beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Jorge D'Elia]

Dear GFortran developers,

With the beta gfortran 14.x.y versions we are noticing some runtime 
errors in a production code. 

One type of runtime errors is related to the concurrent use of the 
intrinsic "repeat" when the source code is compiled with the 
flag -march=native, please, see below:

$ cat test.f90
program test
  implicit none
  integer      , parameter :: iin = kind (1)
  integer (iin), parameter :: pp = 32
  character (len=pp) :: cc
  cc (1:pp) = repeat (" ",pp) 
  write (*,*)" cc : " // cc
end program test

$ gfortran --version
GNU Fortran (GCC) 14.0.0 20230808 (experimental)
Copyright (C) 2023 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

a) Compiling with the flag -mtune=native only, without -march=native, the test is ok:

$ gfortran -mtune=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o test.exe test.f90
$ test.exe 
  cc :                                 

b) However, compiling with the flag -march=native:

$ gfortran -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o test.exe test.f90
$ test.exe 

Program received signal SIGILL: Illegal instruction.

Backtrace for this error:
#0  0x14fae277fb1f in ???
#1  0x4011ad in ???
#2  0x401272 in ???
#3  0x14fae276a50f in ???
#4  0x14fae276a5c8 in ???
#5  0x4010c4 in ???
#6  0xffffffffffffffff in ???
Illegal instruction (core dumped)

On the other hand, compiling with the system version (GNU Fortran (GCC) 
12.3.1 20230508 (Red Hat 12.3.1-1)) or replacing the intrinsic repeat 
with:

do kk = 1, pp
  cc (kk:kk) = " "
end do

both tests are ok. The error occurs on any of Intel or AMD computers, 
e.g. in the present case:

$ lscpu 
Architecture:           x86_64
  CPU op-mode(s):       32-bit, 64-bit
  Address sizes:        46 bits physical, 48 bits virtual
  Byte Order:           Little Endian
CPU(s):                 6
  On-line CPU(s) list:  0-5
Vendor ID:              GenuineIntel
  Model name:           Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
    CPU family:         6
    Model:              45
    Thread(s) per core: 1
    Core(s) per socket: 6
    Socket(s):          1
    Stepping:           7
    BogoMIPS:           6400.22
Caches (sum of all):    
  L1d:                  192 KiB (6 instances)
  L1i:                  192 KiB (6 instances)
  L2:                   1.5 MiB (6 instances)
  L3:                   12 MiB (1 instance)
NUMA:                   
  NUMA node(s):         1
  NUMA node0 CPU(s):    0-5


By the way, I do not know if the -march=native flag or the intrinsic 
repeat would be of deprecated use (or not)...


Regards.
Jorge.
--
CIMEC (UNL-CONICET), cimec.conicet.gov.ar, www.cimec.org.ar
Predio CONICET-Santa Fe, Colec. Ruta Nac. 168, 
Paraje El Pozo, 3000, Santa Fe, ARGENTINA. 
Tel +54-342-4511594/95 ext 7062, fax: +54-342-4511169

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Jerry D]

