[multiple changes]

2004-05-03  Olivier Hainque  <hainque@act-europe.fr>

	PR ada/15152

	* exp_ch2.adb (Expand_Current_Value): Leave Machine_Code Asm arguments
	alone. Replacing object references by literals is inappropriate in a
	so low level context.

2004-05-03  Arnaud Charlet  <charlet@act-europe.fr>

	* a-exexpr.adb: Add comments

2004-05-03  Joel Brobecker  <brobecker@gnat.com>

	* a-tags.adb (Tag_Table): Add Index_Check pragma Suppress. Allows us to
	declare the Ancestor_Tags array in Type_Specific_Data with a small size
	without risking a bounds check error when accessing one of its
	components.
	(Type_Specific_Data): Define Ancestor_Tags as a small array.
	This prevents us from hitting a limitation during the debug info
	generation when using stabs.

	* a-tags.adb (Dispatch_Table): Define the Prims_Ptr component as a
	small array.
	This prevents us from hitting a limitation during the debug info
	generation when using stabs.

2004-05-03  Eric Botcazou  <ebotcazou@act-europe.fr>

	lang-specs.h: Remove -gnatz* from specs.

2004-05-03  Vincent Celier  <celier@gnat.com>

	* gprmake.adb, makegpr.ads, makegpr.adb: New files.

	* Make-lang.in, Makefile.in: Add gprmake

2004-05-03  Thomas Quinot  <quinot@act-europe.fr>

	* sem_aggr.adb: Fix typo in comment.

2004-05-03  Robert Dewar  <dewar@gnat.com>

	* make.adb: Minor reformatting

	* rtsfind.ads, rtsfind.adb: (RTU_Loaded): New function

	* sem_attr.adb (Eval_Attribute, case Type_Class): Fix check for address
	so that it works when address is not a private type.

	* sem_ch13.adb (Check_Expr_Constants, case N_Integer_Literal): Deal
	properly with rewritten unchecked conversions. This prevents
	order-of-elaboration issues that can otherwise arise.
	(Minimum_Size): Don't check size of access types under VMS

	* sem_ch4.adb (Remove_Address_Interpretation): New circuit to remove
	interpretations of integer literals as type System.Address.

	* sem_util.ads, sem_util.adb (Is_Descendent_Of_Address): New function
	(Is_Descendent_Of): New function

2004-05-03  Jose Ruiz  <ruiz@act-europe.fr>

	* sem_prag.adb: Boolean_Entry_Barriers is a synonym of Simple_Barriers.
	Max_Entry_Queue_Depth is a synonym of Max_Entry_Queue_Length.
	No_Dynamic_Interrupts is a synonym of No_Dynamic_Attachment.

	* sem_res.adb: Use the new restriction Max_Entry_Queue_Length instead
	of the old Max_Entry_Queue_Depth.

	* snames.adb: Boolean_Entry_Barriers is a synonym of Simple_Barriers.
	Max_Entry_Queue_Depth is a synonym of Max_Entry_Queue_Length
	No_Dynamic_Interrupts is a synonym of No_Dynamic_Attachment

	* snames.ads: New entry for proper handling of Boolean_Entry_Barriers.
	New entry for proper handling of Max_Entry_Queue_Depth.
	New entry for proper handling of No_Dynamic_Interrupts.

	* s-rident.ads: Adding restriction Simple_Barriers (AI-00249) that
	supersedes the GNAT specific restriction Boolean_Entry_Barriers.
	Adding restriction Max_Entry_Queue_Length (AI-00249) that supersedes
	the GNAT specific restriction Max_Entry_Queue_Depth.
	Adding restriction No_Dynamic_Attachment (AI-00249) that supersedes
	the GNAT specific restriction No_Dynamic_Interrupts.

	* restrict.ads, restrict.adb: Use the new restriction Simple_Barriers
	instead of the old Boolean_Entry_Barriers.
	Use the new restriction No_Dynamic_Attachment instead of the old
	No_Dynamic_Interrupts.

	* exp_ch9.adb: Check restriction Simple_Barriers (AI-00249) that
	supersedes the GNAT specific restriction Boolean_Entry_Barriers.

	* gnatbind.adb: Use the new restriction Max_Entry_Queue_Length instead
	of the old Max_Entry_Queue_Depth.

2004-05-03  GNAT Script  <nobody@gnat.com>

	* Make-lang.in: Makefile automatically updated

From-SVN: r81429

1 files changed, 25 insertions, 4 deletions

diff --git a/gcc/ada/sem_ch13.adb b/gcc/ada/sem_ch13.adb
index e89041a..1da9566 100644
--- a/gcc/ada/sem_ch13.adb
+++ b/gcc/ada/sem_ch13.adb
@@ -45,6 +45,7 @@ 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;
@@ -2699,8 +2700,19 @@ package body Sem_Ch13 is
                   end if;
                end if;
 
-            when N_Integer_Literal   |
-                 N_Real_Literal      |
+            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;
@@ -3068,10 +3080,19 @@ package body Sem_Ch13 is
       then
          return 0;
 
-      --  Access types
+         --  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 the back end to do the check.
 
       elsif Is_Access_Type (T) then
-         return System_Address_Size;
+         if OpenVMS_On_Target then
+            return 0;
+         else
+            return System_Address_Size;
+         end if;
 
       --  Floating-point types