam);
end if;
if VM_Target = No_VM then
Set_Has_External_Tag_Rep_Clause (U_Ent);
else
Error_Msg_Name_1 := Attr;
Error_Msg_N
("% attribute unsupported in this configuration", Nam);
end if;
if not Is_Library_Level_Entity (U_Ent) then
Error_Msg_NE
("?non-unique external tag supplied for &", N, U_Ent);
Error_Msg_N
("?\same external tag applies to all subprogram calls", N);
Error_Msg_N
("?\corresponding internal tag cannot be obtained", N);
end if;
end External_Tag;
-----------
-- Input --
-----------
when Attribute_Input =>
Analyze_Stream_TSS_Definition (TSS_Stream_Input);
Set_Has_Specified_Stream_Input (Ent);
-------------------
-- Machine_Radix --
-------------------
-- Machine radix attribute definition clause
when Attribute_Machine_Radix => Machine_Radix : declare
Radix : constant Uint := Static_Integer (Expr);
begin
if not Is_Decimal_Fixed_Point_Type (U_Ent) then
Error_Msg_N ("decimal fixed-point type expected for &", Nam);
elsif Has_Machine_Radix_Clause (U_Ent) then
Error_Msg_Sloc := Sloc (Alignment_Clause (U_Ent));
Error_Msg_N ("machine radix clause previously given#", N);
elsif Radix /= No_Uint then
Set_Has_Machine_Radix_Clause (U_Ent);
Set_Has_Non_Standard_Rep (Base_Type (U_Ent));
if Radix = 2 then
null;
elsif Radix = 10 then
Set_Machine_Radix_10 (U_Ent);
else
Error_Msg_N ("machine radix value must be 2 or 10", Expr);
end if;
end if;
end Machine_Radix;
-----------------
-- Object_Size --
-----------------
-- Object_Size attribute definition clause
when Attribute_Object_Size => Object_Size : declare
Size : constant Uint := Static_Integer (Expr);
Biased : Boolean;
pragma Warnings (Off, Biased);
begin
if not Is_Type (U_Ent) then
Error_Msg_N ("Object_Size cannot be given for &", Nam);
elsif Has_Object_Size_Clause (U_Ent) then
Error_Msg_N ("Object_Size already given for &", Nam);
else
Check_Size (Expr, U_Ent, Size, Biased);
if Size /= 8
and then
Size /= 16
and then
Size /= 32
and then
UI_Mod (Size, 64) /= 0
then
Error_Msg_N
("Object_Size must be 8, 16, 32, or multiple of 64",
Expr);
end if;
Set_Esize (U_Ent, Size);
Set_Has_Object_Size_Clause (U_Ent);
Alignment_Check_For_Esize_Change (U_Ent);
end if;
end Object_Size;
------------
-- Output --
------------
when Attribute_Output =>
Analyze_Stream_TSS_Definition (TSS_Stream_Output);
Set_Has_Specified_Stream_Output (Ent);
----------
-- Read --
----------
when Attribute_Read =>
Analyze_Stream_TSS_Definition (TSS_Stream_Read);
Set_Has_Specified_Stream_Read (Ent);
----------
-- Size --
----------
-- Size attribute definition clause
when Attribute_Size => Size : declare
Size : constant Uint := Static_Integer (Expr);
Etyp : Entity_Id;
Biased : Boolean;
begin
FOnly := True;
if Has_Size_Clause (U_Ent) then
Error_Msg_N ("size already given for &", Nam);
elsif not Is_Type (U_Ent)
and then Ekind (U_Ent) /= E_Variable
and then Ekind (U_Ent) /= E_Constant
then
Error_Msg_N ("size cannot be given for &", Nam);
elsif Is_Array_Type (U_Ent)
and then not Is_Constrained (U_Ent)
then
Error_Msg_N
("size cannot be given for unconstrained array", Nam);
elsif Size /= No_Uint then
if Is_Type (U_Ent) then
Etyp := U_Ent;
else
Etyp := Etype (U_Ent);
end if;
-- Check size, note that Gigi is in charge of checking that the
-- size of an array or record type is OK. Also we do not check
-- the size in the ordinary fixed-point case, since it is too
-- early to do so (there may be subsequent small clause that
-- affects the size). We can check the size if a small clause
-- has already been given.
if not Is_Ordinary_Fixed_Point_Type (U_Ent)
or else Has_Small_Clause (U_Ent)
then
Check_Size (Expr, Etyp, Size, Biased);
Set_Has_Biased_Representation (U_Ent, Biased);
if Biased and Warn_On_Biased_Representation then
Error_Msg_N
("?size clause forces biased representation", N);
end if;
end if;
-- For types set RM_Size and Esize if possible
if Is_Type (U_Ent) then
Set_RM_Size (U_Ent, Size);
-- For scalar types, increase Object_Size to power of 2, but
-- not less than a storage unit in any case (i.e., normally
-- this means it will be byte addressable).
if Is_Scalar_Type (U_Ent) then
if Size <= System_Storage_Unit then
Init_Esize (U_Ent, System_Storage_Unit);
elsif Size <= 16 then
Init_Esize (U_Ent, 16);
elsif Size <= 32 then
Init_Esize (U_Ent, 32);
else
Set_Esize (U_Ent, (Size + 63) / 64 * 64);
end if;
-- For all other types, object size = value size. The
-- backend will adjust as needed.
else
Set_Esize (U_Ent, Size);
end if;
Alignment_Check_For_Esize_Change (U_Ent);
-- For objects, set Esize only
else
if Is_Elementary_Type (Etyp) then
if Size /= System_Storage_Unit
and then
Size /= System_Storage_Unit * 2
and then
Size /= System_Storage_Unit * 4
and then
Size /= System_Storage_Unit * 8
then
Error_Msg_Uint_1 := UI_From_Int (System_Storage_Unit);
Error_Msg_Uint_2 := Error_Msg_Uint_1 * 8;
Error_Msg_N
("size for primitive object must be a power of 2"
& " in the range ^-^", N);
end if;
end if;
Set_Esize (U_Ent, Size);
end if;
Set_Has_Size_Clause (U_Ent);
end if;
end Size;
-----------
-- Small --
-----------
-- Small attribute definition clause
when Attribute_Small => Small : declare
Implicit_Base : constant Entity_Id := Base_Type (U_Ent);
Small : Ureal;
begin
Analyze_And_Resolve (Expr, Any_Real);
if Etype (Expr) = Any_Type then
return;
elsif not Is_Static_Expression (Expr) then
Flag_Non_Static_Expr
("small requires static expression!", Expr);
return;
else
Small := Expr_Value_R (Expr);
if Small <= Ureal_0 then
Error_Msg_N ("small value must be greater than zero", Expr);
return;
end if;
end if;
if not Is_Ordinary_Fixed_Point_Type (U_Ent) then
Error_Msg_N
("small requires an ordinary fixed point type", Nam);
elsif Has_Small_Clause (U_Ent) then
Error_Msg_N ("small already given for &", Nam);
elsif Small > Delta_Value (U_Ent) then
Error_Msg_N
("small value must not be greater then delta value", Nam);
else
Set_Small_Value (U_Ent, Small);
Set_Small_Value (Implicit_Base, Small);
Set_Has_Small_Clause (U_Ent);
Set_Has_Small_Clause (Implicit_Base);
Set_Has_Non_Standard_Rep (Implicit_Base);
end if;
end Small;
------------------
-- Storage_Pool --
------------------
-- Storage_Pool attribute definition clause
when Attribute_Storage_Pool => Storage_Pool : declare
Pool : Entity_Id;
T : Entity_Id;
begin
if Ekind (U_Ent) = E_Access_Subprogram_Type then
Error_Msg_N
("storage pool cannot be given for access-to-subprogram type",
Nam);
return;
elsif Ekind (U_Ent) /= E_Access_Type
and then Ekind (U_Ent) /= E_General_Access_Type
then
Error_Msg_N
("storage pool can only be given for access types", Nam);
return;
elsif Is_Derived_Type (U_Ent) then
Error_Msg_N
("storage pool cannot be given for a derived access type",
Nam);
elsif Has_Storage_Size_Clause (U_Ent) then
Error_Msg_N ("storage size already given for &", Nam);
return;
elsif Present (Associated_Storage_Pool (U_Ent)) then
Error_Msg_N ("storage pool already given for &", Nam);
return;
end if;
Analyze_And_Resolve
(Expr, Class_Wide_Type (RTE (RE_Root_Storage_Pool)));
if not Denotes_Variable (Expr) then
Error_Msg_N ("storage pool must be a variable", Expr);
return;
end if;
if Nkind (Expr) = N_Type_Conversion then
T := Etype (Expression (Expr));
else
T := Etype (Expr);
end if;
-- The Stack_Bounded_Pool is used internally for implementing
-- access types with a Storage_Size. Since it only work
-- properly when used on one specific type, we need to check
-- that it is not hijacked improperly:
-- type T is access Integer;
-- for T'Storage_Size use n;
-- type Q is access Float;
-- for Q'Storage_Size use T'Storage_Size; -- incorrect
if RTE_Available (RE_Stack_Bounded_Pool)
and then Base_Type (T) = RTE (RE_Stack_Bounded_Pool)
then
Error_Msg_N ("non-shareable internal Pool", Expr);
return;
end if;
-- If the argument is a name that is not an entity name, then
-- we construct a renaming operation to define an entity of
-- type storage pool.
if not Is_Entity_Name (Expr)
and then Is_Object_Reference (Expr)
then
Pool :=
Make_Defining_Identifier (Loc,
Chars => New_Internal_Name ('P'));
declare
Rnode : constant Node_Id :=
Make_Object_Renaming_Declaration (Loc,
Defining_Identifier => Pool,
Subtype_Mark =>
New_Occurrence_Of (Etype (Expr), Loc),
Name => Expr);
begin
Insert_Before (N, Rnode);
Analyze (Rnode);
Set_Associated_Storage_Pool (U_Ent, Pool);
end;
elsif Is_Entity_Name (Expr) then
Pool := Entity (Expr);
-- If pool is a renamed object, get original one. This can
-- happen with an explicit renaming, and within instances.
