[Ada] Do not use exponentiation for common bases in floating-point Value

gcc/ada/

	* Makefile.rtl (GNATRTL_NONTASKING_OBJS): Likewise.
	* exp_imgv.adb (Expand_Value_Attribute): Use RE_Value_Long_Float in
	lieu of RE_Value_Long_Long_Float as fallback for fixed-point types.
	Also use it for Long_Long_Float if it has same size as Long_Float.
	* libgnat/s-imgrea.adb: Replace Powten_Table with Powen_LLF.
	* libgnat/s-powflt.ads: New file.
	* libgnat/s-powlfl.ads: Likewise.
	* libgnat/s-powtab.ads: Rename to...
	* libgnat/s-powllf.ads: ...this.
	* libgnat/s-valflt.ads: Add with clause for System.Powten_Flt and
	pass its table as actual parameter to System.Val_Real.
	* libgnat/s-vallfl.ads: Likewise for System.Powten_LFlt.
	* libgnat/s-valllf.ads: Likewise for System.Powten_LLF.
	* libgnat/s-valrea.ads: Add Maxpow and Powten_Address parameters.
	* libgnat/s-valrea.adb: Add pragma Warnings (Off).
	(Need_Extra): New boolean constant.
	(Precision_Limit): Set it according to Need_Extra.
	(Impl): Adjust actual parameter.
	(Integer_to_Rea): Add assertion on the machine radix. Take into
	account the extra digit only if Need_Extra is true.  Reimplement
	the computation of the final value for bases 2, 4, 8, 10 and 16.
	* libgnat/s-valued.adb (Impl): Adjust actual parameter.
	(Scan_Decimal): Add pragma Unreferenced.
	(Value_Decimal): Likewise.
	* libgnat/s-valuef.adb (Impl): Adjust actual parameter.
	* libgnat/s-valuer.ads (Floating): Remove.
	(Round): New formal parameter.
	* libgnat/s-valuer.adb (Round_Extra): New procedure.
	(Scan_Decimal_Digits): Use it to round the extra digit if Round
	is set to True in the instantiation.
	(Scan_Integral_Digits): Likewise.

13 files changed, 637 insertions, 68 deletions

diff --git a/gcc/ada/libgnat/s-imgrea.adb b/gcc/ada/libgnat/s-imgrea.adb
index 03d30bd..2ec6a1a 100644
--- a/gcc/ada/libgnat/s-imgrea.adb
+++ b/gcc/ada/libgnat/s-imgrea.adb
@@ -29,9 +29,9 @@
 --                                                                          --
 ------------------------------------------------------------------------------
 
-with System.Img_LLU;      use System.Img_LLU;
-with System.Img_Uns;      use System.Img_Uns;
-with System.Powten_Table; use System.Powten_Table;
+with System.Img_LLU;    use System.Img_LLU;
+with System.Img_Uns;    use System.Img_Uns;
+with System.Powten_LLF; use System.Powten_LLF;
 with System.Float_Control;
 
