GNAT Compilation Problems

GNAT Compilation Problems

[Nigel J. Tracey]

I am having problems with compiling the below code using GNAT
3.05 on Linux 2.0.20 I have identified the line which causes
the problem but do not know why. With the program as it stands
I get a the following:

gcc: Internal compiler error: program gnat1 got fatal signal 11

Any help would be much appreciated. (Please also e-mail replies)

Thanks a lot,

	Nigel


--------------------
Nigel J. Tracey MEng, Research Associate (High Integrity Systems Group).
Department of Computer Science,   Office: X/D016
University of York,               Tel   : [0|+44]1904 432769
York, England.                    E-Mail: njt@minster.york.ac.uk
YO1 5DD.                          URL   : http://sv1pc161.cs.york.ac.uk/~njt



with Lists;
with String_Pkg;
with TEXT_IO; use TEXT_IO;

procedure variant is


   type Basic_Type is (NONE, ENUM, INT, BOOL, FLOAT_NUM, FIXED_NUM, CHAR);

   type Type_Information_Type(Base_Type : Basic_Type := NONE) is
      record
         Type_Name : String_Pkg.String_Type;
         case Base_Type is
            when INT | CHAR | ENUM =>
               Min_Integer : Integer;
               Max_Integer : Integer;
            when BOOL =>
               null;
            when FLOAT_NUM =>
               Float_Precision : Integer;
               Min_Float : Float;
               Max_Float : Float;
            when FIXED_NUM =>
               Fixed_Precision : Float;
               Min_Fixed : Float;
               Max_Fixed : Float;
            when NONE =>
               null;
         end case;
      end record;

   package Type_List is new
      Lists(Type_Information_Type);

   TypeLst : Type_List.List;

   Standard_Ada_Types : constant array(1..4) of Type_Information_Type :=
      ((INT, String_pkg.Create("Integer"), -2147483647, 2147483647),
       (BOOL, String_pkg.Create("Boolean")),
       (FLOAT_NUM, String_pkg.Create("Float"), 6, -3.40282E+38, 3.40282E+38),
       (INT, String_pkg.Create("Character"), 0, 255));

   procedure Add_Subtype
     (Type_Name : String_Pkg.String_Type;
      Subtype_Name : String_Pkg.String_Type)
   is

      List_Iterator : Type_List.ListIter;
      Temp : Type_Information_Type;

   begin

      List_Iterator := Type_List.MakeListIter(TypeLst);

      while Type_List.More(List_Iterator)
      loop
         Type_List.Next(List_Iterator, Temp);
         if String_Pkg.Equal(Temp.Type_Name, Subtype_Name) then
            Temp.Type_Name := Type_Name;
            Type_List.Attach(TypeLst, Temp);
            return;
         end if;
      end loop;

   end Add_Subtype;

	
begin
   TypeLst := Type_List.Create;
   for Counter in Standard_Ada_Types'Range
   loop
      Type_List.Attach(TypeLst, Standard_Ada_Types(Counter));
   end loop;
end Variant;

GNAT Compilation Problems

[Nigel J. Tracey]

I am having problems with compiling the below code using GNAT
3.05 on Linux 2.0.20 I have identified the line which causes
the problem but do not know why. With the program as it stands
I get a the following:

gcc: Internal compiler error: program gnat1 got fatal signal 11

The line which seems to cause the problem is higlighed below
as -- *****************
With this line commented out, things seem to compile fine.
This line is however vitial!!! :)

Any help would be much appreciated. (Please also e-mail replies)

Thanks a lot,

	Nigel


--------------------
Nigel J. Tracey MEng, Research Associate (High Integrity Systems Group).
Department of Computer Science,   Office: X/D016
University of York,               Tel   : [0|+44]1904 432769
York, England.                    E-Mail: njt@minster.york.ac.uk
YO1 5DD.                          URL   : http://sv1pc161.cs.york.ac.uk/~njt



with Lists;
with String_Pkg;
with TEXT_IO; use TEXT_IO;

procedure variant is


   type Basic_Type is (NONE, ENUM, INT, BOOL, FLOAT_NUM, FIXED_NUM, CHAR);

   type Type_Information_Type(Base_Type : Basic_Type := NONE) is
      record
         Type_Name : String_Pkg.String_Type;
         case Base_Type is
            when INT | CHAR | ENUM =>
               Min_Integer : Integer;
               Max_Integer : Integer;
            when BOOL =>
               null;
            when FLOAT_NUM =>
               Float_Precision : Integer;
               Min_Float : Float;
               Max_Float : Float;
            when FIXED_NUM =>
               Fixed_Precision : Float;
               Min_Fixed : Float;
               Max_Fixed : Float;
            when NONE =>
               null;
         end case;
      end record;

   package Type_List is new
      Lists(Type_Information_Type);