while Present (Renamed_Object (Pool))
and then Is_Entity_Name (Renamed_Object (Pool))
loop
Pool := Entity (Renamed_Object (Pool));
end loop;
if Present (Renamed_Object (Pool))
and then Nkind (Renamed_Object (Pool)) = N_Type_Conversion
and then Is_Entity_Name (Expression (Renamed_Object (Pool)))
then
Pool := Entity (Expression (Renamed_Object (Pool)));
end if;
Set_Associated_Storage_Pool (U_Ent, Pool);
elsif Nkind (Expr) = N_Type_Conversion
and then Is_Entity_Name (Expression (Expr))
and then Nkind (Original_Node (Expr)) = N_Attribute_Reference
then
Pool := Entity (Expression (Expr));
Set_Associated_Storage_Pool (U_Ent, Pool);
else
Error_Msg_N ("incorrect reference to a Storage Pool", Expr);
return;
end if;
end Storage_Pool;
------------------
-- Storage_Size --
------------------
-- Storage_Size attribute definition clause
when Attribute_Storage_Size => Storage_Size : declare
Btype : constant Entity_Id := Base_Type (U_Ent);
Sprag : Node_Id;
begin
if Is_Task_Type (U_Ent) then
Check_Restriction (No_Obsolescent_Features, N);
if Warn_On_Obsolescent_Feature then
Error_Msg_N
("storage size clause for task is an " &
"obsolescent feature (RM J.9)?", N);
Error_Msg_N
("\use Storage_Size pragma instead?", N);
end if;
FOnly := True;
end if;
if not Is_Access_Type (U_Ent)
and then Ekind (U_Ent) /= E_Task_Type
then
Error_Msg_N ("storage size cannot be given for &", Nam);
elsif Is_Access_Type (U_Ent) and Is_Derived_Type (U_Ent) then
Error_Msg_N
("storage size cannot be given for a derived access type",
Nam);
elsif Has_Storage_Size_Clause (Btype) then
Error_Msg_N ("storage size already given for &", Nam);
else
Analyze_And_Resolve (Expr, Any_Integer);
if Is_Access_Type (U_Ent) then
if Present (Associated_Storage_Pool (U_Ent)) then
Error_Msg_N ("storage pool already given for &", Nam);
return;
end if;
if Compile_Time_Known_Value (Expr)
and then Expr_Value (Expr) = 0
then
Set_No_Pool_Assigned (Btype);
end if;
else -- Is_Task_Type (U_Ent)
Sprag := Get_Rep_Pragma (Btype, Name_Storage_Size);
if Present (Sprag) then
Error_Msg_Sloc := Sloc (Sprag);
Error_Msg_N
("Storage_Size already specified#", Nam);
return;
end if;
end if;
Set_Has_Storage_Size_Clause (Btype);
end if;
end Storage_Size;
-----------------
-- Stream_Size --
-----------------
when Attribute_Stream_Size => Stream_Size : declare
Size : constant Uint := Static_Integer (Expr);
begin
if Ada_Version <= Ada_95 then
Check_Restriction (No_Implementation_Attributes, N);
end if;
if Has_Stream_Size_Clause (U_Ent) then
Error_Msg_N ("Stream_Size already given for &", Nam);
elsif Is_Elementary_Type (U_Ent) then
if Size /= System_Storage_Unit
and then
Size /= System_Storage_Unit * 2
and then
Size /= System_Storage_Unit * 4
and then
Size /= System_Storage_Unit * 8
then
Error_Msg_Uint_1 := UI_From_Int (System_Storage_Unit);
Error_Msg_N
("stream size for elementary type must be a"
& " power of 2 and at least ^", N);
elsif RM_Size (U_Ent) > Size then
Error_Msg_Uint_1 := RM_Size (U_Ent);
Error_Msg_N
("stream size for elementary type must be a"
& " power of 2 and at least ^", N);
end if;
Set_Has_Stream_Size_Clause (U_Ent);
else
Error_Msg_N ("Stream_Size cannot be given for &", Nam);
end if;
end Stream_Size;
----------------
-- Value_Size --
----------------
-- Value_Size attribute definition clause
when Attribute_Value_Size => Value_Size : declare
Size : constant Uint := Static_Integer (Expr);
Biased : Boolean;
begin
if not Is_Type (U_Ent) then
Error_Msg_N ("Value_Size cannot be given for &", Nam);
elsif Present
(Get_Attribute_Definition_Clause
(U_Ent, Attribute_Value_Size))
then
Error_Msg_N ("Value_Size already given for &", Nam);
elsif Is_Array_Type (U_Ent)
and then not Is_Constrained (U_Ent)
then
Error_Msg_N
("Value_Size cannot be given for unconstrained array", Nam);
else
if Is_Elementary_Type (U_Ent) then
Check_Size (Expr, U_Ent, Size, Biased);
Set_Has_Biased_Representation (U_Ent, Biased);
if Biased and Warn_On_Biased_Representation then
Error_Msg_N
("?value size clause forces biased representation", N);
end if;
end if;
Set_RM_Size (U_Ent, Size);
end if;
end Value_Size;
-----------
-- Write --
-----------
when Attribute_Write =>
Analyze_Stream_TSS_Definition (TSS_Stream_Write);
Set_Has_Specified_Stream_Write (Ent);
-- All other attributes cannot be set
when others =>
Error_Msg_N
("attribute& cannot be set with definition clause", N);
end case;
-- The test for the type being frozen must be performed after
-- any expression the clause has been analyzed since the expression
-- itself might cause freezing that makes the clause illegal.
if Rep_Item_Too_Late (U_Ent, N, FOnly) then
return;
end if;
end Analyze_Attribute_Definition_Clause;
----------------------------
-- Analyze_Code_Statement --
----------------------------
procedure Analyze_Code_Statement (N : Node_Id) is
HSS : constant Node_Id := Parent (N);
SBody : constant Node_Id := Parent (HSS);
Subp : constant Entity_Id := Current_Scope;
Stmt : Node_Id;
Decl : Node_Id;
StmtO : Node_Id;
DeclO : Node_Id;
begin
-- Analyze and check we get right type, note that this implements the
-- requirement (RM 13.8(1)) that Machine_Code be with'ed, since that
-- is the only way that Asm_Insn could possibly be visible.
Analyze_And_Resolve (Expression (N));
if Etype (Expression (N)) = Any_Type then
return;
elsif Etype (Expression (N)) /= RTE (RE_Asm_Insn) then
Error_Msg_N ("incorrect type for code statement", N);
return;
end if;
Check_Code_Statement (N);
-- Make sure we appear in the handled statement sequence of a
-- subprogram (RM 13.8(3)).
if Nkind (HSS) /= N_Handled_Sequence_Of_Statements
or else Nkind (SBody) /= N_Subprogram_Body
then
Error_Msg_N
("code statement can only appear in body of subprogram", N);
return;
end if;
-- Do remaining checks (RM 13.8(3)) if not already done
if not Is_Machine_Code_Subprogram (Subp) then
Set_Is_Machine_Code_Subprogram (Subp);
-- No exception handlers allowed
if Present (Exception_Handlers (HSS)) then
Error_Msg_N
("exception handlers not permitted in machine code subprogram",
First (Exception_Handlers (HSS)));
end if;
-- No declarations other than use clauses and pragmas (we allow
-- certain internally generated declarations as well).
Decl := First (Declarations (SBody));
while Present (Decl) loop
DeclO := Original_Node (Decl);
if Comes_From_Source (DeclO)
and not Nkind_In (DeclO, N_Pragma,
N_Use_Package_Clause,
N_Use_Type_Clause,
N_Implicit_Label_Declaration)
then
Error_Msg_N
("this declaration not allowed in machine code subprogram",
DeclO);
end if;
Next (Decl);
end loop;
-- No statements other than code statements, pragmas, and labels.
-- Again we allow certain internally generated statements.
Stmt := First (Statements (HSS));
while Present (Stmt) loop
StmtO := Original_Node (Stmt);
if Comes_From_Source (StmtO)
and then not Nkind_In (StmtO, N_Pragma,
N_Label,
N_Code_Statement)
then
Error_Msg_N
("this statement is not allowed in machine code subprogram",
StmtO);
end if;
Next (Stmt);
end loop;
end if;
end Analyze_Code_Statement;
-----------------------------------------------
-- Analyze_Enumeration_Representation_Clause --
-----------------------------------------------
procedure Analyze_Enumeration_Representation_Clause (N : Node_Id) is
Ident : constant Node_Id := Identifier (N);
Aggr : constant Node_Id := Array_Aggregate (N);
Enumtype : Entity_Id;
Elit : Entity_Id;
Expr : Node_Id;
Assoc : Node_Id;
Choice : Node_Id;
Val : Uint;
Err : Boolean := False;
Lo : constant Uint := Expr_Value (Type_Low_Bound (Universal_Integer));
Hi : constant Uint := Expr_Value (Type_High_Bound (Universal_Integer));
Min : Uint;
Max : Uint;
begin
if Ignore_Rep_Clauses then
return;
end if;
-- First some basic error checks
Find_Type (Ident);
Enumtype := Entity (Ident);
if Enumtype = Any_Type
or else Rep_Item_Too_Early (Enumtype, N)
then
return;
else
Enumtype := Underlying_Type (Enumtype);
end if;
if not Is_Enumeration_Type (Enumtype) then
Error_Msg_NE
("enumeration type required, found}",
Ident, First_Subtype (Enumtype));
return;
end if;
-- Ignore rep clause on generic actual type. This will already have
-- been flagged on the template as an error, and this is the safest
-- way to ensure we don't get a junk cascaded message in the instance.
if Is_Generic_Actual_Type (Enumtype) then
return;
-- Type must be in current scope
elsif Scope (Enumtype) /= Current_Scope then
Error_Msg_N ("type must be declared in this scope", Ident);
return;
-- Type must be a first subtype
elsif not Is_First_Subtype (Enumtype) then
Error_Msg_N ("cannot give enumeration rep clause for subtype", N);
return;
-- Ignore duplicate rep clause
elsif Has_Enumeration_Rep_Clause (Enumtype) then
Error_Msg_N ("duplicate enumeration rep clause ignored", N);
return;
-- Don't allow rep clause for standard [wide_[wide_]]character
elsif Is_Standard_Character_Type (Enumtype) then
Error_Msg_N ("enumeration rep clause not allowed for this type", N);
return;
-- Check that the expression is a proper aggregate (no parentheses)
elsif Paren_Count (Aggr) /= 0 then
Error_Msg
("extra parentheses surrounding aggregate not allowed",
First_Sloc (Aggr));
return;
-- All tests passed, so set rep clause in place
else
Set_Has_Enumeration_Rep_Clause (Enumtype);
Set_Has_Enumeration_Rep_Clause (Base_Type (Enumtype));
end if;
-- Now we process the aggregate. Note that we don't use the normal
-- aggregate code for this purpose, because we don't want any of the
-- normal expansion activities, and a number of special semantic
-- rules apply (including the component type being any integer type)
Elit := First_Literal (Enumtype);
-- First the positional entries if any
if Present (Expressions (Aggr)) then
Expr := First (Expressions (Aggr));
while Present (Expr) loop
if No (Elit) then
Error_Msg_N ("too many entries in aggregate", Expr);
return;
end if;
Val := Static_Integer (Expr);
-- Err signals that we found some incorrect entries processing
-- the list. The final checks for completeness and ordering are
-- skipped in this case.
if Val = No_Uint then
Err := True;
elsif Val < Lo or else Hi < Val then
Error_Msg_N ("value outside permitted range", Expr);
Err := True;
end if;
Set_Enumeration_Rep (Elit, Val);
Set_Enumeration_Rep_Expr (Elit, Expr);
Next (Expr);
Next (Elit);
end loop;
end if;
-- Now process the named entries if present
if Present (Component_Associations (Aggr)) then
Assoc := First (Component_Associations (Aggr));
while Present (Assoc) loop
Choice := First (Choices (Assoc));
if Present (Next (Choice)) then
Error_Msg_N
("multiple choice not allowed here", Next (Choice));
Err := True;
end if;
if Nkind (Choice) = N_Others_Choice then
Error_Msg_N ("others choice not allowed here", Choice);
Err := True;
elsif Nkind (Choice) = N_Range then
-- ??? should allow zero/one element range here
Error_Msg_N ("range not allowed here", Choice);
Err := True;
else
Analyze_And_Resolve (Choice, Enumtype);
if Is_Entity_Name (Choice)
and then Is_Type (Entity (Choice))
then
Error_Msg_N ("subtype name not allowed here", Choice);
Err := True;
-- ??? should allow static subtype with zero/one entry
elsif Etype (Choice) = Base_Type (Enumtype) then
if not Is_Static_Expression (Choice) then
Flag_Non_Static_Expr
("non-static expression used for choice!", Choice);
Err := True;
else
Elit := Expr_Value_E (Choice);
if Present (Enumeration_Rep_Expr (Elit)) then
Error_Msg_Sloc := Sloc (Enumeration_Rep_Expr (Elit));
Error_Msg_NE
("representation for& previously given#",
Choice, Elit);
Err := True;
end if;
Set_Enumeration_Rep_Expr (Elit, Choice);
Expr := Expression (Assoc);
Val := Static_Integer (Expr);
if Val = No_Uint then
Err := True;
elsif Val < Lo or else Hi < Val then
Error_Msg_N ("value outside permitted range", Expr);
Err := True;
end if;
Set_Enumeration_Rep (Elit, Val);
end if;
end if;
end if;
Next (Assoc);
end loop;
end if;
-- Aggregate is fully processed. Now we check that a full set of
-- representations was given, and that they are in range and in order.
