URL
https://opencores.org/ocsvn/openrisc/openrisc/trunk
Subversion Repositories openrisc
[/] [openrisc/] [trunk/] [gnu-dev/] [or1k-gcc/] [gcc/] [ada/] [sem_ch13.ads] - Rev 707
Go to most recent revision | Compare with Previous | Blame | View Log
------------------------------------------------------------------------------ -- -- -- GNAT COMPILER COMPONENTS -- -- -- -- S E M _ C H 1 3 -- -- -- -- S p e c -- -- -- -- Copyright (C) 1992-2010, 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 Table; with Types; use Types; with Uintp; use Uintp; package Sem_Ch13 is procedure Analyze_At_Clause (N : Node_Id); procedure Analyze_Attribute_Definition_Clause (N : Node_Id); procedure Analyze_Enumeration_Representation_Clause (N : Node_Id); procedure Analyze_Free_Statement (N : Node_Id); procedure Analyze_Freeze_Entity (N : Node_Id); procedure Analyze_Record_Representation_Clause (N : Node_Id); procedure Analyze_Code_Statement (N : Node_Id); procedure Analyze_Aspect_Specifications (N : Node_Id; E : Entity_Id); -- This procedure is called to analyze aspect specifications for node N. E -- is the corresponding entity declared by the declaration node N. Callers -- should check that Has_Aspects (N) is True before calling this routine. procedure Adjust_Record_For_Reverse_Bit_Order (R : Entity_Id); -- Called from Freeze where R is a record entity for which reverse bit -- order is specified and there is at least one component clause. Adjusts -- component positions according to either Ada 95 or Ada 2005 (AI-133). procedure Build_Invariant_Procedure (Typ : Entity_Id; N : Node_Id); -- Typ is a private type with invariants (indicated by Has_Invariants being -- set for Typ, indicating the presence of pragma Invariant entries on the -- rep chain, note that Invariant aspects have already been converted to -- pragma Invariant), then this procedure builds the spec and body for the -- corresponding Invariant procedure, inserting them at appropriate points -- in the package specification N. Invariant_Procedure is set for Typ. Note -- that this procedure is called at the end of processing the declarations -- in the visible part (i.e. the right point for visibility analysis of -- the invariant expression). procedure Check_Record_Representation_Clause (N : Node_Id); -- This procedure completes the analysis of a record representation clause -- N. It is called at freeze time after adjustment of component clause bit -- positions for possible non-standard bit order. In the case of Ada 2005 -- (machine scalar) mode, this adjustment can make substantial changes, so -- some checks, in particular for component overlaps cannot be done at the -- time the record representation clause is first seen, but must be delayed -- till freeze time, and in particular is called after calling the above -- procedure for adjusting record bit positions for reverse bit order. procedure Initialize; -- Initialize internal tables for new compilation procedure Set_Enum_Esize (T : Entity_Id); -- This routine sets the Esize field for an enumeration type T, based -- on the current representation information available for T. Note that -- the setting of the RM_Size field is not affected. This routine also -- initializes the alignment field to zero. function Minimum_Size (T : Entity_Id; Biased : Boolean := False) return Nat; -- Given an elementary type, determines the minimum number of bits required -- to represent all values of the type. This function may not be called -- with any other types. If the flag Biased is set True, then the minimum -- size calculation that biased representation is used in the case of a -- discrete type, e.g. the range 7..8 gives a minimum size of 4 with -- Biased set to False, and 1 with Biased set to True. Note that the -- biased parameter only has an effect if the type is not biased, it -- causes Minimum_Size to indicate the minimum size of an object with -- the given type, of the size the type would have if it were biased. If -- the type is already biased, then Minimum_Size returns the biased size, -- regardless of the setting of Biased. Also, fixed-point types are never -- biased in the current implementation. If the size is not known at -- compile time, this function returns 0. procedure Check_Constant_Address_Clause (Expr : Node_Id; U_Ent : Entity_Id); -- Expr is an expression for an address clause. This procedure checks -- that the expression is constant, in the limited sense that it is safe -- to evaluate it at the point the object U_Ent is declared, rather than -- at the point of the address clause. The condition for this to be true -- is that the expression has no variables, no constants declared after -- U_Ent, and no calls to non-pure functions. If this condition is not -- met, then an appropriate error message is posted. This check is applied -- at the point an object with an address clause is frozen, as well as for -- address clauses for tasks and entries. procedure Check_Size (N : Node_Id; T : Entity_Id; Siz : Uint; Biased : out Boolean); -- Called