   TypeLst : Type_List.List;

   Standard_Ada_Types : constant array(1..4) of Type_Information_Type :=
      ((INT, String_pkg.Create("Integer"), -2147483647, 2147483647),
       (BOOL, String_pkg.Create("Boolean")),
       (FLOAT_NUM, String_pkg.Create("Float"), 6, -3.40282E+38, 3.40282E+38),
       (INT, String_pkg.Create("Character"), 0, 255));

   procedure Add_Subtype
     (Type_Name : String_Pkg.String_Type;
      Subtype_Name : String_Pkg.String_Type)
   is

      List_Iterator : Type_List.ListIter;
      Temp : Type_Information_Type;

   begin

      List_Iterator := Type_List.MakeListIter(TypeLst);

      while Type_List.More(List_Iterator)
      loop
      	-- ************************
      	-- The line below causes the problems
      	-- ************************
         Type_List.Next(List_Iterator, Temp);

         if String_Pkg.Equal(Temp.Type_Name, Subtype_Name) then
            Temp.Type_Name := Type_Name;
            Type_List.Attach(TypeLst, Temp);
            return;
         end if;
      end loop;

   end Add_Subtype;

	
begin
   TypeLst := Type_List.Create;
   for Counter in Standard_Ada_Types'Range
   loop
      Type_List.Attach(TypeLst, Standard_Ada_Types(Counter));
   end loop;
end Variant;

Re: GNAT Compilation Problems

[Robert Dewar]

Nigel said

"Nigel J. Tracey: GNAT Compilation Problems                   20 Sep 1996
I am having problems with compiling the below code using GNAT
3.05 on Linux 2.0.20 I have identified the line which causes
the problem but do not know why. With the program as it stands
I get a the following:"

This certainly looks like a bug, but this is not, as far as we are
concerned, a bug report, because it is in the wrong place. Please
be sure to follow the documentation and send a copy of the bug report
to report@gnat.com. Make sure you exactly follow the instructions
in gnatinfo.txt when you submit the report.

Robert Dewar
Ada Core Technologies

Re: GNAT Compilation Problems

[Robert Dewar]

Reposting article removed by rogue canceller.

Nigel said

"Nigel J. Tracey: GNAT Compilation Problems                   20 Sep 1996
I am having problems with compiling the below code using GNAT
3.05 on Linux 2.0.20 I have identified the line which causes
the problem but do not know why. With the program as it stands
I get a the following:"

This certainly looks like a bug, but this is not, as far as we are
concerned, a bug report, because it is in the wrong place. Please
be sure to follow the documentation and send a copy of the bug report
to report@gnat.com. Make sure you exactly follow the instructions
in gnatinfo.txt when you submit the report.

Robert Dewar
Ada Core Technologies

Re: GNAT Compilation Problems

[David C. Hoos, Sr.]

Nigel J. Tracey <njt@minster.york.ac.uk> wrote in article
<51u63c$80c@netty.york.ac.uk>...
> I am having problems with compiling the below code using GNAT
> 3.05 on Linux 2.0.20 I have identified the line which causes
> the problem but do not know why. With the program as it stands
> I get a the following:
> 
> gcc: Internal compiler error: program gnat1 got fatal signal 11
> 

It's not really possible to solve this problem without the units which you
"with", particularly since the problem line involves one of these "withed"
units.
-- 
David C. Hoos, Sr.,
http://www.dbhwww.com
http://www.ada95.com


> 

Re: GNAT Compilation Problems

[Graffiti]

In article <01bba753$d2824ee0$1c8371a5@dhoossr.iquest.com>,
David C. Hoos, Sr. <david.c.hoos.sr@ada95.com> wrote:
>Nigel J. Tracey <njt@minster.york.ac.uk> wrote in article
><51u63c$80c@netty.york.ac.uk>...
[snip]
>> 
>> gcc: Internal compiler error: program gnat1 got fatal signal 11

Although this post would be more appropriate in a linux newsgroup, I'll
answer it anyways.  Sig 11 indicates hardware trouble.  This comes up
so often, it's not only in a FAQ, there's a web page dedicated to it.

You'll find it at http://www.bitwizard.nl/sig11

obilynoh@ktb.net

Re: GNAT Compilation Problems

[David C. Hoos, Sr.]

Reposting article removed by rogue canceller.