On 8/11/23 5:34 AM, Jorge D'Elia via Fortran wrote:
> Dear GFortran developers,
> 
> With the beta gfortran 14.x.y versions we are noticing some runtime
> errors in a production code.
> 
> One type of runtime errors is related to the concurrent use of the
> intrinsic "repeat" when the source code is compiled with the
> flag -march=native, please, see below:
> 
> $ cat test.f90
> program test
>    implicit none
>    integer      , parameter :: iin = kind (1)
>    integer (iin), parameter :: pp = 32
>    character (len=pp) :: cc
>    cc (1:pp) = repeat (" ",pp)
>    write (*,*)" cc : " // cc
> end program test
> 
> $ gfortran --version
> GNU Fortran (GCC) 14.0.0 20230808 (experimental)
> Copyright (C) 2023 Free Software Foundation, Inc.
> This is free software; see the source for copying conditions.  There is NO
> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
> 
> a) Compiling with the flag -mtune=native only, without -march=native, the test is ok:
> 
> $ gfortran -mtune=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o test.exe test.f90
> $ test.exe
>    cc :
> 
> b) However, compiling with the flag -march=native:
> 
> $ gfortran -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o test.exe test.f90
> $ test.exe
> 
> Program received signal SIGILL: Illegal instruction.
> 
> Backtrace for this error:
> #0  0x14fae277fb1f in ???
> #1  0x4011ad in ???
> #2  0x401272 in ???
> #3  0x14fae276a50f in ???
> #4  0x14fae276a5c8 in ???
> #5  0x4010c4 in ???
> #6  0xffffffffffffffff in ???
> Illegal instruction (core dumped)
> 
> On the other hand, compiling with the system version (GNU Fortran (GCC)
> 12.3.1 20230508 (Red Hat 12.3.1-1)) or replacing the intrinsic repeat
> with:
> 
> do kk = 1, pp
>    cc (kk:kk) = " "
> end do
> 
> both tests are ok. The error occurs on any of Intel or AMD computers,
> e.g. in the present case:
> 
> $ lscpu
> Architecture:           x86_64
>    CPU op-mode(s):       32-bit, 64-bit
>    Address sizes:        46 bits physical, 48 bits virtual
>    Byte Order:           Little Endian
> CPU(s):                 6
>    On-line CPU(s) list:  0-5
> Vendor ID:              GenuineIntel
>    Model name:           Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
>      CPU family:         6
>      Model:              45
>      Thread(s) per core: 1
>      Core(s) per socket: 6
>      Socket(s):          1
>      Stepping:           7
>      BogoMIPS:           6400.22
> Caches (sum of all):
>    L1d:                  192 KiB (6 instances)
>    L1i:                  192 KiB (6 instances)
>    L2:                   1.5 MiB (6 instances)
>    L3:                   12 MiB (1 instance)
> NUMA:
>    NUMA node(s):         1
>    NUMA node0 CPU(s):    0-5
> 
> 
> By the way, I do not know if the -march=native flag or the intrinsic
> repeat would be of deprecated use (or not)...

Regardless, we should never segfault.  Thanks for the code example. We 
need to get a bug report opened on this.  I am on travel this morning, 
but if I have time i will do so this afternnon if someone else does not 
beat me to it.

There were some recent patches in this area IIRC.

Jerry

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Paul Richard Thomas]

I wonder why the development 14.0.0 doesn't exhibit this behaviour?

Could you please rerun with the compile options -g -fdump-tree-original .

The later should generate a file *.original with the content:

void test ()
{
  character(kind=1) cc[1:32];

  __builtin_memmove ((void *) &cc, (void *) &"
       "[1]{lb: 1 sz: 1}, 32);
  {
    struct __st_parameter_dt dt_parm.0;

    dt_parm.0.common.filename = &"test_repeat.f90"[1]{lb: 1 sz: 1};
    dt_parm.0.common.line = 7;
    dt_parm.0.common.flags = 128;
    dt_parm.0.common.unit = 6;
    _gfortran_st_write (&dt_parm.0);
    {
      character(kind=1) str.1[38];
      character(kind=1) str.2[41];

      _gfortran_concat_string (38, (character(kind=1)[1:] *) &str.1,
6, &" cc : "[1]{lb: 1 sz: 1}, 32, &cc);
      _gfortran_concat_string (41, (character(kind=1)[1:] *) &str.2,
38, (character(kind=1)[1:] *) &str.1, 3, &"end"[1]{lb: 1 sz: 1});
      _gfortran_transfer_character_write (&dt_parm.0,
(character(kind=1)[1:] *) &str.2, 41);
    }
    _gfortran_st_write_done (&dt_parm.0);
  }
}

Note that repeat is reduced to the builtin memmove with 32 spaces going to cc.