 package body System.Img_Real is
diff --git a/gcc/ada/libgnat/s-powflt.ads b/gcc/ada/libgnat/s-powflt.ads
new file mode 100644
index 0000000..9d58967
--- /dev/null
+++ b/gcc/ada/libgnat/s-powflt.ads
@@ -0,0 +1,85 @@
+------------------------------------------------------------------------------
+--                                                                          --
+--                         GNAT COMPILER COMPONENTS                         --
+--                                                                          --
+--                    S Y S T E M . P O W T E N _ F L T                     --
+--                                                                          --
+--                                 S p e c                                  --
+--                                                                          --
+--             Copyright (C) 2020, 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.                                     --
+--                                                                          --
+-- As a special exception under Section 7 of GPL version 3, you are granted --
+-- additional permissions described in the GCC Runtime Library Exception,   --
+-- version 3.1, as published by the Free Software Foundation.               --
+--                                                                          --
+-- You should have received a copy of the GNU General Public License and    --
+-- a copy of the GCC Runtime Library Exception along with this program;     --
+-- see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see    --
+-- <http://www.gnu.org/licenses/>.                                          --
+--                                                                          --
+-- GNAT was originally developed  by the GNAT team at  New York University. --
+-- Extensive contributions were provided by Ada Core Technologies Inc.      --
+--                                                                          --
+------------------------------------------------------------------------------
+
+--  This package provides a powers of ten table used for real conversions
+
+package System.Powten_Flt is
+   pragma Pure;
+
+   Maxpow : constant := 38;
+   --  Largest power of ten representable with Float
+
+   Maxpow_Exact : constant := 10;
+   --  Largest power of ten exactly representable with Float. It is equal to
+   --  floor (M * log 2 / log 5), when M is the size of the mantissa (24).
+
+   Powten : constant array (0 .. Maxpow) of Float :=
+      (00 => 1.0E+00,
+       01 => 1.0E+01,
+       02 => 1.0E+02,
+       03 => 1.0E+03,
+       04 => 1.0E+04,
+       05 => 1.0E+05,
+       06 => 1.0E+06,
+       07 => 1.0E+07,
+       08 => 1.0E+08,
+       09 => 1.0E+09,
+       10 => 1.0E+10,
+       11 => 1.0E+11,
+       12 => 1.0E+12,
+       13 => 1.0E+13,
+       14 => 1.0E+14,
+       15 => 1.0E+15,
+       16 => 1.0E+16,
+       17 => 1.0E+17,
+       18 => 1.0E+18,
+       19 => 1.0E+19,
+       20 => 1.0E+20,
+       21 => 1.0E+21,
+       22 => 1.0E+22,
+       23 => 1.0E+23,
+       24 => 1.0E+24,
+       25 => 1.0E+25,
+       26 => 1.0E+26,
+       27 => 1.0E+27,
+       28 => 1.0E+28,
+       29 => 1.0E+29,
+       30 => 1.0E+30,
+       31 => 1.0E+31,
+       32 => 1.0E+32,
+       33 => 1.0E+33,
+       34 => 1.0E+34,
+       35 => 1.0E+35,
+       36 => 1.0E+36,
+       37 => 1.0E+37,
+       38 => 1.0E+38);
+
+end System.Powten_Flt;
diff --git a/gcc/ada/libgnat/s-powlfl.ads b/gcc/ada/libgnat/s-powlfl.ads
new file mode 100644
index 0000000..d191eff
--- /dev/null
+++ b/gcc/ada/libgnat/s-powlfl.ads
@@ -0,0 +1,355 @@
+------------------------------------------------------------------------------
+--                                                                          --
+--                         GNAT COMPILER COMPONENTS                         --
+--                                                                          --
+--                   S Y S T E M . P O W T E N _ L F L T                    --
+--                                                                          --
+--                                 S p e c                                  --
+--                                                                          --
+--             Copyright (C) 2020, 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.                                     --
+--                                                                          --
+-- As a special exception under Section 7 of GPL version 3, you are granted --
+-- additional permissions described in the GCC Runtime Library Exception,   --
+-- version 3.1, as published by the Free Software Foundation.               --
+--                                                                          --