when size Siz is specified for subtype T. This subprogram checks -- that the size is appropriate, posting errors on node N as required. -- This check is effective for elementary types and bit-packed arrays. -- For other non-elementary types, a check is only made if an explicit -- size has been given for the type (and the specified size must match). -- The parameter Biased is set False if the size specified did not require -- the use of biased representation, and True if biased representation -- was required to meet the size requirement. Note that Biased is only -- set if the type is not currently biased, but biasing it is the only -- way to meet the requirement. If the type is currently biased, then -- this biased size is used in the initial check, and Biased is False. -- If the size is too small, and an error message is given, then both -- Esize and RM_Size are reset to the allowed minimum value in T. function Rep_Item_Too_Early (T : Entity_Id; N : Node_Id) return Boolean; -- Called at the start of processing a representation clause or a -- representation pragma. Used to check that the representation item -- is not being applied to an incomplete type or to a generic formal -- type or a type derived from a generic formal type. Returns False if -- no such error occurs. If this error does occur, appropriate error -- messages are posted on node N, and True is returned. function Rep_Item_Too_Late (T : Entity_Id; N : Node_Id; FOnly : Boolean := False) return Boolean; -- Called at the start of processing a representation clause or a -- representation pragma. Used to check that a representation item -- for entity T does not appear too late (according to the rules in -- RM 13.1(9) and RM 13.1(10)). N is the associated node, which in -- the pragma case is the pragma or representation clause itself, used -- for placing error messages if the item is too late. -- -- Fonly is a flag that causes only the freezing rule (para 9) to be -- applied, and the tests of para 10 are skipped. This is appropriate -- for both subtype related attributes (Alignment and Size) and for -- stream attributes, which, although certainly not subtype related -- attributes, clearly should not be subject to the para 10 restrictions -- (see AI95-00137). Similarly, we also skip the para 10 restrictions for -- the Storage_Size case where they also clearly do not apply, and for -- Stream_Convert which is in the same category as the stream attributes. -- -- If the rep item is too late, an appropriate message is output and -- True is returned, which is a signal that the caller should abandon -- processing for the item. If the item is not too late, then False -- is returned, and the caller can continue processing the item. -- -- If no error is detected, this call also as a side effect links the -- representation item onto the head of the representation item chain -- (referenced by the First_Rep_Item field of the entity). -- -- Note: Rep_Item_Too_Late must be called with the underlying type in -- the case of a private or incomplete type. The protocol is to first -- check for Rep_Item_Too_Early using the initial entity, then take the -- underlying type, then call Rep_Item_Too_Late on the result. -- -- Note: Calls to Rep_Item_Too_Late are ignored for the case of attribute -- definition clauses which have From_Aspect_Specification set. This is -- because such clauses are linked on to the Rep_Item chain in procedure -- Sem_Ch13.Analyze_Aspect_Specifications. See that procedure for details. function Same_Representation (Typ1, Typ2 : Entity_Id) return Boolean; -- Given two types, where the two types are related by possible derivation, -- determines if the two types have the same representation, or different -- representations, requiring the special processing for representation -- change. A False result is possible only for array, enumeration or -- record types. procedure Validate_Unchecked_Conversion (N : Node_Id; Act_Unit : Entity_Id); -- Validate a call to unchecked conversion. N is the node for the actual -- instantiation, which is used only for error messages. Act_Unit is the -- entity for the instantiation, from which the actual types etc. for this -- instantiation can be determined. This procedure makes an entry in a -- table and/or generates an N_Validate_Unchecked_Conversion node. The -- actual checking is done in Validate_Unchecked_Conversions or in the -- back end as required. procedure Validate_Unchecked_Conversions; -- This routine is called after calling the backend to validate unchecked -- conversions for size and alignment appropriateness. The reason it is -- called that late is to take advantage of any back-annotation of size -- and alignment performed by the backend. procedure Validate_Address_Clauses; -- This is called after the back end has been called (and thus after the -- alignments of objects have been back annotated). It goes through the -- table of saved address clauses checking for suspicious alignments and -- if necessary issuing