Nigel J. Tracey <njt@minster.york.ac.uk> wrote in article
<51u63c$80c@netty.york.ac.uk>...
> I am having problems with compiling the below code using GNAT
> 3.05 on Linux 2.0.20 I have identified the line which causes
> the problem but do not know why. With the program as it stands
> I get a the following:
> 
> gcc: Internal compiler error: program gnat1 got fatal signal 11
> 

It's not really possible to solve this problem without the units which you
"with", particularly since the problem line involves one of these "withed"
units.
-- 
David C. Hoos, Sr.,
http://www.dbhwww.com
http://www.ada95.com


> 

Re: GNAT Compilation Problems

[Graffiti]

Reposting article removed by rogue canceller.

In article <01bba753$d2824ee0$1c8371a5@dhoossr.iquest.com>,
David C. Hoos, Sr. <david.c.hoos.sr@ada95.com> wrote:
>Nigel J. Tracey <njt@minster.york.ac.uk> wrote in article
><51u63c$80c@netty.york.ac.uk>...
[snip]
>> 
>> gcc: Internal compiler error: program gnat1 got fatal signal 11

Although this post would be more appropriate in a linux newsgroup, I'll
answer it anyways.  Sig 11 indicates hardware trouble.  This comes up
so often, it's not only in a FAQ, there's a web page dedicated to it.

You'll find it at http://www.bitwizard.nl/sig11

obilynoh@ktb.net

Re: GNAT Compilation Problems

[Keith Thompson]

In <520gjr$tf6@ktb2.ktb.net> obilynoh@ktb.net (Graffiti) writes:
> Although this post would be more appropriate in a linux newsgroup, I'll
> answer it anyways.  Sig 11 indicates hardware trouble.

Signal 11 is SIGSEGV, a segmentation violation.  I've seen this a number
of times, but never due to hardware trouble.  There are a number of
software problems that can cause a segmentation violation, including
dereferencing a null or garbage pointer.  (On the other hand, I've never
used Linux; my experience has been on other Unix-like operating systems.)

-- 
Keith Thompson (The_Other_Keith) kst@thomsoft.com <*>
TeleSoft^H^H^H^H^H^H^H^H Alsys^H^H^H^H^H Thomson Software Products
10251 Vista Sorrento Parkway, Suite 300, San Diego, CA, USA, 92121-2706
FIJAGDWOL

Re: GNAT Compilation Problems

[Simon Wright]

kst@thomsoft.com (Keith Thompson) writes:
> In <520gjr$tf6@ktb2.ktb.net> obilynoh@ktb.net (Graffiti) writes:
> > Although this post would be more appropriate in a linux newsgroup, I'll
> > answer it anyways.  Sig 11 indicates hardware trouble.
> 
> Signal 11 is SIGSEGV, a segmentation violation.  I've seen this a number
> of times, but never due to hardware trouble.  There are a number of
> software problems that can cause a segmentation violation, including
> dereferencing a null or garbage pointer.  (On the other hand, I've never
> used Linux; my experience has been on other Unix-like operating systems.)

I had trouble with building the Linux kernel/gcc/gnat/emacs/things
that required a lot of resources to compile. This sort of thing has
been attributed to hardware (eg, doubtful memory) in the Linux
groups. It went away for me after a recent kernel upgrade (2.0.17);
but since I had the box apart in the same timescale and took the
opportunity to jiggle the SIMMs I think this is a dubious datum :-)

-- 
Simon Wright                    Work Email: simon.j.wright@gecm.com
Ferranti Naval Systems                     Voice: +44(0)1705-701778
GEC-Marconi S3I Combat Systems Division      FAX: +44(0)1705-701800

Re: GNAT Compilation Problems

[Kevin English]

What's the command to compile Ada programs from dos prompt with the gnat?


: > 

--
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
{ This message was sent by :    |   "In short, life has been made into a }
{                               |      nonstop, commercially prepackaged }
{   Kevin                       |       masterbational fantasy."         }
{     William                   |                                        }
{       English                 |        -- Alan Bloom on MTV and Music  }
{                               |        -- The Closing of the American  }
{                               |                  Mind                  }
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Re: GNAT Compilation Problems

[Robert Dewar]

Kevin asked

"What's the command to compile Ada programs from dos prompt with the gnat?"

Kevin, if you are using gnat, you must read gnatinfo.txt, which answers
this and many other relevant questions.

Re: GNAT Compilation Problems

[Gandalf]

Robert Dewar wrote:
> 
> Kevin asked
> 
> "What's the command to compile Ada programs from dos prompt with the gnat?"
> 
> Kevin, if you are using gnat, you must read gnatinfo.txt, which answers
> this and many other relevant questions.