+-- You should have received a copy of the GNU General Public License and    --
+-- a copy of the GCC Runtime Library Exception along with this program;     --
+-- see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see    --
+-- <http://www.gnu.org/licenses/>.                                          --
+--                                                                          --
+-- GNAT was originally developed  by the GNAT team at  New York University. --
+-- Extensive contributions were provided by Ada Core Technologies Inc.      --
+--                                                                          --
+------------------------------------------------------------------------------
+
+--  This package provides a powers of ten table used for real conversions
+
+package System.Powten_LFlt is
+   pragma Pure;
+
+   Maxpow : constant := 308;
+   --  Largest power of ten representable with Long_Float
+
+   Maxpow_Exact : constant := 22;
+   --  Largest power of ten exactly representable with Long_Float. It is equal
+   --  to floor (M * log 2 / log 5), when M is the size of the mantissa (53).
+
+   Powten : constant array (0 .. Maxpow) of Long_Float :=
+      (00 => 1.0E+00,
+       01 => 1.0E+01,
+       02 => 1.0E+02,
+       03 => 1.0E+03,
+       04 => 1.0E+04,
+       05 => 1.0E+05,
+       06 => 1.0E+06,
+       07 => 1.0E+07,
+       08 => 1.0E+08,
+       09 => 1.0E+09,
+       10 => 1.0E+10,
+       11 => 1.0E+11,
+       12 => 1.0E+12,
+       13 => 1.0E+13,
+       14 => 1.0E+14,
+       15 => 1.0E+15,
+       16 => 1.0E+16,
+       17 => 1.0E+17,
+       18 => 1.0E+18,
+       19 => 1.0E+19,
+       20 => 1.0E+20,
+       21 => 1.0E+21,
+       22 => 1.0E+22,
+       23 => 1.0E+23,
+       24 => 1.0E+24,
+       25 => 1.0E+25,
+       26 => 1.0E+26,
+       27 => 1.0E+27,
+       28 => 1.0E+28,
+       29 => 1.0E+29,
+       30 => 1.0E+30,
+       31 => 1.0E+31,
+       32 => 1.0E+32,
+       33 => 1.0E+33,
+       34 => 1.0E+34,
+       35 => 1.0E+35,
+       36 => 1.0E+36,
+       37 => 1.0E+37,
+       38 => 1.0E+38,
+       39 => 1.0E+39,
+       40 => 1.0E+40,
+       41 => 1.0E+41,
+       42 => 1.0E+42,
+       43 => 1.0E+43,
+       44 => 1.0E+44,
+       45 => 1.0E+45,
+       46 => 1.0E+46,
+       47 => 1.0E+47,
+       48 => 1.0E+48,
+       49 => 1.0E+49,
+       50 => 1.0E+50,
+       51 => 1.0E+51,
+       52 => 1.0E+52,
+       53 => 1.0E+53,
+       54 => 1.0E+54,
+       55 => 1.0E+55,
+       56 => 1.0E+56,
+       57 => 1.0E+57,
+       58 => 1.0E+58,
+       59 => 1.0E+59,
+       60 => 1.0E+60,
+       61 => 1.0E+61,
+       62 => 1.0E+62,
+       63 => 1.0E+63,
+       64 => 1.0E+64,
+       65 => 1.0E+65,
+       66 => 1.0E+66,
+       67 => 1.0E+67,
+       68 => 1.0E+68,
+       69 => 1.0E+69,
+       70 => 1.0E+70,
+       71 => 1.0E+71,
+       72 => 1.0E+72,
+       73 => 1.0E+73,
+       74 => 1.0E+74,
+       75 => 1.0E+75,
+       76 => 1.0E+76,
+       77 => 1.0E+77,
+       78 => 1.0E+78,
+       79 => 1.0E+79,
+       80 => 1.0E+80,
+       81 => 1.0E+81,
+       82 => 1.0E+82,
+       83 => 1.0E+83,
+       84 => 1.0E+84,
+       85 => 1.0E+85,
+       86 => 1.0E+86,
+       87 => 1.0E+87,
+       88 => 1.0E+88,
+       89 => 1.0E+89,
+       90 => 1.0E+90,
+       91 => 1.0E+91,
+       92 => 1.0E+92,
+       93 => 1.0E+93,
+       94 => 1.0E+94,
+       95 => 1.0E+95,
+       96 => 1.0E+96,
+       97 => 1.0E+97,
+       98 => 1.0E+98,
+       99 => 1.0E+99,
+       100 => 1.0E+100,
+       101 => 1.0E+101,
+       102 => 1.0E+102,
+       103 => 1.0E+103,
+       104 => 1.0E+104,
+       105 => 1.0E+105,
+       106 => 1.0E+106,
+       107 => 1.0E+107,
+       108 => 1.0E+108,
+       109 => 1.0E+109,
+       110 => 1.0E+110,
+       111 => 1.0E+111,
+       112 => 1.0E+112,
+       113 => 1.0E+113,
+       114 => 1.0E+114,
+       115 => 1.0E+115,
+       116 => 1.0E+116,
+       117 => 1.0E+117,
+       118 => 1.0E+118,
+       119 => 1.0E+119,
+       120 => 1.0E+120,
+       121 => 1.0E+121,
+       122 => 1.0E+122,
+       123 => 1.0E+123,
+       124 => 1.0E+124,
+       125 => 1.0E+125,
+       126 => 1.0E+126,
+       127 => 1.0E+127,
+       128 => 1.0E+128,
+       129 => 1.0E+129,
+       130 => 1.0E+130,
+       131 => 1.0E+131,
+       132 => 1.0E+132,
+       133 => 1.0E+133,
+       134 => 1.0E+134,
+       135 => 1.0E+135,
+       136 => 1.0E+136,
+       137 => 1.0E+137,
+       138 => 1.0E+138,
+       139 => 1.0E+139,
+       140 => 1.0E+140,
+       141 => 1.0E+141,
+       142 => 1.0E+142,
+       143 => 1.0E+143,
+       144 => 1.0E+144,
+       145 => 1.0E+145,
+       146 => 1.0E+146,
+       147 => 1.0E+147,
+       148 => 1.0E+148,
+       149 => 1.0E+149,
+       150 => 1.0E+150,
+       151 => 1.0E+151,