-- These checks are only done if no other errors occurred.
if not Err then
Min := No_Uint;
Max := No_Uint;
Elit := First_Literal (Enumtype);
while Present (Elit) loop
if No (Enumeration_Rep_Expr (Elit)) then
Error_Msg_NE ("missing representation for&!", N, Elit);
else
Val := Enumeration_Rep (Elit);
if Min = No_Uint then
Min := Val;
end if;
if Val /= No_Uint then
if Max /= No_Uint and then Val <= Max then
Error_Msg_NE
("enumeration value for& not ordered!",
Enumeration_Rep_Expr (Elit), Elit);
end if;
Max := Val;
end if;
-- If there is at least one literal whose representation
-- is not equal to the Pos value, then note that this
-- enumeration type has a non-standard representation.
if Val /= Enumeration_Pos (Elit) then
Set_Has_Non_Standard_Rep (Base_Type (Enumtype));
end if;
end if;
Next (Elit);
end loop;
-- Now set proper size information
declare
Minsize : Uint := UI_From_Int (Minimum_Size (Enumtype));
begin
if Has_Size_Clause (Enumtype) then
if Esize (Enumtype) >= Minsize then
null;
else
Minsize :=
UI_From_Int (Minimum_Size (Enumtype, Biased => True));
if Esize (Enumtype) < Minsize then
Error_Msg_N ("previously given size is too small", N);
else
Set_Has_Biased_Representation (Enumtype);
end if;
end if;
else
Set_RM_Size (Enumtype, Minsize);
Set_Enum_Esize (Enumtype);
end if;
Set_RM_Size (Base_Type (Enumtype), RM_Size (Enumtype));
Set_Esize (Base_Type (Enumtype), Esize (Enumtype));
Set_Alignment (Base_Type (Enumtype), Alignment (Enumtype));
end;
end if;
-- We repeat the too late test in case it froze itself!
if Rep_Item_Too_Late (Enumtype, N) then
null;
end if;
end Analyze_Enumeration_Representation_Clause;
----------------------------
-- Analyze_Free_Statement --
----------------------------
procedure Analyze_Free_Statement (N : Node_Id) is
begin
Analyze (Expression (N));
end Analyze_Free_Statement;
---------------------------
-- Analyze_Freeze_Entity --
---------------------------
procedure Analyze_Freeze_Entity (N : Node_Id) is
E : constant Entity_Id := Entity (N);
begin
-- For tagged types covering interfaces add internal entities that link
-- the primitives of the interfaces with the primitives that cover them.
-- Note: These entities were originally generated only when generating
-- code because their main purpose was to provide support to initialize
-- the secondary dispatch tables. They are now generated also when
-- compiling with no code generation to provide ASIS the relationship
-- between interface primitives and tagged type primitives.
if Ada_Version >= Ada_05
and then Ekind (E) = E_Record_Type
and then Is_Tagged_Type (E)
and then not Is_Interface (E)
and then Has_Interfaces (E)
then
Add_Internal_Interface_Entities (E);
end if;
end Analyze_Freeze_Entity;
------------------------------------------
-- Analyze_Record_Representation_Clause --
------------------------------------------
procedure Analyze_Record_Representation_Clause (N : Node_Id) is
Loc : constant Source_Ptr := Sloc (N);
Ident : constant Node_Id := Identifier (N);
Rectype : Entity_Id;
Fent : Entity_Id;
CC : Node_Id;
Posit : Uint;
Fbit : Uint;
Lbit : Uint;
Hbit : Uint := Uint_0;
Comp : Entity_Id;
Ocomp : Entity_Id;
Pcomp : Entity_Id;
Biased : Boolean;
Max_Bit_So_Far : Uint;
-- Records the maximum bit position so far. If all field positions
-- are monotonically increasing, then we can skip the circuit for
-- checking for overlap, since no overlap is possible.
Tagged_Parent : Entity_Id := Empty;
-- This is set in the case of a derived tagged type for which we have
-- Is_Fully_Repped_Tagged_Type True (indicating that all components are
-- positioned by record representation clauses). In this case we must
-- check for overlap between components of this tagged type, and the
-- components of its parent. Tagged_Parent will point to this parent
-- type. For all other cases Tagged_Parent is left set to Empty.
Parent_Last_Bit : Uint;
-- Relevant only if Tagged_Parent is set, Parent_Last_Bit indicates the
-- last bit position for any field in the parent type. We only need to
-- check overlap for fields starting below this point.
Overlap_Check_Required : Boolean;
-- Used to keep track of whether or not an overlap check is required
Ccount : Natural := 0;
-- Number of component clauses in record rep clause
CR_Pragma : Node_Id := Empty;
-- Points to N_Pragma node if Complete_Representation pragma present
begin
if Ignore_Rep_Clauses then
return;
end if;
Find_Type (Ident);
Rectype := Entity (Ident);
if Rectype = Any_Type
or else Rep_Item_Too_Early (Rectype, N)
then
return;
else
Rectype := Underlying_Type (Rectype);
end if;
-- First some basic error checks
if not Is_Record_Type (Rectype) then
Error_Msg_NE
("record type required, found}", Ident, First_Subtype (Rectype));
return;
elsif Is_Unchecked_Union (Rectype) then
Error_Msg_N
("record rep clause not allowed for Unchecked_Union", N);
elsif Scope (Rectype) /= Current_Scope then
Error_Msg_N ("type must be declared in this scope", N);
return;
elsif not Is_First_Subtype (Rectype) then
Error_Msg_N ("cannot give record rep clause for subtype", N);
return;
elsif Has_Record_Rep_Clause (Rectype) then
Error_Msg_N ("duplicate record rep clause ignored", N);
return;
elsif Rep_Item_Too_Late (Rectype, N) then
return;
end if;
if Present (Mod_Clause (N)) then
declare
Loc : constant Source_Ptr := Sloc (N);
M : constant Node_Id := Mod_Clause (N);
P : constant List_Id := Pragmas_Before (M);
AtM_Nod : Node_Id;
Mod_Val : Uint;
pragma Warnings (Off, Mod_Val);
begin
Check_Restriction (No_Obsolescent_Features, Mod_Clause (N));
if Warn_On_Obsolescent_Feature then
Error_Msg_N
("mod clause is an obsolescent feature (RM J.8)?", N);
Error_Msg_N
("\use alignment attribute definition clause instead?", N);
end if;
if Present (P) then
Analyze_List (P);
end if;
-- In ASIS_Mode mode, expansion is disabled, but we must convert
-- the Mod clause into an alignment clause anyway, so that the
-- back-end can compute and back-annotate properly the size and
-- alignment of types that may include this record.
-- This seems dubious, this destroys the source tree in a manner
-- not detectable by ASIS ???
if Operating_Mode = Check_Semantics
and then ASIS_Mode
then
AtM_Nod :=
Make_Attribute_Definition_Clause (Loc,
Name => New_Reference_To (Base_Type (Rectype), Loc),
Chars => Name_Alignment,
Expression => Relocate_Node (Expression (M)));
Set_From_At_Mod (AtM_Nod);
Insert_After (N, AtM_Nod);
Mod_Val := Get_Alignment_Value (Expression (AtM_Nod));
Set_Mod_Clause (N, Empty);
else
-- Get the alignment value to perform error checking
Mod_Val := Get_Alignment_Value (Expression (M));
end if;
end;
end if;
-- For untagged types, clear any existing component clauses for the
-- type. If the type is derived, this is what allows us to override
-- a rep clause for the parent. For type extensions, the representation
-- of the inherited components is inherited, so we want to keep previous
-- component clauses for completeness.
if not Is_Tagged_Type (Rectype) then
Comp := First_Component_Or_Discriminant (Rectype);
while Present (Comp) loop
Set_Component_Clause (Comp, Empty);
Next_Component_Or_Discriminant (Comp);
end loop;
end if;
-- See if we have a fully repped derived tagged type
declare
PS : constant Entity_Id := Parent_Subtype (Rectype);
begin
if Present (PS) and then Is_Fully_Repped_Tagged_Type (PS) then
Tagged_Parent := PS;
-- Find maximum bit of any component of the parent type
Parent_Last_Bit := UI_From_Int (System_Address_Size - 1);
Pcomp := First_Entity (Tagged_Parent);
while Present (Pcomp) loop
if Ekind (Pcomp) = E_Discriminant
or else
Ekind (Pcomp) = E_Component
then
if Component_Bit_Offset (Pcomp) /= No_Uint
and then Known_Static_Esize (Pcomp)
then
Parent_Last_Bit :=
UI_Max
(Parent_Last_Bit,
Component_Bit_Offset (Pcomp) + Esize (Pcomp) - 1);
end if;
Next_Entity (Pcomp);
end if;
end loop;
end if;
end;
-- All done if no component clauses
CC := First (Component_Clauses (N));
if No (CC) then
return;
end if;
-- If a tag is present, then create a component clause that places it
-- at the start of the record (otherwise gigi may place it after other
-- fields that have rep clauses).
Fent := First_Entity (Rectype);
if Nkind (Fent) = N_Defining_Identifier
and then Chars (Fent) = Name_uTag
then
Set_Component_Bit_Offset (Fent, Uint_0);
Set_Normalized_Position (Fent, Uint_0);
Set_Normalized_First_Bit (Fent, Uint_0);
Set_Normalized_Position_Max (Fent, Uint_0);
Init_Esize (Fent, System_Address_Size);
Set_Component_Clause (Fent,
Make_Component_Clause (Loc,
Component_Name =>
Make_Identifier (Loc,
Chars => Name_uTag),
Position =>
Make_Integer_Literal (Loc,
Intval => Uint_0),
First_Bit =>
Make_Integer_Literal (Loc,
Intval => Uint_0),
Last_Bit =>
Make_Integer_Literal (Loc,
UI_From_Int (System_Address_Size))));
Ccount := Ccount + 1;
end if;
-- A representation like this applies to the base type
Set_Has_Record_Rep_Clause (Base_Type (Rectype));
Set_Has_Non_Standard_Rep (Base_Type (Rectype));
Set_Has_Specified_Layout (Base_Type (Rectype));
Max_Bit_So_Far := Uint_Minus_1;
Overlap_Check_Required := False;
-- Process the component clauses
while Present (CC) loop
-- Pragma
if Nkind (CC) = N_Pragma then
Analyze (CC);
-- The only pragma of interest is Complete_Representation
if Pragma_Name (CC) = Name_Complete_Representation then
CR_Pragma := CC;
end if;
-- Processing for real component clause
else
Ccount := Ccount + 1;
Posit := Static_Integer (Position (CC));
Fbit := Static_Integer (First_Bit (CC));
Lbit := Static_Integer (Last_Bit (CC));
if Posit /= No_Uint
and then Fbit /= No_Uint
and then Lbit /= No_Uint
then
if Posit < 0 then
Error_Msg_N
("position cannot be negative", Position (CC));
elsif Fbit < 0 then
Error_Msg_N
("first bit cannot be negative", First_Bit (CC));
-- The Last_Bit specified in a component clause must not be
-- less than the First_Bit minus one (RM-13.5.1(10)).
elsif Lbit < Fbit - 1 then
Error_Msg_N
("last bit cannot be less than first bit minus one",
Last_Bit (CC));
-- Values look OK, so find the corresponding record component
-- Even though the syntax allows an attribute reference for
-- implementation-defined components, GNAT does not allow the
-- tag to get an explicit position.
elsif Nkind (Component_Name (CC)) = N_Attribute_Reference then
if Attribute_Name (Component_Name (CC)) = Name_Tag then
Error_Msg_N ("position of tag cannot be specified", CC);
else
Error_Msg_N ("illegal component name", CC);
end if;
else
Comp := First_Entity (Rectype);
while Present (Comp) loop
exit when Chars (Comp) = Chars (Component_Name (CC));
Next_Entity (Comp);
end loop;
if No (Comp) then
-- Maybe component of base type that is absent from
-- statically constrained first subtype.