While Robert is correct that gnatinfo.txt has some great information,
the command lines for compilation using gnat for DO$ is

gcc -c 'your_file_name_goes_here'


There are a miriad of other switches with which you can compile, best to
read the referenced readin material to get a grasp onn them all.  To
link your Ada program, you'll need to type

gnatbl 'filename'

where 'filename' should be substituted with the name of your main
program.  Again there are some switches you can use which are outlined
in the reference reading material.  Good luck, read the group often for
more useful information!

Re: GNAT Compilation Problems

[Robert Dewar]

Gandalf said

"While Robert is correct that gnatinfo.txt has some great information,
the command lines for compilation using gnat for DO$ is

gcc -c 'your_file_name_goes_here'
"

This is very bad advice, as you will find out from reading gnatinfo.txt,
the proper way to compile programs using GNAT from the DOS prompt or 
anywhere else is to use gnatmake. There is very seldom a reason for typing
in a gcc command.

Re: GNAT Compilation Problems

[Robert Dewar]

Michael says

"If your intent is to check that a component is entirely up to date
(including bodies), use "gnatmake". If you have only to check
the legality and consistency of one component, then "gnatmake" has
the harmful side-effect of attempting at compiling bodies, and "gcc"
is better.

So, I assert that "gcc" is more often the good choice than "gnatmake"."



The original question was about compiling, and most people think of
generating code when they are compiling. If you just want to check
legality, the best tool is gnatf.

But for most purposes, most people should use gnatmake all the time and
not use gcc directly. The typical cycle is

1. Edit some sources
2. gnatmake the main unit
3. run and test

The reason that gnatf is better than gcc for checking legality is that
(a) you do not have to bother with saying -gnatc
(b) you can check multiple units at the same time
(c) you can generate cross reference information
(d) or minimally, you can check for unused entities

Re: GNAT Compilation Problems

[Michael Feldman]

In article <dewar.844905925@merv>, Robert Dewar <dewar@merv.cs.nyu.edu> wrote:

>This is very bad advice, as you will find out from reading gnatinfo.txt,
>the proper way to compile programs using GNAT from the DOS prompt or 
>anywhere else is to use gnatmake. There is very seldom a reason for typing
>in a gcc command.

I agree with Robert here in general. OTOH, for beginners who have 
started out with GNAT-DOS using the ez2load package, we provide some
simplified commands (gcompile/glink/gexecute) that behave consistenly 
with the AdaCAPS editor.

For example, gcompile produces a listing file but, more important, they 
compile and link so that our special backtrace stuff works.

In this case, the answer for the student is

gcompile foo.adb
glink foo.ali
gexecute foo.exe

This is documented in the ez2load documentation.

Our Solaris scripts work the same way. Robert, I think you may even be
using those scripts in your course at NYU.:-)

Mike Feldman

Re: GNAT Compilation Problems

[Michel Gauthier]

In article <dewar.844905925@merv>, dewar@merv.cs.nyu.edu (Robert Dewar) wrote:

>>  Gandalf said
>>  
>>  "While Robert is correct that gnatinfo.txt has some great information,
>>  the command lines for compilation using gnat for DO$ is
>>  
>>  gcc -c 'your_file_name_goes_here'
>>  "
>>  This is very bad advice, as you will find out from reading gnatinfo.txt,
>>  the proper way to compile programs using GNAT from the DOS prompt or 
>>  anywhere else is to use gnatmake. There is very seldom a reason for typing
>>  in a gcc command.

Why ? "gcc" and "gnatmake" are suitable for different cases.

If your intent is to check that a component is entirely up to date
(including bodies), use "gnatmake". If you have only to check
the legality and consistency of one component, then "gnatmake" has
the harmful side-effect of attempting at compiling bodies, and "gcc"
is better.

So, I assert that "gcc" is more often the good choice than "gnatmake".

----------          ----------          ----------          ---------- 
Michel Gauthier / Laboratoire d'informatique
123 avenue Albert Thomas / F-87060 Limoges
telephone +33 () 55457335 [or ~ 7232]
    a partir du 19 octobre, depuis la France  : 05 55 45 73 35 
     from october,19 on, and from abroad : +33 5 55 45 73 35
fax +33 ()  55457315  [or ~7201]  
    and similar evolution
----------          ----------          ----------          ----------
Si l'an 2000 est pour vous un mysticisme stupide, utilisez la base 9
If you feel year 2000 a stupid mystic craze, use numeration base 9
----------          ----------          ----------          ----------