+       152 => 1.0E+152,
+       153 => 1.0E+153,
+       154 => 1.0E+154,
+       155 => 1.0E+155,
+       156 => 1.0E+156,
+       157 => 1.0E+157,
+       158 => 1.0E+158,
+       159 => 1.0E+159,
+       160 => 1.0E+160,
+       161 => 1.0E+161,
+       162 => 1.0E+162,
+       163 => 1.0E+163,
+       164 => 1.0E+164,
+       165 => 1.0E+165,
+       166 => 1.0E+166,
+       167 => 1.0E+167,
+       168 => 1.0E+168,
+       169 => 1.0E+169,
+       170 => 1.0E+170,
+       171 => 1.0E+171,
+       172 => 1.0E+172,
+       173 => 1.0E+173,
+       174 => 1.0E+174,
+       175 => 1.0E+175,
+       176 => 1.0E+176,
+       177 => 1.0E+177,
+       178 => 1.0E+178,
+       179 => 1.0E+179,
+       180 => 1.0E+180,
+       181 => 1.0E+181,
+       182 => 1.0E+182,
+       183 => 1.0E+183,
+       184 => 1.0E+184,
+       185 => 1.0E+185,
+       186 => 1.0E+186,
+       187 => 1.0E+187,
+       188 => 1.0E+188,
+       189 => 1.0E+189,
+       190 => 1.0E+190,
+       191 => 1.0E+191,
+       192 => 1.0E+192,
+       193 => 1.0E+193,
+       194 => 1.0E+194,
+       195 => 1.0E+195,
+       196 => 1.0E+196,
+       197 => 1.0E+197,
+       198 => 1.0E+198,
+       199 => 1.0E+199,
+       200 => 1.0E+200,
+       201 => 1.0E+201,
+       202 => 1.0E+202,
+       203 => 1.0E+203,
+       204 => 1.0E+204,
+       205 => 1.0E+205,
+       206 => 1.0E+206,
+       207 => 1.0E+207,
+       208 => 1.0E+208,
+       209 => 1.0E+209,
+       210 => 1.0E+210,
+       211 => 1.0E+211,
+       212 => 1.0E+212,
+       213 => 1.0E+213,
+       214 => 1.0E+214,
+       215 => 1.0E+215,
+       216 => 1.0E+216,
+       217 => 1.0E+217,
+       218 => 1.0E+218,
+       219 => 1.0E+219,
+       220 => 1.0E+220,
+       221 => 1.0E+221,
+       222 => 1.0E+222,
+       223 => 1.0E+223,
+       224 => 1.0E+224,
+       225 => 1.0E+225,
+       226 => 1.0E+226,
+       227 => 1.0E+227,
+       228 => 1.0E+228,
+       229 => 1.0E+229,
+       230 => 1.0E+230,
+       231 => 1.0E+231,
+       232 => 1.0E+232,
+       233 => 1.0E+233,
+       234 => 1.0E+234,
+       235 => 1.0E+235,
+       236 => 1.0E+236,
+       237 => 1.0E+237,
+       238 => 1.0E+238,
+       239 => 1.0E+239,
+       240 => 1.0E+240,
+       241 => 1.0E+241,
+       242 => 1.0E+242,
+       243 => 1.0E+243,
+       244 => 1.0E+244,
+       245 => 1.0E+245,
+       246 => 1.0E+246,
+       247 => 1.0E+247,
+       248 => 1.0E+248,
+       249 => 1.0E+249,
+       250 => 1.0E+250,
+       251 => 1.0E+251,
+       252 => 1.0E+252,
+       253 => 1.0E+253,
+       254 => 1.0E+254,
+       255 => 1.0E+255,
+       256 => 1.0E+256,
+       257 => 1.0E+257,
+       258 => 1.0E+258,
+       259 => 1.0E+259,
+       260 => 1.0E+260,
+       261 => 1.0E+261,
+       262 => 1.0E+262,
+       263 => 1.0E+263,
+       264 => 1.0E+264,
+       265 => 1.0E+265,
+       266 => 1.0E+266,
+       267 => 1.0E+267,
+       268 => 1.0E+268,
+       269 => 1.0E+269,
+       270 => 1.0E+270,
+       271 => 1.0E+271,
+       272 => 1.0E+272,
+       273 => 1.0E+273,
+       274 => 1.0E+274,
+       275 => 1.0E+275,
+       276 => 1.0E+276,
+       277 => 1.0E+277,
+       278 => 1.0E+278,
+       279 => 1.0E+279,
+       280 => 1.0E+280,
+       281 => 1.0E+281,
+       282 => 1.0E+282,
+       283 => 1.0E+283,
+       284 => 1.0E+284,
+       285 => 1.0E+285,
+       286 => 1.0E+286,
+       287 => 1.0E+287,
+       288 => 1.0E+288,
+       289 => 1.0E+289,
+       290 => 1.0E+290,
+       291 => 1.0E+291,
+       292 => 1.0E+292,
+       293 => 1.0E+293,
+       294 => 1.0E+294,
+       295 => 1.0E+295,
+       296 => 1.0E+296,
+       297 => 1.0E+297,
+       298 => 1.0E+298,
+       299 => 1.0E+299,
+       300 => 1.0E+300,
+       301 => 1.0E+301,
+       302 => 1.0E+302,
+       303 => 1.0E+303,
+       304 => 1.0E+304,
+       305 => 1.0E+305,
+       306 => 1.0E+306,
+       307 => 1.0E+307,
+       308 => 1.0E+308);
+
+end System.Powten_LFlt;
diff --git a/gcc/ada/libgnat/s-powtab.ads b/gcc/ada/libgnat/s-powllf.ads
index 79982b9..c5c42a1 100644
--- a/gcc/ada/libgnat/s-powtab.ads
+++ b/gcc/ada/libgnat/s-powllf.ads
@@ -2,7 +2,7 @@
 --                                                                          --
 --                         GNAT COMPILER COMPONENTS                         --
 --                                                                          --
---                  S Y S T E M . P O W T E N _ T A B L E                   --
+--                    S Y S T E M . P O W T E N _ L L F                     --
 --                                                                          --
 --                                 S p e c                                  --
 --                                                                          --
@@ -31,7 +31,7 @@
 