Comp := First_Entity (Base_Type (Rectype));
while Present (Comp) loop
exit when Chars (Comp) = Chars (Component_Name (CC));
Next_Entity (Comp);
end loop;
end if;
if No (Comp) then
Error_Msg_N
("component clause is for non-existent field", CC);
elsif Present (Component_Clause (Comp)) then
-- Diagnose duplicate rep clause, or check consistency
-- if this is an inherited component. In a double fault,
-- there may be a duplicate inconsistent clause for an
-- inherited component.
if Scope (Original_Record_Component (Comp)) = Rectype
or else Parent (Component_Clause (Comp)) = N
then
Error_Msg_Sloc := Sloc (Component_Clause (Comp));
Error_Msg_N ("component clause previously given#", CC);
else
declare
Rep1 : constant Node_Id := Component_Clause (Comp);
begin
if Intval (Position (Rep1)) /=
Intval (Position (CC))
or else Intval (First_Bit (Rep1)) /=
Intval (First_Bit (CC))
or else Intval (Last_Bit (Rep1)) /=
Intval (Last_Bit (CC))
then
Error_Msg_N ("component clause inconsistent "
& "with representation of ancestor", CC);
elsif Warn_On_Redundant_Constructs then
Error_Msg_N ("?redundant component clause "
& "for inherited component!", CC);
end if;
end;
end if;
-- Normal case where this is the first component clause we
-- have seen for this entity, so set it up properly.
else
-- Make reference for field in record rep clause and set
-- appropriate entity field in the field identifier.
Generate_Reference
(Comp, Component_Name (CC), Set_Ref => False);
Set_Entity (Component_Name (CC), Comp);
-- Update Fbit and Lbit to the actual bit number
Fbit := Fbit + UI_From_Int (SSU) * Posit;
Lbit := Lbit + UI_From_Int (SSU) * Posit;
if Fbit <= Max_Bit_So_Far then
Overlap_Check_Required := True;
else
Max_Bit_So_Far := Lbit;
end if;
if Has_Size_Clause (Rectype)
and then Esize (Rectype) <= Lbit
then
Error_Msg_N
("bit number out of range of specified size",
Last_Bit (CC));
else
Set_Component_Clause (Comp, CC);
Set_Component_Bit_Offset (Comp, Fbit);
Set_Esize (Comp, 1 + (Lbit - Fbit));
Set_Normalized_First_Bit (Comp, Fbit mod SSU);
Set_Normalized_Position (Comp, Fbit / SSU);
Set_Normalized_Position_Max
(Fent, Normalized_Position (Fent));
if Is_Tagged_Type (Rectype)
and then Fbit < System_Address_Size
then
Error_Msg_NE
("component overlaps tag field of&",
Component_Name (CC), Rectype);
end if;
-- This information is also set in the corresponding
-- component of the base type, found by accessing the
-- Original_Record_Component link if it is present.
Ocomp := Original_Record_Component (Comp);
if Hbit < Lbit then
Hbit := Lbit;
end if;
Check_Size
(Component_Name (CC),
Etype (Comp),
Esize (Comp),
Biased);
Set_Has_Biased_Representation (Comp, Biased);
if Biased and Warn_On_Biased_Representation then
Error_Msg_F
("?component clause forces biased "
& "representation", CC);
end if;
if Present (Ocomp) then
Set_Component_Clause (Ocomp, CC);
Set_Component_Bit_Offset (Ocomp, Fbit);
Set_Normalized_First_Bit (Ocomp, Fbit mod SSU);
Set_Normalized_Position (Ocomp, Fbit / SSU);
Set_Esize (Ocomp, 1 + (Lbit - Fbit));
Set_Normalized_Position_Max
(Ocomp, Normalized_Position (Ocomp));
Set_Has_Biased_Representation
(Ocomp, Has_Biased_Representation (Comp));
end if;
if Esize (Comp) < 0 then
Error_Msg_N ("component size is negative", CC);
end if;
end if;
-- If OK component size, check parent type overlap if
-- this component might overlap a parent field.
if Present (Tagged_Parent)
and then Fbit <= Parent_Last_Bit
then
Pcomp := First_Entity (Tagged_Parent);
while Present (Pcomp) loop
if (Ekind (Pcomp) = E_Discriminant
or else
Ekind (Pcomp) = E_Component)
and then not Is_Tag (Pcomp)
and then Chars (Pcomp) /= Name_uParent
then
Check_Component_Overlap (Comp, Pcomp);
end if;
Next_Entity (Pcomp);
end loop;
end if;
end if;
end if;
end if;
end if;
Next (CC);
end loop;
-- Now that we have processed all the component clauses, check for
-- overlap. We have to leave this till last, since the components can
-- appear in any arbitrary order in the representation clause.
-- We do not need this check if all specified ranges were monotonic,
-- as recorded by Overlap_Check_Required being False at this stage.
-- This first section checks if there are any overlapping entries at
-- all. It does this by sorting all entries and then seeing if there are
-- any overlaps. If there are none, then that is decisive, but if there
-- are overlaps, they may still be OK (they may result from fields in
-- different variants).
if Overlap_Check_Required then
Overlap_Check1 : declare
OC_Fbit : array (0 .. Ccount) of Uint;
-- First-bit values for component clauses, the value is the offset
-- of the first bit of the field from start of record. The zero
-- entry is for use in sorting.
OC_Lbit : array (0 .. Ccount) of Uint;
-- Last-bit values for component clauses, the value is the offset
-- of the last bit of the field from start of record. The zero
-- entry is for use in sorting.
OC_Count : Natural := 0;
-- Count of entries in OC_Fbit and OC_Lbit
function OC_Lt (Op1, Op2 : Natural) return Boolean;
-- Compare routine for Sort
procedure OC_Move (From : Natural; To : Natural);
-- Move routine for Sort
package Sorting is new GNAT.Heap_Sort_G (OC_Move, OC_Lt);
-----------
-- OC_Lt --
-----------
function OC_Lt (Op1, Op2 : Natural) return Boolean is
begin
return OC_Fbit (Op1) < OC_Fbit (Op2);
end OC_Lt;
-------------
-- OC_Move --
-------------
procedure OC_Move (From : Natural; To : Natural) is
begin
OC_Fbit (To) := OC_Fbit (From);
OC_Lbit (To) := OC_Lbit (From);
end OC_Move;
-- Start of processing for Overlap_Check
begin
CC := First (Component_Clauses (N));
while Present (CC) loop
if Nkind (CC) /= N_Pragma then
Posit := Static_Integer (Position (CC));
Fbit := Static_Integer (First_Bit (CC));
Lbit := Static_Integer (Last_Bit (CC));
if Posit /= No_Uint
and then Fbit /= No_Uint
and then Lbit /= No_Uint
then
OC_Count := OC_Count + 1;
Posit := Posit * SSU;
OC_Fbit (OC_Count) := Fbit + Posit;
OC_Lbit (OC_Count) := Lbit + Posit;
end if;
end if;
Next (CC);
end loop;
Sorting.Sort (OC_Count);
Overlap_Check_Required := False;
for J in 1 .. OC_Count - 1 loop
if OC_Lbit (J) >= OC_Fbit (J + 1) then
Overlap_Check_Required := True;
exit;
end if;
end loop;
end Overlap_Check1;
end if;
-- If Overlap_Check_Required is still True, then we have to do the full
-- scale overlap check, since we have at least two fields that do
-- overlap, and we need to know if that is OK since they are in
-- different variant, or whether we have a definite problem.
if Overlap_Check_Required then
Overlap_Check2 : declare
C1_Ent, C2_Ent : Entity_Id;
-- Entities of components being checked for overlap
Clist : Node_Id;
-- Component_List node whose Component_Items are being checked
Citem : Node_Id;
-- Component declaration for component being checked
begin
C1_Ent := First_Entity (Base_Type (Rectype));
-- Loop through all components in record. For each component check
-- for overlap with any of the preceding elements on the component
-- list containing the component and also, if the component is in
-- a variant, check against components outside the case structure.
-- This latter test is repeated recursively up the variant tree.
Main_Component_Loop : while Present (C1_Ent) loop
if Ekind (C1_Ent) /= E_Component
and then Ekind (C1_Ent) /= E_Discriminant
then
goto Continue_Main_Component_Loop;
end if;
-- Skip overlap check if entity has no declaration node. This
-- happens with discriminants in constrained derived types.
-- Probably we are missing some checks as a result, but that
-- does not seem terribly serious ???
if No (Declaration_Node (C1_Ent)) then
goto Continue_Main_Component_Loop;
end if;
Clist := Parent (List_Containing (Declaration_Node (C1_Ent)));
-- Loop through component lists that need checking. Check the
-- current component list and all lists in variants above us.
Component_List_Loop : loop
-- If derived type definition, go to full declaration
-- If at outer level, check discriminants if there are any.
if Nkind (Clist) = N_Derived_Type_Definition then
Clist := Parent (Clist);
end if;
-- Outer level of record definition, check discriminants
if Nkind_In (Clist, N_Full_Type_Declaration,
N_Private_Type_Declaration)
then
if Has_Discriminants (Defining_Identifier (Clist)) then
C2_Ent :=
First_Discriminant (Defining_Identifier (Clist));
while Present (C2_Ent) loop
exit when C1_Ent = C2_Ent;
Check_Component_Overlap (C1_Ent, C2_Ent);
Next_Discriminant (C2_Ent);
end loop;
end if;
-- Record extension case
elsif Nkind (Clist) = N_Derived_Type_Definition then
Clist := Empty;
-- Otherwise check one component list
else
Citem := First (Component_Items (Clist));
while Present (Citem) loop
if Nkind (Citem) = N_Component_Declaration then
C2_Ent := Defining_Identifier (Citem);
exit when C1_Ent = C2_Ent;
Check_Component_Overlap (C1_Ent, C2_Ent);
end if;
Next (Citem);
end loop;
end if;
-- Check for variants above us (the parent of the Clist can
-- be a variant, in which case its parent is a variant part,
-- and the parent of the variant part is a component list
-- whose components must all be checked against the current
-- component for overlap).
if Nkind (Parent (Clist)) = N_Variant then
Clist := Parent (Parent (Parent (Clist)));
-- Check for possible discriminant part in record, this is
-- treated essentially as another level in the recursion.
-- For this case the parent of the component list is the
-- record definition, and its parent is the full type
-- declaration containing the discriminant specifications.
elsif Nkind (Parent (Clist)) = N_Record_Definition then
Clist := Parent (Parent ((Clist)));
-- If neither of these two cases, we are at the top of
-- the tree.
else
exit Component_List_Loop;
end if;
end loop Component_List_Loop;
<<Continue_Main_Component_Loop>>
Next_Entity (C1_Ent);
end loop Main_Component_Loop;
end Overlap_Check2;
end if;
-- For records that have component clauses for all components, and whose
-- size is less than or equal to 32, we need to know the size in the
-- front end to activate possible packed array processing where the
-- component type is a record.
-- At this stage Hbit + 1 represents the first unused bit from all the
-- component clauses processed, so if the component clauses are
-- complete, then this is the length of the record.