warnings. procedure Validate_Independence; -- This is called after the back end has been called (and thus after the -- layout of components has been back annotated). It goes through the -- table of saved pragma Independent[_Component] entries, checking that -- independence can be achieved, and if necessary issuing error messages. ------------------------------------- -- Table for Validate_Independence -- ------------------------------------- -- If a legal pragma Independent or Independent_Components is given for -- an entity, then an entry is made in this table, to be checked by a -- call to Validate_Independence after back annotation of layout is done. type Independence_Check_Record is record N : Node_Id; -- The pragma Independent or Independent_Components E : Entity_Id; -- The entity to which it applies end record; package Independence_Checks is new Table.Table ( Table_Component_Type => Independence_Check_Record, Table_Index_Type => Int, Table_Low_Bound => 1, Table_Initial => 20, Table_Increment => 200, Table_Name => "Independence_Checks"); ----------------------------------- -- Handling of Aspect Visibility -- ----------------------------------- -- The visibility of aspects is tricky. First, the visibility is delayed -- to the freeze point. This is not too complicated, what we do is simply -- to leave the aspect "laying in wait" for the freeze point, and at that -- point materialize and analye the corresponding attribute definition -- clause or pragma. There is some special processing for preconditions -- and postonditions, where the pragmas themselves deal with the required -- delay, but basically the approach is the same, delay analysis of the -- expression to the freeze point. -- Much harder is the requirement for diagnosing cases in which an early -- freeze causes a change in visibility. Consider: -- package AspectVis is -- R_Size : constant Integer := 32; -- -- package Inner is -- type R is new Integer with -- Size => R_Size; -- F : R; -- freezes -- R_Size : constant Integer := 64; -- S : constant Integer := R'Size; -- 32 not 64 -- end Inner; -- end AspectVis; -- Here the 32 not 64 shows what would be expected if this program were -- legal, since the evaluation of R_Size has to be done at the freeze -- point and gets the outer definition not the inner one. -- But the language rule requires this program to be diagnosed as illegal -- because the visibility changes between the freeze point and the end of -- the declarative region. -- To meet this requirement, we first note that the Expression field of the -- N_Aspect_Specification node holds the raw unanalyzed expression, which -- will get used in processing the aspect. At the time of analyzing the -- N_Aspect_Specification node, we create a complete copy of the expression -- and store it in the entity field of the Identifier (an odd usage, but -- the identifier is not used except to identify the aspect, so its Entity -- field is otherwise unused, and we are short of room in the node). -- This copy stays unanalyzed up to the freeze point, where we analyze the -- resulting pragma or attribute definition clause, except that in the -- case of invariants and predicates, we mark occurrences of the subtype -- name as having the entity of the subprogram parameter, so that they -- will not cause trouble in the following steps. -- Then at the freeze point, we create another copy of this unanalyzed -- expression. By this time we no longer need the Expression field for -- other purposes, so we can store it there. Now we have two copies of -- the original unanalyzed expression. One of them gets preanalyzed at -- the freeze point to capture the visibility at the freeze point. -- Now when we hit the freeze all at the end of the declarative part, if -- we come across a frozen entity with delayed aspects, we still have one -- copy of the unanalyzed expression available in the node, and we again -- do a preanalysis using that copy and the visibility at the end of the -- declarative part. Now we have two preanalyzed expression (preanalysis -- is good enough, since we are only interested in referenced entities). -- One captures the visibility at the freeze point, the other captures the -- visibility at the end of the declarative part. We see if the entities -- in these two expressions are the same, by seeing if the two expressions -- are fully conformant, and if not, issue appropriate error messages. -- Quite an awkward procedure, but this is an awkard requirement! procedure Check_Aspect_At_Freeze_Point (ASN : Node_Id); -- Performs the processing described above at the freeze point, ASN is the -- N_Aspect_Specification node for the aspect. procedure Check_Aspect_At_End_Of_Declarations (ASN : Node_Id); -- Performs the processing described above at the freeze all point, and -- issues appropriate error messages if the visibility has indeed changed. -- Again, ASN is the N_Aspect_Specification node for the aspect. end Sem_Ch13;
Go to most recent revision | Compare with Previous | Blame | View Log