 --  This package provides a powers of ten table used for real conversions
 
-package System.Powten_Table is
+package System.Powten_LLF is
    pragma Pure;
 
    Maxpow : constant := 22;
@@ -67,4 +67,4 @@ package System.Powten_Table is
        21 => 1.0E+21,
        22 => 1.0E+22);
 
-end System.Powten_Table;
+end System.Powten_LLF;
diff --git a/gcc/ada/libgnat/s-valflt.ads b/gcc/ada/libgnat/s-valflt.ads
index 476a251..5806d58 100644
--- a/gcc/ada/libgnat/s-valflt.ads
+++ b/gcc/ada/libgnat/s-valflt.ads
@@ -33,12 +33,17 @@
 --  type Float, for use in Text_IO.Float_IO and the Value attribute.
 
 with Interfaces;
+with System.Powten_Flt;
 with System.Val_Real;
 
 package System.Val_Flt is
    pragma Preelaborate;
 
-   package Impl is new Val_Real (Float, Interfaces.Unsigned_32);
+   package Impl is new Val_Real
+     (Float,
+      Interfaces.Unsigned_32,
+      System.Powten_Flt.Maxpow,
+      System.Powten_Flt.Powten'Address);
 
    function Scan_Float
      (Str : String;
diff --git a/gcc/ada/libgnat/s-vallfl.ads b/gcc/ada/libgnat/s-vallfl.ads
index 5bb6da4..c612f75 100644
--- a/gcc/ada/libgnat/s-vallfl.ads
+++ b/gcc/ada/libgnat/s-vallfl.ads
@@ -33,12 +33,17 @@
 --  type Long_Float, for use in Text_IO.Float_IO and the Value attribute.
 
 with Interfaces;
+with System.Powten_LFlt;
 with System.Val_Real;
 
 package System.Val_LFlt is
    pragma Preelaborate;
 
-   package Impl is new Val_Real (Long_Float, Interfaces.Unsigned_64);
+   package Impl is new Val_Real
+     (Long_Float,
+      Interfaces.Unsigned_64,
+      System.Powten_LFlt.Maxpow,
+      System.Powten_LFlt.Powten'Address);
 
    function Scan_Long_Float
      (Str : String;
diff --git a/gcc/ada/libgnat/s-valllf.ads b/gcc/ada/libgnat/s-valllf.ads
index 715f6ac..46a311b 100644
--- a/gcc/ada/libgnat/s-valllf.ads
+++ b/gcc/ada/libgnat/s-valllf.ads
@@ -33,12 +33,17 @@
 --  type Long_Long_Float, for use in Text_IO.Float_IO and the Value attribute.
 
 with Interfaces;
+with System.Powten_LLF;
 with System.Val_Real;
 
 package System.Val_LLF is
    pragma Preelaborate;
 
-   package Impl is new Val_Real (Long_Long_Float, Interfaces.Unsigned_64);
+   package Impl is new Val_Real
+     (Long_Long_Float,
+      Interfaces.Unsigned_64,
+      System.Powten_LLF.Maxpow,
+      System.Powten_LLF.Powten'Address);
 
    function Scan_Long_Long_Float
      (Str : String;
diff --git a/gcc/ada/libgnat/s-valrea.adb b/gcc/ada/libgnat/s-valrea.adb
index 5ce3642..9614760 100644
--- a/gcc/ada/libgnat/s-valrea.adb
+++ b/gcc/ada/libgnat/s-valrea.adb
@@ -34,15 +34,26 @@ with System.Unsigned_Types; use System.Unsigned_Types;
 with System.Val_Util;       use System.Val_Util;
 with System.Value_R;
 
+pragma Warnings (Off, "non-static constant in preelaborated unit");
+--  Every constant is static given our instantiation model
+
 package body System.Val_Real is
 
    pragma Assert (Num'Machine_Mantissa <= Uns'Size);
    --  We need an unsigned type large enough to represent the mantissa
 