-- For records longer than System.Storage_Unit, and for those where not
-- all components have component clauses, the back end determines the
-- length (it may for example be appropriate to round up the size
-- to some convenient boundary, based on alignment considerations, etc).
if Unknown_RM_Size (Rectype) and then Hbit + 1 <= 32 then
-- Nothing to do if at least one component has no component clause
Comp := First_Component_Or_Discriminant (Rectype);
while Present (Comp) loop
exit when No (Component_Clause (Comp));
Next_Component_Or_Discriminant (Comp);
end loop;
-- If we fall out of loop, all components have component clauses
-- and so we can set the size to the maximum value.
if No (Comp) then
Set_RM_Size (Rectype, Hbit + 1);
end if;
end if;
-- Check missing components if Complete_Representation pragma appeared
if Present (CR_Pragma) then
Comp := First_Component_Or_Discriminant (Rectype);
while Present (Comp) loop
if No (Component_Clause (Comp)) then
Error_Msg_NE
("missing component clause for &", CR_Pragma, Comp);
end if;
Next_Component_Or_Discriminant (Comp);
end loop;
-- If no Complete_Representation pragma, warn if missing components
elsif Warn_On_Unrepped_Components then
declare
Num_Repped_Components : Nat := 0;
Num_Unrepped_Components : Nat := 0;
begin
-- First count number of repped and unrepped components
Comp := First_Component_Or_Discriminant (Rectype);
while Present (Comp) loop
if Present (Component_Clause (Comp)) then
Num_Repped_Components := Num_Repped_Components + 1;
else
Num_Unrepped_Components := Num_Unrepped_Components + 1;
end if;
Next_Component_Or_Discriminant (Comp);
end loop;
-- We are only interested in the case where there is at least one
-- unrepped component, and at least half the components have rep
-- clauses. We figure that if less than half have them, then the
-- partial rep clause is really intentional. If the component
-- type has no underlying type set at this point (as for a generic
-- formal type), we don't know enough to give a warning on the
-- component.
if Num_Unrepped_Components > 0
and then Num_Unrepped_Components < Num_Repped_Components
then
Comp := First_Component_Or_Discriminant (Rectype);
while Present (Comp) loop
if No (Component_Clause (Comp))
and then Comes_From_Source (Comp)
and then Present (Underlying_Type (Etype (Comp)))
and then (Is_Scalar_Type (Underlying_Type (Etype (Comp)))
or else Size_Known_At_Compile_Time
(Underlying_Type (Etype (Comp))))
and then not Has_Warnings_Off (Rectype)
then
Error_Msg_Sloc := Sloc (Comp);
Error_Msg_NE
("?no component clause given for & declared #",
N, Comp);
end if;
Next_Component_Or_Discriminant (Comp);
end loop;
end if;
end;
end if;
end Analyze_Record_Representation_Clause;
-----------------------------
-- Check_Component_Overlap --
-----------------------------
procedure Check_Component_Overlap (C1_Ent, C2_Ent : Entity_Id) is
begin
if Present (Component_Clause (C1_Ent))
and then Present (Component_Clause (C2_Ent))
then
-- Exclude odd case where we have two tag fields in the same record,
-- both at location zero. This seems a bit strange, but it seems to
-- happen in some circumstances ???
if Chars (C1_Ent) = Name_uTag
and then Chars (C2_Ent) = Name_uTag
then
return;
end if;
-- Here we check if the two fields overlap
declare
S1 : constant Uint := Component_Bit_Offset (C1_Ent);
S2 : constant Uint := Component_Bit_Offset (C2_Ent);
E1 : constant Uint := S1 + Esize (C1_Ent);
E2 : constant Uint := S2 + Esize (C2_Ent);
begin
if E2 <= S1 or else E1 <= S2 then
null;
else
Error_Msg_Node_2 :=
Component_Name (Component_Clause (C2_Ent));
Error_Msg_Sloc := Sloc (Error_Msg_Node_2);
Error_Msg_Node_1 :=
Component_Name (Component_Clause (C1_Ent));
Error_Msg_N
("component& overlaps & #",
Component_Name (Component_Clause (C1_Ent)));
end if;
end;
end if;
end Check_Component_Overlap;
-----------------------------------
-- Check_Constant_Address_Clause --
-----------------------------------
procedure Check_Constant_Address_Clause
(Expr : Node_Id;
U_Ent : Entity_Id)
is
procedure Check_At_Constant_Address (Nod : Node_Id);
-- Checks that the given node N represents a name whose 'Address is
-- constant (in the same sense as OK_Constant_Address_Clause, i.e. the
-- address value is the same at the point of declaration of U_Ent and at
-- the time of elaboration of the address clause.
procedure Check_Expr_Constants (Nod : Node_Id);
-- Checks that Nod meets the requirements for a constant address clause
-- in the sense of the enclosing procedure.
procedure Check_List_Constants (Lst : List_Id);
-- Check that all elements of list Lst meet the requirements for a
-- constant address clause in the sense of the enclosing procedure.
-------------------------------
-- Check_At_Constant_Address --
-------------------------------
procedure Check_At_Constant_Address (Nod : Node_Id) is
begin
if Is_Entity_Name (Nod) then
if Present (Address_Clause (Entity ((Nod)))) then
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
Error_Msg_NE
("address for& cannot" &
" depend on another address clause! (RM 13.1(22))!",
Nod, U_Ent);
elsif In_Same_Source_Unit (Entity (Nod), U_Ent)
and then Sloc (U_Ent) < Sloc (Entity (Nod))
then
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
Error_Msg_Node_2 := U_Ent;
Error_Msg_NE
("\& must be defined before & (RM 13.1(22))!",
Nod, Entity (Nod));
end if;
elsif Nkind (Nod) = N_Selected_Component then
declare
T : constant Entity_Id := Etype (Prefix (Nod));
begin
if (Is_Record_Type (T)
and then Has_Discriminants (T))
or else
(Is_Access_Type (T)
and then Is_Record_Type (Designated_Type (T))
and then Has_Discriminants (Designated_Type (T)))
then
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
Error_Msg_N
("\address cannot depend on component" &
" of discriminated record (RM 13.1(22))!",
Nod);
else
Check_At_Constant_Address (Prefix (Nod));
end if;
end;
elsif Nkind (Nod) = N_Indexed_Component then
Check_At_Constant_Address (Prefix (Nod));
Check_List_Constants (Expressions (Nod));
else
Check_Expr_Constants (Nod);
end if;
end Check_At_Constant_Address;
--------------------------
-- Check_Expr_Constants --
--------------------------
procedure Check_Expr_Constants (Nod : Node_Id) is
Loc_U_Ent : constant Source_Ptr := Sloc (U_Ent);
Ent : Entity_Id := Empty;
begin
if Nkind (Nod) in N_Has_Etype
and then Etype (Nod) = Any_Type
then
return;
end if;
case Nkind (Nod) is
when N_Empty | N_Error =>
return;
when N_Identifier | N_Expanded_Name =>
Ent := Entity (Nod);
-- We need to look at the original node if it is different
-- from the node, since we may have rewritten things and
-- substituted an identifier representing the rewrite.
if Original_Node (Nod) /= Nod then
Check_Expr_Constants (Original_Node (Nod));
-- If the node is an object declaration without initial
-- value, some code has been expanded, and the expression
-- is not constant, even if the constituents might be
-- acceptable, as in A'Address + offset.
if Ekind (Ent) = E_Variable
and then
Nkind (Declaration_Node (Ent)) = N_Object_Declaration
and then
No (Expression (Declaration_Node (Ent)))
then
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
-- If entity is constant, it may be the result of expanding
-- a check. We must verify that its declaration appears
-- before the object in question, else we also reject the
-- address clause.
elsif Ekind (Ent) = E_Constant
and then In_Same_Source_Unit (Ent, U_Ent)
and then Sloc (Ent) > Loc_U_Ent
then
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
end if;
return;
end if;
-- Otherwise look at the identifier and see if it is OK
if Ekind (Ent) = E_Named_Integer
or else
Ekind (Ent) = E_Named_Real
or else
Is_Type (Ent)
then
return;
elsif
Ekind (Ent) = E_Constant
or else
Ekind (Ent) = E_In_Parameter
then
-- This is the case where we must have Ent defined before
-- U_Ent. Clearly if they are in different units this
-- requirement is met since the unit containing Ent is
-- already processed.
if not In_Same_Source_Unit (Ent, U_Ent) then
return;
-- Otherwise location of Ent must be before the location
-- of U_Ent, that's what prior defined means.
elsif Sloc (Ent) < Loc_U_Ent then
return;
else
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
Error_Msg_Node_2 := U_Ent;
Error_Msg_NE
("\& must be defined before & (RM 13.1(22))!",
Nod, Ent);
end if;
elsif Nkind (Original_Node (Nod)) = N_Function_Call then
Check_Expr_Constants (Original_Node (Nod));
else
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
if Comes_From_Source (Ent) then
Error_Msg_NE
("\reference to variable& not allowed"
& " (RM 13.1(22))!", Nod, Ent);
else
Error_Msg_N
("non-static expression not allowed"
& " (RM 13.1(22))!", Nod);
end if;
end if;
when N_Integer_Literal =>
-- If this is a rewritten unchecked conversion, in a system
-- where Address is an integer type, always use the base type
-- for a literal value. This is user-friendly and prevents
-- order-of-elaboration issues with instances of unchecked
-- conversion.
if Nkind (Original_Node (Nod)) = N_Function_Call then
Set_Etype (Nod, Base_Type (Etype (Nod)));
end if;
when N_Real_Literal |
N_String_Literal |
N_Character_Literal =>
return;
when N_Range =>
Check_Expr_Constants (Low_Bound (Nod));
Check_Expr_Constants (High_Bound (Nod));
when N_Explicit_Dereference =>
Check_Expr_Constants (Prefix (Nod));
when N_Indexed_Component =>
Check_Expr_Constants (Prefix (Nod));
Check_List_Constants (Expressions (Nod));
when N_Slice =>
Check_Expr_Constants (Prefix (Nod));
Check_Expr_Constants (Discrete_Range (Nod));
when N_Selected_Component =>
Check_Expr_Constants (Prefix (Nod));
when N_Attribute_Reference =>
if Attribute_Name (Nod) = Name_Address
or else
Attribute_Name (Nod) = Name_Access
or else
Attribute_Name (Nod) = Name_Unchecked_Access
or else
Attribute_Name (Nod) = Name_Unrestricted_Access
then
Check_At_Constant_Address (Prefix (Nod));
else
Check_Expr_Constants (Prefix (Nod));
Check_List_Constants (Expressions (Nod));
end if;
when N_Aggregate =>
Check_List_Constants (Component_Associations (Nod));
Check_List_Constants (Expressions (Nod));
when N_Component_Association =>
Check_Expr_Constants (Expression (Nod));
when N_Extension_Aggregate =>
Check_Expr_Constants (Ancestor_Part (Nod));
Check_List_Constants (Component_Associations (Nod));
Check_List_Constants (Expressions (Nod));
when N_Null =>
return;
when N_Binary_Op | N_Short_Circuit | N_Membership_Test =>
Check_Expr_Constants (Left_Opnd (Nod));
Check_Expr_Constants (Right_Opnd (Nod));
when N_Unary_Op =>
Check_Expr_Constants (Right_Opnd (Nod));
when N_Type_Conversion |
N_Qualified_Expression |
N_Allocator =>
Check_Expr_Constants (Expression (Nod));
when N_Unchecked_Type_Conversion =>
Check_Expr_Constants (Expression (Nod));
-- If this is a rewritten unchecked conversion, subtypes in
-- this node are those created within the instance. To avoid
-- order of elaboration issues, replace them with their base
-- types. Note that address clauses can cause order of
-- elaboration problems because they are elaborated by the
-- back-end at the point of definition, and may mention
-- entities declared in between (as long as everything is
-- static). It is user-friendly to allow unchecked conversions
-- in this context.
