New Ada build problem on ia64

New Ada build problem on ia64

[Andreas Schwab]

There is a new build problem with the Ada compiler on ia64:

gcc -c -g    -W -Wall -gnatpg -gnata -I- -I. -I../../../gcc/ada ../../../gcc/ada/ali.adb
ali.adb:52:27: warning: loop range is null, loop will not execute
ali.adb:473:18: warning: loop range is null, loop will not execute
ali.adb:637:19: warning: loop range is null, loop will not execute
ali.adb:757:27: warning: loop range is null, loop will not execute
ali.adb:1040:30: warning: loop range is null, loop will not execute

The compiler has been built by a ia32->ia64 cross compiler, using CVS from
earlier today.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

Re: New Ada build problem on ia64

[Andreas Schwab]

Andreas Schwab <schwab@suse.de> writes:

|> There is a new build problem with the Ada compiler on ia64:
|> 
|> gcc -c -g    -W -Wall -gnatpg -gnata -I- -I. -I../../../gcc/ada ../../../gcc/ada/ali.adb
|> ali.adb:52:27: warning: loop range is null, loop will not execute
|> ali.adb:473:18: warning: loop range is null, loop will not execute
|> ali.adb:637:19: warning: loop range is null, loop will not execute
|> ali.adb:757:27: warning: loop range is null, loop will not execute
|> ali.adb:1040:30: warning: loop range is null, loop will not execute
|> 
|> The compiler has been built by a ia32->ia64 cross compiler, using CVS from
|> earlier today.

Update: the problem vanishes if I build the compiler with -O0.  I'm now
bootstrapping natively to see whether the problem reappears.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

Re: New Ada build problem on ia64

[Andreas Schwab]

Andreas Schwab <schwab@suse.de> writes:

|> Andreas Schwab <schwab@suse.de> writes:
|> 
|> |> There is a new build problem with the Ada compiler on ia64:
|> |> 
|> |> gcc -c -g    -W -Wall -gnatpg -gnata -I- -I. -I../../../gcc/ada ../../../gcc/ada/ali.adb
|> |> ali.adb:52:27: warning: loop range is null, loop will not execute
|> |> ali.adb:473:18: warning: loop range is null, loop will not execute
|> |> ali.adb:637:19: warning: loop range is null, loop will not execute
|> |> ali.adb:757:27: warning: loop range is null, loop will not execute
|> |> ali.adb:1040:30: warning: loop range is null, loop will not execute
|> |> 
|> |> The compiler has been built by a ia32->ia64 cross compiler, using CVS from
|> |> earlier today.
|> 
|> Update: the problem vanishes if I build the compiler with -O0.  I'm now
|> bootstrapping natively to see whether the problem reappears.

The problem reappeared with the stage2 compiler, so it seems like it got
miscompiled by the stage1 compiler.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

Re: New Ada build problem on ia64

[guerby]

> The problem reappeared with the stage2 compiler, so it seems like it got
> miscompiled by the stage1 compiler.

If you have an IA32 to IA64 compiler able to produce IA64 executables
from Ada sources, may be it would be better to start by compiling
simple Ada programs (at least simpler than the whole Ada compiler :)
to check if they work. I attached a standalone Ada program to start
with, if you're able to make this work, then I can feed you with more
tests (tomorrow Paris time, need to sleep right now :).

I have included a C wrapper if you have problem binding and/or linking
Ada with the cross tools.

-- 
Laurent Guerby <guerby@acm.org>

$ cat t1c.c
extern void _ada_t1 (void);
int main (void) {
  _ada_t1 ();
  return 0;
}

$ cat t1.adb
procedure T1 is
   type AC is access all Character;
   type Str is array (Positive range <>) of aliased Character;

   procedure P (X : in AC);
   pragma Import (C, P, "printf");

   LF : constant Character := Character'Val (10);
   Z : constant Character := Character'Val (0);
   S1 : aliased Str := "Hello" & LF & Z;
   S2 : aliased Str := "Loop" & LF & Z;
begin
   P (S1 (1)'Access);

   for I in 1 .. 2 loop
      P (S2 (1)'Access);
   end loop;
end T1;

$ gcc -c t1.adb
$ gcc -o t1 t1c.c t1.o
$ ./t1
Hello
Loop
Loop
$ 

Re: New Ada build problem on ia64

[Andreas Schwab]

<guerby@acm.org> writes:

|> > The problem reappeared with the stage2 compiler, so it seems like it got
|> > miscompiled by the stage1 compiler.
|> 
|> If you have an IA32 to IA64 compiler able to produce IA64 executables
|> from Ada sources, may be it would be better to start by compiling
|> simple Ada programs (at least simpler than the whole Ada compiler :)
|> to check if they work. I attached a standalone Ada program to start
|> with, if you're able to make this work, then I can feed you with more
|> tests (tomorrow Paris time, need to sleep right now :).