-   Precision_Limit : constant Uns := 2**Num'Machine_Mantissa - 1;
-   --  We use the precision of the floating-point type
+   Need_Extra : constant Boolean := Num'Machine_Mantissa > Uns'Size - 4;
+   --  If the mantissa of the floating-point type is almost as large as that
+   --  of the unsigned type, we do not have enough space for an extra digit
+   --  in the unsigned type so we handle the extra digit separately, at the
+   --  cost of a potential roundoff error.
+
+   Precision_Limit : constant Uns :=
+     (if Need_Extra then 2**Num'Machine_Mantissa - 1 else 2**Uns'Size - 1);
+   --  If we handle the extra digit separately, we use the precision of the
+   --  floating-point type so that the conversion is exact.
 
-   package Impl is new Value_R (Uns, Precision_Limit, Floating => True);
+   package Impl is new Value_R (Uns, Precision_Limit, Round => Need_Extra);
 
    subtype Base_T is Unsigned range 2 .. 16;
 
@@ -88,6 +99,8 @@ package body System.Val_Real is
    is
       pragma Assert (Base in 2 .. 16);
 
+      pragma Assert (Num'Machine_Radix = 2);
+
       pragma Unsuppress (Range_Check);
 
       Maxexp : constant Positive :=
@@ -112,29 +125,98 @@ package body System.Val_Real is
          System.Float_Control.Reset;
       end if;
 
-      --  Take into account the extra digit
+      --  Do the conversion
 
       R_Val := Num (Val);
-      if Extra > 0 then
+
+      --  Take into account the extra digit, if need be. In this case, the
+      --  three operands are exact, so using an FMA would be ideal.
+
+      if Need_Extra and then Extra > 0 then
          R_Val := R_Val * B + Num (Extra);
          S := S - 1;
       end if;
 
-      --  Compute the final value. When the exponent is positive, we can do the
-      --  computation directly because, if the exponentiation overflows, then
-      --  the final value overflows as well. But when the exponent is negative,
-      --  we may need to do it in two steps to avoid an artificial underflow.
+      --  Compute the final value
+
+      if R_Val /= 0.0 and then S /= 0 then
+         case Base is
+            --  If the base is a power of two, we use the efficient Scaling
+            --  attribute with an overflow check, if it is not 2, to catch
+            --  ludicrous exponents that would result in an infinity or zero.
+
+            when 2 =>
+               R_Val := Num'Scaling (R_Val, S);
+
+            when 4 =>
+               if Integer'First / 2 <= S and then S <= Integer'Last / 2 then
+                  S := S * 2;
+               end if;
+
+               R_Val := Num'Scaling (R_Val, S);
+
+            when 8 =>
+               if Integer'First / 3 <= S and then S <= Integer'Last / 3 then
+                  S := S * 3;
+               end if;
+
+               R_Val := Num'Scaling (R_Val, S);
+
+            when 16 =>
+               if Integer'First / 4 <= S and then S <= Integer'Last / 4 then
+                  S := S * 4;
+               end if;
+
+               R_Val := Num'Scaling (R_Val, S);
+
+            --  If the base is 10, we use a table of powers for accuracy's sake
+
+            when 10 =>
+               declare
+                  Powten : constant array (0 .. Maxpow) of Num;
+                  pragma Import (Ada, Powten);
+                  for Powten'Address use Powten_Address;
+
+               begin
+                  if S > 0 then
+                     while S > Maxpow loop
+                        R_Val := R_Val * Powten (Maxpow);
+                        S := S - Maxpow;
+                     end loop;
+
+                     R_Val := R_Val * Powten (S);
+
+                  else
+                     while S < -Maxpow loop
+                        R_Val := R_Val / Powten (Maxpow);
+                        S := S + Maxpow;
+                     end loop;
+
+                     R_Val := R_Val / Powten (-S);
+                  end if;
+               end;
+
+            --  Implementation for other bases with exponentiation
+
+            --  When the exponent is positive, we can do the computation
+            --  directly because, if the exponentiation overflows, then
+            --  the final value overflows as well. But when the exponent
+            --  is negative, we may need to do it in two steps to avoid
+            --  an artificial underflow.
 