if Nkind (Original_Node (Nod)) = N_Function_Call then
Set_Etype (Expression (Nod),
Base_Type (Etype (Expression (Nod))));
Set_Etype (Nod, Base_Type (Etype (Nod)));
end if;
when N_Function_Call =>
if not Is_Pure (Entity (Name (Nod))) then
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
Error_Msg_NE
("\function & is not pure (RM 13.1(22))!",
Nod, Entity (Name (Nod)));
else
Check_List_Constants (Parameter_Associations (Nod));
end if;
when N_Parameter_Association =>
Check_Expr_Constants (Explicit_Actual_Parameter (Nod));
when others =>
Error_Msg_NE
("invalid address clause for initialized object &!",
Nod, U_Ent);
Error_Msg_NE
("\must be constant defined before& (RM 13.1(22))!",
Nod, U_Ent);
end case;
end Check_Expr_Constants;
--------------------------
-- Check_List_Constants --
--------------------------
procedure Check_List_Constants (Lst : List_Id) is
Nod1 : Node_Id;
begin
if Present (Lst) then
Nod1 := First (Lst);
while Present (Nod1) loop
Check_Expr_Constants (Nod1);
Next (Nod1);
end loop;
end if;
end Check_List_Constants;
-- Start of processing for Check_Constant_Address_Clause
begin
Check_Expr_Constants (Expr);
end Check_Constant_Address_Clause;
----------------
-- Check_Size --
----------------
procedure Check_Size
(N : Node_Id;
T : Entity_Id;
Siz : Uint;
Biased : out Boolean)
is
UT : constant Entity_Id := Underlying_Type (T);
M : Uint;
begin
Biased := False;
-- Dismiss cases for generic types or types with previous errors
if No (UT)
or else UT = Any_Type
or else Is_Generic_Type (UT)
or else Is_Generic_Type (Root_Type (UT))
then
return;
-- Check case of bit packed array
elsif Is_Array_Type (UT)
and then Known_Static_Component_Size (UT)
and then Is_Bit_Packed_Array (UT)
then
declare
Asiz : Uint;
Indx : Node_Id;
Ityp : Entity_Id;
begin
Asiz := Component_Size (UT);
Indx := First_Index (UT);
loop
Ityp := Etype (Indx);
-- If non-static bound, then we are not in the business of
-- trying to check the length, and indeed an error will be
-- issued elsewhere, since sizes of non-static array types
-- cannot be set implicitly or explicitly.
if not Is_Static_Subtype (Ityp) then
return;
end if;
-- Otherwise accumulate next dimension
Asiz := Asiz * (Expr_Value (Type_High_Bound (Ityp)) -
Expr_Value (Type_Low_Bound (Ityp)) +
Uint_1);
Next_Index (Indx);
exit when No (Indx);
end loop;
if Asiz <= Siz then
return;
else
Error_Msg_Uint_1 := Asiz;
Error_Msg_NE
("size for& too small, minimum allowed is ^", N, T);
Set_Esize (T, Asiz);
Set_RM_Size (T, Asiz);
end if;
end;
-- All other composite types are ignored
elsif Is_Composite_Type (UT) then
return;
-- For fixed-point types, don't check minimum if type is not frozen,
-- since we don't know all the characteristics of the type that can
-- affect the size (e.g. a specified small) till freeze time.
elsif Is_Fixed_Point_Type (UT)
and then not Is_Frozen (UT)
then
null;
-- Cases for which a minimum check is required
else
-- Ignore if specified size is correct for the type
if Known_Esize (UT) and then Siz = Esize (UT) then
return;
end if;
-- Otherwise get minimum size
M := UI_From_Int (Minimum_Size (UT));
if Siz < M then
-- Size is less than minimum size, but one possibility remains
-- that we can manage with the new size if we bias the type.
M := UI_From_Int (Minimum_Size (UT, Biased => True));
if Siz < M then
Error_Msg_Uint_1 := M;
Error_Msg_NE
("size for& too small, minimum allowed is ^", N, T);
Set_Esize (T, M);
Set_RM_Size (T, M);
else
Biased := True;
end if;
end if;
end if;
end Check_Size;
-------------------------
-- Get_Alignment_Value --
-------------------------
function Get_Alignment_Value (Expr : Node_Id) return Uint is
Align : constant Uint := Static_Integer (Expr);
begin
if Align = No_Uint then
return No_Uint;
elsif Align <= 0 then
Error_Msg_N ("alignment value must be positive", Expr);
return No_Uint;
else
for J in Int range 0 .. 64 loop
declare
M : constant Uint := Uint_2 ** J;
begin
exit when M = Align;
if M > Align then
Error_Msg_N
("alignment value must be power of 2", Expr);
return No_Uint;
end if;
end;
end loop;
return Align;
end if;
end Get_Alignment_Value;
----------------
-- Initialize --
----------------
procedure Initialize is
begin
Unchecked_Conversions.Init;
end Initialize;
-------------------------
-- Is_Operational_Item --
-------------------------
function Is_Operational_Item (N : Node_Id) return Boolean is
begin
if Nkind (N) /= N_Attribute_Definition_Clause then
return False;
else
declare
Id : constant Attribute_Id := Get_Attribute_Id (Chars (N));
begin
return Id = Attribute_Input
or else Id = Attribute_Output
or else Id = Attribute_Read
or else Id = Attribute_Write
or else Id = Attribute_External_Tag;
end;
end if;
end Is_Operational_Item;
------------------
-- Minimum_Size --
------------------
function Minimum_Size
(T : Entity_Id;
Biased : Boolean := False) return Nat
is
Lo : Uint := No_Uint;
Hi : Uint := No_Uint;
LoR : Ureal := No_Ureal;
HiR : Ureal := No_Ureal;
LoSet : Boolean := False;
HiSet : Boolean := False;
B : Uint;
S : Nat;
Ancest : Entity_Id;
R_Typ : constant Entity_Id := Root_Type (T);
begin
-- If bad type, return 0
if T = Any_Type then
return 0;
-- For generic types, just return zero. There cannot be any legitimate
-- need to know such a size, but this routine may be called with a
-- generic type as part of normal processing.
elsif Is_Generic_Type (R_Typ)
or else R_Typ = Any_Type
then
return 0;
-- Access types. Normally an access type cannot have a size smaller
-- than the size of System.Address. The exception is on VMS, where
-- we have short and long addresses, and it is possible for an access
-- type to have a short address size (and thus be less than the size
-- of System.Address itself). We simply skip the check for VMS, and
-- leave it to the back end to do the check.
elsif Is_Access_Type (T) then
if OpenVMS_On_Target then
return 0;
else
return System_Address_Size;
end if;
-- Floating-point types
elsif Is_Floating_Point_Type (T) then
return UI_To_Int (Esize (R_Typ));
-- Discrete types
elsif Is_Discrete_Type (T) then
-- The following loop is looking for the nearest compile time known
-- bounds following the ancestor subtype chain. The idea is to find
-- the most restrictive known bounds information.
Ancest := T;
loop
if Ancest = Any_Type or else Etype (Ancest) = Any_Type then
return 0;
end if;
if not LoSet then
if Compile_Time_Known_Value (Type_Low_Bound (Ancest)) then
Lo := Expr_Rep_Value (Type_Low_Bound (Ancest));
LoSet := True;
exit when HiSet;
end if;
end if;
if not HiSet then
if Compile_Time_Known_Value (Type_High_Bound (Ancest)) then
Hi := Expr_Rep_Value (Type_High_Bound (Ancest));
HiSet := True;
exit when LoSet;
end if;
end if;
Ancest := Ancestor_Subtype (Ancest);
if No (Ancest) then
Ancest := Base_Type (T);
if Is_Generic_Type (Ancest) then
return 0;
end if;
end if;
end loop;
-- Fixed-point types. We can't simply use Expr_Value to get the
-- Corresponding_Integer_Value values of the bounds, since these do not
-- get set till the type is frozen, and this routine can be called
-- before the type is frozen. Similarly the test for bounds being static
-- needs to include the case where we have unanalyzed real literals for
-- the same reason.
elsif Is_Fixed_Point_Type (T) then
-- The following loop is looking for the nearest compile time known
-- bounds following the ancestor subtype chain. The idea is to find
-- the most restrictive known bounds information.
Ancest := T;
loop
if Ancest = Any_Type or else Etype (Ancest) = Any_Type then
return 0;
end if;
-- Note: In the following two tests for LoSet and HiSet, it may
-- seem redundant to test for N_Real_Literal here since normally
-- one would assume that the test for the value being known at
-- compile time includes this case. However, there is a glitch.
-- If the real literal comes from folding a non-static expression,
-- then we don't consider any non- static expression to be known
-- at compile time if we are in configurable run time mode (needed
-- in some cases to give a clearer definition of what is and what
-- is not accepted). So the test is indeed needed. Without it, we
-- would set neither Lo_Set nor Hi_Set and get an infinite loop.
if not LoSet then
if Nkind (Type_Low_Bound (Ancest)) = N_Real_Literal
or else Compile_Time_Known_Value (Type_Low_Bound (Ancest))
then
LoR := Expr_Value_R (Type_Low_Bound (Ancest));
LoSet := True;
exit when HiSet;
end if;
end if;
if not HiSet then
if Nkind (Type_High_Bound (Ancest)) = N_Real_Literal
or else Compile_Time_Known_Value (Type_High_Bound (Ancest))
then
HiR := Expr_Value_R (Type_High_Bound (Ancest));
HiSet := True;
exit when LoSet;
end if;
end if;
Ancest := Ancestor_Subtype (Ancest);
if No (Ancest) then
Ancest := Base_Type (T);
if Is_Generic_Type (Ancest) then
return 0;
end if;
end if;
end loop;
Lo := UR_To_Uint (LoR / Small_Value (T));
Hi := UR_To_Uint (HiR / Small_Value (T));
-- No other types allowed
else
raise Program_Error;
end if;
-- Fall through with Hi and Lo set. Deal with biased case
if (Biased
and then not Is_Fixed_Point_Type (T)
and then not (Is_Enumeration_Type (T)
and then Has_Non_Standard_Rep (T)))
or else Has_Biased_Representation (T)
then
Hi := Hi - Lo;
Lo := Uint_0;
end if;
-- Signed case. Note that we consider types like range 1 .. -1 to be
-- signed for the purpose of computing the size, since the bounds have
-- to be accommodated in the base type.
if Lo < 0 or else Hi < 0 then
S := 1;
B := Uint_1;
-- S = size, B = 2 ** (size - 1) (can accommodate -B .. +(B - 1))
-- Note that we accommodate the case where the bounds cross. This
-- can happen either because of the way the bounds are declared
-- or because of the algorithm in Freeze_Fixed_Point_Type.
while Lo < -B
or else Hi < -B
or else Lo >= B
or else Hi >= B
loop
B := Uint_2 ** S;
S := S + 1;
end loop;
-- Unsigned case
else
-- If both bounds are positive, make sure that both are represen-
-- table in the case where the bounds are crossed. This can happen
-- either because of the way the bounds are declared, or because of
-- the algorithm in Freeze_Fixed_Point_Type.
if Lo > Hi then
Hi := Lo;
end if;
-- S = size, (can accommodate 0 .. (2**size - 1))
S := 0;
while Hi >= Uint_2 ** S loop
S := S + 1;
end loop;
end if;
return S;
end Minimum_Size;
---------------------------
-- New_Stream_Subprogram --
---------------------------
procedure New_Stream_Subprogram
(N : Node_Id;
Ent : Entity_Id;
Subp : Entity_Id;
Nam : TSS_Name_Type)
is
Loc : constant Source_Ptr := Sloc (N);
Sname : constant Name_Id := Make_TSS_Name (Base_Type (Ent), Nam);
Subp_Id : Entity_Id;
Subp_Decl : Node_Id;
F : Entity_Id;
Etyp : Entity_Id;
Defer_Declaration : constant Boolean :=
Is_Tagged_Type (Ent) or else Is_Private_Type (Ent);
-- For a tagged type, there is a declaration for each stream attribute
-- at the freeze point, and we must generate only a completion of this
-- declaration. We do the same for private types, because the full view
-- might be tagged. Otherwise we generate a declaration at the point of
-- the attribute definition clause.
function Build_Spec return Node_Id;
-- Used for declaration and renaming declaration, so that this is
-- treated as a renaming_as_body.