The sample works correctly, even with -O2.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

Re: New Ada build problem on ia64

[Andreas Schwab]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 879 bytes --]

<guerby@acm.org> writes:

|> > The problem reappeared with the stage2 compiler, so it seems like it got
|> > miscompiled by the stage1 compiler.
|> 
|> If you have an IA32 to IA64 compiler able to produce IA64 executables
|> from Ada sources, may be it would be better to start by compiling
|> simple Ada programs (at least simpler than the whole Ada compiler :)
|> to check if they work. I attached a standalone Ada program to start
|> with, if you're able to make this work, then I can feed you with more
|> tests (tomorrow Paris time, need to sleep right now :).

The sample works correctly, even with -O2.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

Re: New Ada build problem on ia64

[guerby]

> The problem reappeared with the stage2 compiler, so it seems like it got
> miscompiled by the stage1 compiler.

If you have an IA32 to IA64 compiler able to produce IA64 executables
from Ada sources, may be it would be better to start by compiling
simple Ada programs (at least simpler than the whole Ada compiler :)
to check if they work. I attached a standalone Ada program to start
with, if you're able to make this work, then I can feed you with more
tests (tomorrow Paris time, need to sleep right now :).

I have included a C wrapper if you have problem binding and/or linking
Ada with the cross tools.

-- 
Laurent Guerby <guerby@acm.org>

$ cat t1c.c
extern void _ada_t1 (void);
int main (void) {
  _ada_t1 ();
  return 0;
}

$ cat t1.adb
procedure T1 is
   type AC is access all Character;
   type Str is array (Positive range <>) of aliased Character;

   procedure P (X : in AC);
   pragma Import (C, P, "printf");

   LF : constant Character := Character'Val (10);
   Z : constant Character := Character'Val (0);
   S1 : aliased Str := "Hello" & LF & Z;
   S2 : aliased Str := "Loop" & LF & Z;
begin
   P (S1 (1)'Access);

   for I in 1 .. 2 loop
      P (S2 (1)'Access);
   end loop;
end T1;

$ gcc -c t1.adb
$ gcc -o t1 t1c.c t1.o
$ ./t1
Hello
Loop
Loop
$ 

Re: New Ada build problem on ia64

[Andreas Schwab]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1252 bytes --]

Andreas Schwab <schwab@suse.de> writes:

|> Andreas Schwab <schwab@suse.de> writes:
|> 
|> |> There is a new build problem with the Ada compiler on ia64:
|> |> 
|> |> gcc -c -g    -W -Wall -gnatpg -gnata -I- -I. -I../../../gcc/ada ../../../gcc/ada/ali.adb
|> |> ali.adb:52:27: warning: loop range is null, loop will not execute
|> |> ali.adb:473:18: warning: loop range is null, loop will not execute
|> |> ali.adb:637:19: warning: loop range is null, loop will not execute
|> |> ali.adb:757:27: warning: loop range is null, loop will not execute
|> |> ali.adb:1040:30: warning: loop range is null, loop will not execute
|> |> 
|> |> The compiler has been built by a ia32->ia64 cross compiler, using CVS from
|> |> earlier today.
|> 
|> Update: the problem vanishes if I build the compiler with -O0.  I'm now
|> bootstrapping natively to see whether the problem reappears.

The problem reappeared with the stage2 compiler, so it seems like it got
miscompiled by the stage1 compiler.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

Re: New Ada build problem on ia64

[Andreas Schwab]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1053 bytes --]

Andreas Schwab <schwab@suse.de> writes:

|> There is a new build problem with the Ada compiler on ia64:
|> 
|> gcc -c -g    -W -Wall -gnatpg -gnata -I- -I. -I../../../gcc/ada ../../../gcc/ada/ali.adb
|> ali.adb:52:27: warning: loop range is null, loop will not execute
|> ali.adb:473:18: warning: loop range is null, loop will not execute
|> ali.adb:637:19: warning: loop range is null, loop will not execute
|> ali.adb:757:27: warning: loop range is null, loop will not execute
|> ali.adb:1040:30: warning: loop range is null, loop will not execute
|> 
|> The compiler has been built by a ia32->ia64 cross compiler, using CVS from
|> earlier today.

Update: the problem vanishes if I build the compiler with -O0.  I'm now
bootstrapping natively to see whether the problem reappears.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

New Ada build problem on ia64

[Andreas Schwab]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 845 bytes --]

There is a new build problem with the Ada compiler on ia64:

gcc -c -g    -W -Wall -gnatpg -gnata -I- -I. -I../../../gcc/ada ../../../gcc/ada/ali.adb
ali.adb:52:27: warning: loop range is null, loop will not execute
ali.adb:473:18: warning: loop range is null, loop will not execute
ali.adb:637:19: warning: loop range is null, loop will not execute
ali.adb:757:27: warning: loop range is null, loop will not execute
ali.adb:1040:30: warning: loop range is null, loop will not execute

The compiler has been built by a ia32->ia64 cross compiler, using CVS from
earlier today.

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5