-      if S > 0 then
-         R_Val := R_Val * B ** S;
+            when others =>
+               if S > 0 then
+                  R_Val := R_Val * B ** S;
 
-      elsif S < 0 then
-         if S < -Maxexp then
-            R_Val := R_Val / B ** Maxexp;
-            S := S + Maxexp;
-         end if;
+               else
+                  if S < -Maxexp then
+                     R_Val := R_Val / B ** Maxexp;
+                     S := S + Maxexp;
+                  end if;
 
-         R_Val := R_Val / B ** (-S);
+                  R_Val := R_Val / B ** (-S);
+               end if;
+         end case;
       end if;
 
       --  Finally deal with initial minus sign, note that this processing is
diff --git a/gcc/ada/libgnat/s-valrea.ads b/gcc/ada/libgnat/s-valrea.ads
index 961c4803..d6ade80 100644
--- a/gcc/ada/libgnat/s-valrea.ads
+++ b/gcc/ada/libgnat/s-valrea.ads
@@ -38,6 +38,10 @@ generic
 
    type Uns is mod <>;
 
+   Maxpow : Positive;
+
+   Powten_Address : System.Address;
+
 package System.Val_Real is
    pragma Preelaborate;
 
diff --git a/gcc/ada/libgnat/s-valued.adb b/gcc/ada/libgnat/s-valued.adb
index 7986ce3..8930752 100644
--- a/gcc/ada/libgnat/s-valued.adb
+++ b/gcc/ada/libgnat/s-valued.adb
@@ -38,7 +38,8 @@ package body System.Value_D is
    pragma Assert (Int'Size <= Uns'Size);
    --  We need an unsigned type large enough to represent the mantissa
 
-   package Impl is new Value_R (Uns, 2**(Int'Size - 1), Floating => False);
+   package Impl is new Value_R (Uns, 2**(Int'Size - 1), Round => False);
+   --  We do not use the Extra digit for decimal fixed-point types
 
    function Integer_to_Decimal
      (Str    : String;
@@ -231,6 +232,7 @@ package body System.Value_D is
       Base   : Unsigned;
       ScaleB : Integer;
       Extra  : Unsigned;
+      pragma Unreferenced (Extra);
       Minus  : Boolean;
       Val    : Uns;
 
@@ -248,6 +250,7 @@ package body System.Value_D is
       Base   : Unsigned;
       ScaleB : Integer;
       Extra  : Unsigned;
+      pragma Unreferenced (Extra);
       Minus  : Boolean;
       Val    : Uns;
 
diff --git a/gcc/ada/libgnat/s-valuef.adb b/gcc/ada/libgnat/s-valuef.adb
index 5a87a7f..d13111a 100644
--- a/gcc/ada/libgnat/s-valuef.adb
+++ b/gcc/ada/libgnat/s-valuef.adb
@@ -46,7 +46,8 @@ package body System.Value_F is
    pragma Assert (Int'Size <= Uns'Size);
    --  We need an unsigned type large enough to represent the mantissa
 
-   package Impl is new Value_R (Uns, 2**(Int'Size - 1), Floating => False);
+   package Impl is new Value_R (Uns, 2**(Int'Size - 1), Round => True);
+   --  We use the Extra digit for ordinary fixed-point types
 
    function Integer_To_Fixed
      (Str    : String;
diff --git a/gcc/ada/libgnat/s-valuer.adb b/gcc/ada/libgnat/s-valuer.adb
index 65a0d50..9e4de3e 100644
--- a/gcc/ada/libgnat/s-valuer.adb
+++ b/gcc/ada/libgnat/s-valuer.adb
@@ -42,6 +42,14 @@ package body System.Value_R is
    function As_Digit (C : Character) return Char_As_Digit;
    --  Given a character return the digit it represents
 
+   procedure Round_Extra
+     (Digit : Char_As_Digit;
+      Value : in out Uns;
+      Scale : in out Integer;
+      Extra : in out Char_As_Digit;
+      Base  : Unsigned);
+   --  Round the triplet (Value, Scale, Extra) according to Digit in Base
+
    procedure Scan_Decimal_Digits
       (Str            : String;
        Index          : in out Integer;
@@ -109,6 +117,45 @@ package body System.Value_R is
       end case;
    end As_Digit;
 
+   -----------------
+   -- Round_Extra --
+   -----------------
+
+   procedure Round_Extra
+     (Digit : Char_As_Digit;
+      Value : in out Uns;
+      Scale : in out Integer;
+      Extra : in out Char_As_Digit;
+      Base  : Unsigned)
+   is
+      B : constant Uns := Uns (Base);
+
+   begin
+      if Digit >= Base / 2 then
+
+         --  If Extra is maximum, round Value
+
+         if Extra = Base - 1 then
+
+            --  If Value is maximum, scale it up
+
+            if Value = Precision_Limit then
+               Extra := Char_As_Digit (Value mod B);
+               Value := Value / B;
+               Scale := Scale + 1;
+               Round_Extra (Digit, Value, Scale, Extra, Base);
+
+            else
+               Extra := 0;
+               Value := Value + 1;
+            end if;
+
+         else
+            Extra := Extra + 1;
+         end if;
+      end if;
+   end Round_Extra;
+
    -------------------------
    -- Scan_Decimal_Digits --
    -------------------------
@@ -140,8 +187,8 @@ package body System.Value_R is
       --  to Precision_Limit.
 