----------------
-- Build_Spec --
----------------
function Build_Spec return Node_Id is
Out_P : constant Boolean := (Nam = TSS_Stream_Read);
Formals : List_Id;
Spec : Node_Id;
T_Ref : constant Node_Id := New_Reference_To (Etyp, Loc);
begin
Subp_Id := Make_Defining_Identifier (Loc, Sname);
-- S : access Root_Stream_Type'Class
Formals := New_List (
Make_Parameter_Specification (Loc,
Defining_Identifier =>
Make_Defining_Identifier (Loc, Name_S),
Parameter_Type =>
Make_Access_Definition (Loc,
Subtype_Mark =>
New_Reference_To (
Designated_Type (Etype (F)), Loc))));
if Nam = TSS_Stream_Input then
Spec := Make_Function_Specification (Loc,
Defining_Unit_Name => Subp_Id,
Parameter_Specifications => Formals,
Result_Definition => T_Ref);
else
-- V : [out] T
Append_To (Formals,
Make_Parameter_Specification (Loc,
Defining_Identifier => Make_Defining_Identifier (Loc, Name_V),
Out_Present => Out_P,
Parameter_Type => T_Ref));
Spec := Make_Procedure_Specification (Loc,
Defining_Unit_Name => Subp_Id,
Parameter_Specifications => Formals);
end if;
return Spec;
end Build_Spec;
-- Start of processing for New_Stream_Subprogram
begin
F := First_Formal (Subp);
if Ekind (Subp) = E_Procedure then
Etyp := Etype (Next_Formal (F));
else
Etyp := Etype (Subp);
end if;
-- Prepare subprogram declaration and insert it as an action on the
-- clause node. The visibility for this entity is used to test for
-- visibility of the attribute definition clause (in the sense of
-- 8.3(23) as amended by AI-195).
if not Defer_Declaration then
Subp_Decl :=
Make_Subprogram_Declaration (Loc,
Specification => Build_Spec);
-- For a tagged type, there is always a visible declaration for each
-- stream TSS (it is a predefined primitive operation), and the
-- completion of this declaration occurs at the freeze point, which is
-- not always visible at places where the attribute definition clause is
-- visible. So, we create a dummy entity here for the purpose of
-- tracking the visibility of the attribute definition clause itself.
else
Subp_Id :=
Make_Defining_Identifier (Loc,
Chars => New_External_Name (Sname, 'V'));
Subp_Decl :=
Make_Object_Declaration (Loc,
Defining_Identifier => Subp_Id,
Object_Definition => New_Occurrence_Of (Standard_Boolean, Loc));
end if;
Insert_Action (N, Subp_Decl);
Set_Entity (N, Subp_Id);
Subp_Decl :=
Make_Subprogram_Renaming_Declaration (Loc,
Specification => Build_Spec,
Name => New_Reference_To (Subp, Loc));
if Defer_Declaration then
Set_TSS (Base_Type (Ent), Subp_Id);
else
Insert_Action (N, Subp_Decl);
Copy_TSS (Subp_Id, Base_Type (Ent));
end if;
end New_Stream_Subprogram;
------------------------
-- Rep_Item_Too_Early --
------------------------
function Rep_Item_Too_Early (T : Entity_Id; N : Node_Id) return Boolean is
begin
-- Cannot apply non-operational rep items to generic types
if Is_Operational_Item (N) then
return False;
elsif Is_Type (T)
and then Is_Generic_Type (Root_Type (T))
then
Error_Msg_N
("representation item not allowed for generic type", N);
return True;
end if;
-- Otherwise check for incomplete type
if Is_Incomplete_Or_Private_Type (T)
and then No (Underlying_Type (T))
then
Error_Msg_N
("representation item must be after full type declaration", N);
return True;
-- If the type has incomplete components, a representation clause is
-- illegal but stream attributes and Convention pragmas are correct.
elsif Has_Private_Component (T) then
if Nkind (N) = N_Pragma then
return False;
else
Error_Msg_N
("representation item must appear after type is fully defined",
N);
return True;
end if;
else
return False;
end if;
end Rep_Item_Too_Early;
-----------------------
-- Rep_Item_Too_Late --
-----------------------
function Rep_Item_Too_Late
(T : Entity_Id;
N : Node_Id;
FOnly : Boolean := False) return Boolean
is
S : Entity_Id;
Parent_Type : Entity_Id;
procedure Too_Late;
-- Output the too late message. Note that this is not considered a
-- serious error, since the effect is simply that we ignore the
-- representation clause in this case.
--------------
-- Too_Late --
--------------
procedure Too_Late is
begin
Error_Msg_N ("|representation item appears too late!", N);
end Too_Late;
-- Start of processing for Rep_Item_Too_Late
begin
-- First make sure entity is not frozen (RM 13.1(9)). Exclude imported
-- types, which may be frozen if they appear in a representation clause
-- for a local type.
if Is_Frozen (T)
and then not From_With_Type (T)
then
Too_Late;
S := First_Subtype (T);
if Present (Freeze_Node (S)) then
Error_Msg_NE
("?no more representation items for }", Freeze_Node (S), S);
end if;
return True;
-- Check for case of non-tagged derived type whose parent either has
-- primitive operations, or is a by reference type (RM 13.1(10)).
elsif Is_Type (T)
and then not FOnly
and then Is_Derived_Type (T)
and then not Is_Tagged_Type (T)
then
Parent_Type := Etype (Base_Type (T));
if Has_Primitive_Operations (Parent_Type) then
Too_Late;
Error_Msg_NE
("primitive operations already defined for&!", N, Parent_Type);
return True;
elsif Is_By_Reference_Type (Parent_Type) then
Too_Late;
Error_Msg_NE
("parent type & is a by reference type!", N, Parent_Type);
return True;
end if;
end if;
-- No error, link item into head of chain of rep items for the entity,
-- but avoid chaining if we have an overloadable entity, and the pragma
-- is one that can apply to multiple overloaded entities.
if Is_Overloadable (T)
and then Nkind (N) = N_Pragma
then
declare
Pname : constant Name_Id := Pragma_Name (N);
begin
if Pname = Name_Convention or else
Pname = Name_Import or else
Pname = Name_Export or else
Pname = Name_External or else
Pname = Name_Interface
then
return False;
end if;
end;
end if;
Record_Rep_Item (T, N);
return False;
end Rep_Item_Too_Late;
-------------------------
-- Same_Representation --
-------------------------
function Same_Representation (Typ1, Typ2 : Entity_Id) return Boolean is
T1 : constant Entity_Id := Underlying_Type (Typ1);
T2 : constant Entity_Id := Underlying_Type (Typ2);
begin
-- A quick check, if base types are the same, then we definitely have
-- the same representation, because the subtype specific representation
-- attributes (Size and Alignment) do not affect representation from
-- the point of view of this test.
if Base_Type (T1) = Base_Type (T2) then
return True;
elsif Is_Private_Type (Base_Type (T2))
and then Base_Type (T1) = Full_View (Base_Type (T2))
then
return True;
end if;
-- Tagged types never have differing representations
if Is_Tagged_Type (T1) then
return True;
end if;
-- Representations are definitely different if conventions differ
if Convention (T1) /= Convention (T2) then
return False;
end if;
-- Representations are different if component alignments differ
if (Is_Record_Type (T1) or else Is_Array_Type (T1))
and then
(Is_Record_Type (T2) or else Is_Array_Type (T2))
and then Component_Alignment (T1) /= Component_Alignment (T2)
then
return False;
end if;
-- For arrays, the only real issue is component size. If we know the
-- component size for both arrays, and it is the same, then that's
-- good enough to know we don't have a change of representation.
if Is_Array_Type (T1) then
if Known_Component_Size (T1)
and then Known_Component_Size (T2)
and then Component_Size (T1) = Component_Size (T2)
then
return True;
end if;
end if;
-- Types definitely have same representation if neither has non-standard
-- representation since default representations are always consistent.
-- If only one has non-standard representation, and the other does not,
-- then we consider that they do not have the same representation. They
-- might, but there is no way of telling early enough.
if Has_Non_Standard_Rep (T1) then
if not Has_Non_Standard_Rep (T2) then
return False;
end if;
else
return not Has_Non_Standard_Rep (T2);
end if;
-- Here the two types both have non-standard representation, and we need
-- to determine if they have the same non-standard representation.
-- For arrays, we simply need to test if the component sizes are the
-- same. Pragma Pack is reflected in modified component sizes, so this
-- check also deals with pragma Pack.
if Is_Array_Type (T1) then
return Component_Size (T1) = Component_Size (T2);
-- Tagged types always have the same representation, because it is not
-- possible to specify different representations for common fields.
elsif Is_Tagged_Type (T1) then
return True;
-- Case of record types
elsif Is_Record_Type (T1) then
-- Packed status must conform
if Is_Packed (T1) /= Is_Packed (T2) then
return False;
-- Otherwise we must check components. Typ2 maybe a constrained
-- subtype with fewer components, so we compare the components
-- of the base types.
else
Record_Case : declare
CD1, CD2 : Entity_Id;
function Same_Rep return Boolean;
-- CD1 and CD2 are either components or discriminants. This
-- function tests whether the two have the same representation
--------------
-- Same_Rep --
--------------
function Same_Rep return Boolean is
begin
if No (Component_Clause (CD1)) then
return No (Component_Clause (CD2));
else
return
Present (Component_Clause (CD2))
and then
Component_Bit_Offset (CD1) = Component_Bit_Offset (CD2)
and then
Esize (CD1) = Esize (CD2);
end if;
end Same_Rep;
-- Start of processing for Record_Case
begin
if Has_Discriminants (T1) then
CD1 := First_Discriminant (T1);
CD2 := First_Discriminant (T2);
-- The number of discriminants may be different if the
-- derived type has fewer (constrained by values). The
-- invisible discriminants retain the representation of
-- the original, so the discrepancy does not per se
-- indicate a different representation.
while Present (CD1)
and then Present (CD2)
loop
if not Same_Rep then
return False;
else
Next_Discriminant (CD1);
Next_Discriminant (CD2);
end if;
end loop;
end if;
CD1 := First_Component (Underlying_Type (Base_Type (T1)));
CD2 := First_Component (Underlying_Type (Base_Type (T2)));
while Present (CD1) loop
if not Same_Rep then
return False;
else
Next_Component (CD1);
Next_Component (CD2);
end if;
end loop;
return True;
end Record_Case;
end if;
-- For enumeration types, we must check each literal to see if the
-- representation is the same. Note that we do not permit enumeration
-- representation clauses for Character and Wide_Character, so these
-- cases were already dealt with.
elsif Is_Enumeration_Type (T1) then
Enumeration_Case : declare
L1, L2 : Entity_Id;
begin
L1 := First_Literal (T1);
L2 := First_Literal (T2);
while Present (L1) loop
if Enumeration_Rep (L1) /= Enumeration_Rep (L2) then
return False;
else
Next_Literal (L1);
Next_Literal (L2);
end if;
end loop;
return True;
end Enumeration_Case;
-- Any other types have the same representation for these purposes
else
return True;
end if;
end Same_Representation;
--------------------
-- Set_Enum_Esize --
--------------------
procedure Set_Enum_Esize (T : Entity_Id) is
Lo : Uint;
Hi : Uint;
Sz : Nat;
begin
Init_Alignment (T);
-- Find the minimum standard size (8,16,32,64) that fits
Lo := Enumeration_Rep (Entity (Type_Low_Bound (T)));
Hi := Enumeration_Rep (Entity (Type_High_Bound (T)));
if Lo < 0 then
if Lo >= -Uint_2**07 and then Hi < Uint_2**07 then
Sz := Standard_Character_Size; -- May be > 8 on some targets
elsif Lo >= -Uint_2**15 and then Hi < Uint_2**15 then
Sz := 16;
elsif Lo >= -Uint_2**31 and then Hi < Uint_2**31 then
Sz := 32;
else pragma Assert (Lo >= -Uint_2**63 and then Hi < Uint_2**63);
Sz := 64;
end if;
else
if Hi < Uint_2**08 then
Sz := Standard_Character_Size; -- May be > 8 on some targets
elsif Hi < Uint_2**16 then
Sz := 16;
elsif Hi < Uint_2**32 then
Sz := 32;
else pragma Assert (Hi < Uint_2**63);
Sz := 64;
end if;
end if;
-- That minimum is the proper size unless we have a foreign convention
-- and the size required is 32 or less, in which case we bump the size
-- up to 32. This is required for C and C++ and seems reasonable for
-- all other foreign conventions.
if Has_Foreign_Convention (T)
and then Esize (T) < Standard_Integer_Size
then
Init_Esize (T, Standard_Integer_Size);
else
Init_Esize (T, Sz);
end if;
end Set_Enum_Esize;
------------------------------
-- Validate_Address_Clauses --
------------------------------
procedure Validate_Address_Clauses is
begin
for J in Address_Clause_Checks.First .. Address_Clause_Checks.Last loop
declare
ACCR : Address_Clause_Check_Record
renames Address_Clause_Checks.Table (J);
Expr : Node_Id;
X_Alignment : Uint;
Y_Alignment : Uint;
X_Size : Uint;
Y_Size : Uint;
begin
-- Skip processing of this entry if warning already posted
if not Address_Warning_Posted (ACCR.N) then
Expr := Original_Node (Expression (ACCR.N));
-- Get alignments
X_Alignment := Alignment (ACCR.X);
Y_Alignment := Alignment (ACCR.Y);
-- Similarly obtain sizes
X_Size := Esize (ACCR.X);
Y_Size := Esize (ACCR.Y);
-- Check for large object overlaying smaller one
if Y_Size > Uint_0
and then X_Size > Uint_0
and then X_Size > Y_Size
then
Error_Msg_NE
("?& overlays smaller object", ACCR.N, ACCR.X);
Error_Msg_N
("\?program execution may be erroneous", ACCR.N);
Error_Msg_Uint_1 := X_Size;
Error_Msg_NE
("\?size of & is ^", ACCR.N, ACCR.X);
Error_Msg_Uint_1 := Y_Size;
Error_Msg_NE
("\?size of & is ^", ACCR.N, ACCR.Y);
-- Check for inadequate alignment, both of the base object
-- and of the offset, if any.
-- Note: we do not check the alignment if we gave a size
-- warning, since it would likely be redundant.
elsif Y_Alignment /= Uint_0
and then (Y_Alignment < X_Alignment
or else (ACCR.Off
and then
Nkind (Expr) = N_Attribute_Reference
and then
Attribute_Name (Expr) = Name_Address
and then
Has_Compatible_Alignment
(ACCR.X, Prefix (Expr))
/= Known_Compatible))
then
Error_Msg_NE
("?specified address for& may be inconsistent "
& "with alignment",
ACCR.N, ACCR.X);
Error_Msg_N
("\?program execution may be erroneous (RM 13.3(27))",
ACCR.N);
Error_Msg_Uint_1 := X_Alignment;
Error_Msg_NE
("\?alignment of & is ^",
ACCR.N, ACCR.X);
Error_Msg_Uint_1 := Y_Alignment;
Error_Msg_NE
("\?alignment of & is ^",
ACCR.N, ACCR.Y);
if Y_Alignment >= X_Alignment then
Error_Msg_N
("\?but offset is not multiple of alignment",
ACCR.N);
end if;
end if;
end if;
end;
end loop;
end Validate_Address_Clauses;
-----------------------------------
-- Validate_Unchecked_Conversion --
-----------------------------------
procedure Validate_Unchecked_Conversion
(N : Node_Id;
Act_Unit : Entity_Id)
is
Source : Entity_Id;
Target : Entity_Id;
Vnode : Node_Id;
begin
-- Obtain source and target types. Note that we call Ancestor_Subtype
-- here because the processing for generic instantiation always makes
-- subtypes, and we want the original frozen actual types.
-- If we are dealing with private types, then do the check on their
-- fully declared counterparts if the full declarations have been
-- encountered (they don't have to be visible, but they must exist!)
Source := Ancestor_Subtype (Etype (First_Formal (Act_Unit)));
if Is_Private_Type (Source)
and then Present (Underlying_Type (Source))
then
Source := Underlying_Type (Source);
end if;
Target := Ancestor_Subtype (Etype (Act_Unit));
-- If either type is generic, the instantiation happens within a generic
-- unit, and there is nothing to check. The proper check
-- will happen when the enclosing generic is instantiated.
if Is_Generic_Type (Source) or else Is_Generic_Type (Target) then
return;
end if;
if Is_Private_Type (Target)
and then Present (Underlying_Type (Target))
then
Target := Underlying_Type (Target);
end if;
-- Source may be unconstrained array, but not target
if Is_Array_Type (Target)
and then not Is_Constrained (Target)
then
Error_Msg_N
("unchecked conversion to unconstrained array not allowed", N);
return;
end if;
-- Warn if conversion between two different convention pointers
if Is_Access_Type (Target)
and then Is_Access_Type (Source)
and then Convention (Target) /= Convention (Source)
and then Warn_On_Unchecked_Conversion
then
-- Give warnings for subprogram pointers only on most targets. The
-- exception is VMS, where data pointers can have different lengths
-- depending on the pointer convention.
if Is_Access_Subprogram_Type (Target)
or else Is_Access_Subprogram_Type (Source)
or else OpenVMS_On_Target
then
Error_Msg_N
("?conversion between pointers with different conventions!", N);
end if;
end if;
-- Warn if one of the operands is Ada.Calendar.Time. Do not emit a
-- warning when compiling GNAT-related sources.
if Warn_On_Unchecked_Conversion
and then not In_Predefined_Unit (N)
and then RTU_Loaded (Ada_Calendar)
and then
(Chars (Source) = Name_Time
or else
Chars (Target) = Name_Time)
then
-- If Ada.Calendar is loaded and the name of one of the operands is
-- Time, there is a good chance that this is Ada.Calendar.Time.
declare
Calendar_Time : constant Entity_Id :=
Full_View (RTE (RO_CA_Time));
begin
pragma Assert (Present (Calendar_Time));
if Source = Calendar_Time
or else Target = Calendar_Time
then
Error_Msg_N
("?representation of 'Time values may change between " &
"'G'N'A'T versions", N);
end if;
end;
end if;
-- Make entry in unchecked conversion table for later processing by
-- Validate_Unchecked_Conversions, which will check sizes and alignments
-- (using values set by the back-end where possible). This is only done
-- if the appropriate warning is active.
if Warn_On_Unchecked_Conversion then
Unchecked_Conversions.Append
(New_Val => UC_Entry'
(Eloc => Sloc (N),
Source => Source,
Target => Target));
-- If both sizes are known statically now, then back end annotation
-- is not required to do a proper check but if either size is not
-- known statically, then we need the annotation.
if Known_Static_RM_Size (Source)
and then Known_Static_RM_Size (Target)
then
null;
else
Back_Annotate_Rep_Info := True;
end if;
end if;
-- If unchecked conversion to access type, and access type is declared
-- in the same unit as the unchecked conversion, then set the
-- No_Strict_Aliasing flag (no strict aliasing is implicit in this
-- situation).
if Is_Access_Type (Target) and then
In_Same_Source_Unit (Target, N)
then
Set_No_Strict_Aliasing (Implementation_Base_Type (Target));
end if;
-- Generate N_Validate_Unchecked_Conversion node for back end in
-- case the back end needs to perform special validation checks.
-- Shouldn't this be in Exp_Ch13, since the check only gets done
-- if we have full expansion and the back end is called ???
Vnode :=
Make_Validate_Unchecked_Conversion (Sloc (N));
Set_Source_Type (Vnode, Source);
Set_Target_Type (Vnode, Target);
-- If the unchecked conversion node is in a list, just insert before it.
-- If not we have some strange case, not worth bothering about.
if Is_List_Member (N) then
Insert_After (N, Vnode);
end if;
end Validate_Unchecked_Conversion;
------------------------------------
-- Validate_Unchecked_Conversions --
------------------------------------
procedure Validate_Unchecked_Conversions is
begin
for N in Unchecked_Conversions.First .. Unchecked_Conversions.Last loop
declare
T : UC_Entry renames Unchecked_Conversions.Table (N);
Eloc : constant Source_Ptr := T.Eloc;
Source : constant Entity_Id := T.Source;
Target : constant Entity_Id := T.Target;
Source_Siz : Uint;
Target_Siz : Uint;
begin
-- This validation check, which warns if we have unequal sizes for
-- unchecked conversion, and thus potentially implementation
-- dependent semantics, is one of the few occasions on which we
-- use the official RM size instead of Esize. See description in
-- Einfo "Handling of Type'Size Values" for details.
if Serious_Errors_Detected = 0
and then Known_Static_RM_Size (Source)
and then Known_Static_RM_Size (Target)
-- Don't do the check if warnings off for either type, note the
-- deliberate use of OR here instead of OR ELSE to get the flag
-- Warnings_Off_Used set for both types if appropriate.
and then not (Has_Warnings_Off (Source)
or
Has_Warnings_Off (Target))
then
Source_Siz := RM_Size (Source);
Target_Siz := RM_Size (Target);
if Source_Siz /= Target_Siz then
Error_Msg
("?types for unchecked conversion have different sizes!",
Eloc);
if All_Errors_Mode then
Error_Msg_Name_1 := Chars (Source);
Error_Msg_Uint_1 := Source_Siz;
Error_Msg_Name_2 := Chars (Target);
Error_Msg_Uint_2 := Target_Siz;
Error_Msg ("\size of % is ^, size of % is ^?", Eloc);
Error_Msg_Uint_1 := UI_Abs (Source_Siz - Target_Siz);
if Is_Discrete_Type (Source)
and then Is_Discrete_Type (Target)
then
if Source_Siz > Target_Siz then
Error_Msg
("\?^ high order bits of source will be ignored!",
Eloc);
elsif Is_Unsigned_Type (Source) then
Error_Msg
("\?source will be extended with ^ high order " &
"zero bits?!", Eloc);
else
Error_Msg
("\?source will be extended with ^ high order " &
"sign bits!",
Eloc);
end if;
elsif Source_Siz < Target_Siz then
if Is_Discrete_Type (Target) then
if Bytes_Big_Endian then
Error_Msg
("\?target value will include ^ undefined " &
"low order bits!",
Eloc);
else
Error_Msg
("\?target value will include ^ undefined " &
"high order bits!",
Eloc);
end if;
else
Error_Msg
("\?^ trailing bits of target value will be " &
"undefined!", Eloc);
end if;
else pragma Assert (Source_Siz > Target_Siz);
Error_Msg
("\?^ trailing bits of source will be ignored!",
Eloc);
end if;
end if;
end if;
end if;
-- If both types are access types, we need to check the alignment.
-- If the alignment of both is specified, we can do it here.
if Serious_Errors_Detected = 0
and then Ekind (Source) in Access_Kind
and then Ekind (Target) in Access_Kind
and then Target_Strict_Alignment
and then Present (Designated_Type (Source))
and then Present (Designated_Type (Target))
then
declare
D_Source : constant Entity_Id := Designated_Type (Source);
D_Target : constant Entity_Id := Designated_Type (Target);
begin
if Known_Alignment (D_Source)
and then Known_Alignment (D_Target)
then
declare
Source_Align : constant Uint := Alignment (D_Source);
Target_Align : constant Uint := Alignment (D_Target);
begin
if Source_Align < Target_Align
and then not Is_Tagged_Type (D_Source)
-- Suppress warning if warnings suppressed on either
-- type or either designated type. Note the use of
-- OR here instead of OR ELSE. That is intentional,
-- we would like to set flag Warnings_Off_Used in
-- all types for which warnings are suppressed.
and then not (Has_Warnings_Off (D_Source)
or
Has_Warnings_Off (D_Target)
or
Has_Warnings_Off (Source)
or
Has_Warnings_Off (Target))
then
Error_Msg_Uint_1 := Target_Align;
Error_Msg_Uint_2 := Source_Align;
Error_Msg_Node_1 := D_Target;
Error_Msg_Node_2 := D_Source;
Error_Msg
("?alignment of & (^) is stricter than " &
"alignment of & (^)!", Eloc);
Error_Msg
("\?resulting access value may have invalid " &
"alignment!", Eloc);
end if;
end;
end if;
end;
end if;
end;
end loop;
end Validate_Unchecked_Conversions;
end Sem_Ch13;