       Precision_Limit_Just_Reached : Boolean;
-      --  Set to True if Precision_Limit_Reached was just set to True
-      --  Only used when Floating = False.
+      --  Set to True if Precision_Limit_Reached was just set to True, but only
+      --  used when Round is True.
 
       Digit : Char_As_Digit;
       --  The current digit
@@ -162,7 +209,7 @@ package body System.Value_R is
          Extra := 0;
       end if;
 
-      if not Floating then
+      if Round then
          Precision_Limit_Just_Reached := False;
       end if;
 
@@ -188,22 +235,12 @@ package body System.Value_R is
 
          --  If precision limit has been reached, just ignore any remaining
          --  digits for the computation of Value and Scale, but store the
-         --  first in Extra and use the second to round Extra if this is for
-         --  a fixed-point type (we skip the rounding for a floating-point
-         --  type to preserve backward compatibility). The scanning should
-         --  continue only to assess the validity of the string.
+         --  first in Extra and use the second to round Extra. The scanning
+         --  should continue only to assess the validity of the string.
 
          if Precision_Limit_Reached then
-            if not Floating and then Precision_Limit_Just_Reached then
-               if Digit >= Base / 2 then
-                  if Extra = Base - 1 then
-                     Extra := 0;
-                     Value := Value + 1;
-                  else
-                     Extra := Extra + 1;
-                  end if;
-               end if;
-
+            if Round and then Precision_Limit_Just_Reached then
+               Round_Extra (Digit, Value, Scale, Extra, Base);
                Precision_Limit_Just_Reached := False;
             end if;
 
@@ -252,8 +289,7 @@ package body System.Value_R is
                else
                   Extra := Digit;
                   Precision_Limit_Reached := True;
-
-                  if not Floating then
+                  if Round then
                      Precision_Limit_Just_Reached := True;
                   end if;
                end if;
@@ -320,8 +356,8 @@ package body System.Value_R is
       --  to Precision_Limit.
 
       Precision_Limit_Just_Reached : Boolean;
-      --  Set to True if Precision_Limit_Reached was just set to True.
-      --  Only used when Floating = False.
+      --  Set to True if Precision_Limit_Reached was just set to True, but only
+      --  used when Round is True.
 
       Digit : Char_As_Digit;
       --  The current digit
@@ -336,7 +372,7 @@ package body System.Value_R is
       Scale := 0;
       Extra := 0;
 
-      if not Floating then
+      if Round then
          Precision_Limit_Just_Reached := False;
       end if;
 
@@ -364,24 +400,14 @@ package body System.Value_R is
 
          --  If precision limit has been reached, just ignore any remaining
          --  digits for the computation of Value and Scale, but store the
-         --  first in Extra and use the second to round Extra if this is for
-         --  a fixed-point type (we skip the rounding for a floating-point
-         --  type to preserve backward compatibility). The scanning should
-         --  continue only to assess the validity of the string.
+         --  first in Extra and use the second to round Extra. The scanning
+         --  should continue only to assess the validity of the string.
 
          if Precision_Limit_Reached then
             Scale := Scale + 1;
 
-            if not Floating and then Precision_Limit_Just_Reached then
-               if Digit >= Base / 2 then
-                  if Extra = Base - 1 then
-                     Extra := 0;
-                     Value := Value + 1;
-                  else
-                     Extra := Extra + 1;
-                  end if;
-               end if;
-
+            if Round and then Precision_Limit_Just_Reached then
+               Round_Extra (Digit, Value, Scale, Extra, Base);
                Precision_Limit_Just_Reached := False;
             end if;
 
@@ -404,11 +430,9 @@ package body System.Value_R is
             else
                Extra := Digit;
                Precision_Limit_Reached := True;
-
-               if not Floating then
+               if Round then
                   Precision_Limit_Just_Reached := True;
                end if;
-
                Scale := Scale + 1;
             end if;
          end if;
diff --git a/gcc/ada/libgnat/s-valuer.ads b/gcc/ada/libgnat/s-valuer.ads
index 06fbe9e..a933859 100644
--- a/gcc/ada/libgnat/s-valuer.ads
+++ b/gcc/ada/libgnat/s-valuer.ads
@@ -40,7 +40,7 @@ generic
 
    Precision_Limit : Uns;
 
-   Floating : Boolean;
+   Round : Boolean;
 
 package System.Value_R is
    pragma Preelaborate;