Regards

Paul

On Fri, 11 Aug 2023 at 17:21, Jerry D via Fortran <fortran@gcc.gnu.org> wrote:
>
> On 8/11/23 5:34 AM, Jorge D'Elia via Fortran wrote:
> > Dear GFortran developers,
> >
> > With the beta gfortran 14.x.y versions we are noticing some runtime
> > errors in a production code.
> >
> > One type of runtime errors is related to the concurrent use of the
> > intrinsic "repeat" when the source code is compiled with the
> > flag -march=native, please, see below:
> >
> > $ cat test.f90
> > program test
> >    implicit none
> >    integer      , parameter :: iin = kind (1)
> >    integer (iin), parameter :: pp = 32
> >    character (len=pp) :: cc
> >    cc (1:pp) = repeat (" ",pp)
> >    write (*,*)" cc : " // cc
> > end program test
> >
> > $ gfortran --version
> > GNU Fortran (GCC) 14.0.0 20230808 (experimental)
> > Copyright (C) 2023 Free Software Foundation, Inc.
> > This is free software; see the source for copying conditions.  There is NO
> > warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
> >
> > a) Compiling with the flag -mtune=native only, without -march=native, the test is ok:
> >
> > $ gfortran -mtune=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o test.exe test.f90
> > $ test.exe
> >    cc :
> >
> > b) However, compiling with the flag -march=native:
> >
> > $ gfortran -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o test.exe test.f90
> > $ test.exe
> >
> > Program received signal SIGILL: Illegal instruction.
> >
> > Backtrace for this error:
> > #0  0x14fae277fb1f in ???
> > #1  0x4011ad in ???
> > #2  0x401272 in ???
> > #3  0x14fae276a50f in ???
> > #4  0x14fae276a5c8 in ???
> > #5  0x4010c4 in ???
> > #6  0xffffffffffffffff in ???
> > Illegal instruction (core dumped)
> >
> > On the other hand, compiling with the system version (GNU Fortran (GCC)
> > 12.3.1 20230508 (Red Hat 12.3.1-1)) or replacing the intrinsic repeat
> > with:
> >
> > do kk = 1, pp
> >    cc (kk:kk) = " "
> > end do
> >
> > both tests are ok. The error occurs on any of Intel or AMD computers,
> > e.g. in the present case:
> >
> > $ lscpu
> > Architecture:           x86_64
> >    CPU op-mode(s):       32-bit, 64-bit
> >    Address sizes:        46 bits physical, 48 bits virtual
> >    Byte Order:           Little Endian
> > CPU(s):                 6
> >    On-line CPU(s) list:  0-5
> > Vendor ID:              GenuineIntel
> >    Model name:           Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
> >      CPU family:         6
> >      Model:              45
> >      Thread(s) per core: 1
> >      Core(s) per socket: 6
> >      Socket(s):          1
> >      Stepping:           7
> >      BogoMIPS:           6400.22
> > Caches (sum of all):
> >    L1d:                  192 KiB (6 instances)
> >    L1i:                  192 KiB (6 instances)
> >    L2:                   1.5 MiB (6 instances)
> >    L3:                   12 MiB (1 instance)
> > NUMA:
> >    NUMA node(s):         1
> >    NUMA node0 CPU(s):    0-5
> >
> >
> > By the way, I do not know if the -march=native flag or the intrinsic
> > repeat would be of deprecated use (or not)...
>
> Regardless, we should never segfault.  Thanks for the code example. We
> need to get a bug report opened on this.  I am on travel this morning,
> but if I have time i will do so this afternnon if someone else does not
> beat me to it.
>
> There were some recent patches in this area IIRC.
>
> Jerry
>

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Jorge D'Elia]

Dear Paul / Jerry,

----- Mensaje original -----
> De: "Paul Richard Thomas" 
> Para: "Jerry D" 
> CC: "Jorge D'Elia" , "Gfortran List" , "Jorge D'Elia"
> 
> Enviado: Viernes, 11 de Agosto 2023 13:55:43
> Asunto: Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL
>
> I wonder why the development 14.0.0 doesn't exhibit this behaviour?
> 
> Could you please rerun with the compile options -g -fdump-tree-original .
> 
> The later should generate a file *.original with the content:
> 
> void test ()
> {
>  character(kind=1) cc[1:32];
> 
>  __builtin_memmove ((void *) &cc, (void *) &"
>       "[1]{lb: 1 sz: 1}, 32);
>  {
>    struct __st_parameter_dt dt_parm.0;
> 
>    dt_parm.0.common.filename = &"test_repeat.f90"[1]{lb: 1 sz: 1};
>    dt_parm.0.common.line = 7;
>    dt_parm.0.common.flags = 128;
>    dt_parm.0.common.unit = 6;
>    _gfortran_st_write (&dt_parm.0);
>    {
>      character(kind=1) str.1[38];
>      character(kind=1) str.2[41];
> 
>      _gfortran_concat_string (38, (character(kind=1)[1:] *) &str.1,
> 6, &" cc : "[1]{lb: 1 sz: 1}, 32, &cc);
>      _gfortran_concat_string (41, (character(kind=1)[1:] *) &str.2,
> 38, (character(kind=1)[1:] *) &str.1, 3, &"end"[1]{lb: 1 sz: 1});
>      _gfortran_transfer_character_write (&dt_parm.0,
> (character(kind=1)[1:] *) &str.2, 41);
>    }
>    _gfortran_st_write_done (&dt_parm.0);
>  }
> }
> 
> Note that repeat is reduced to the builtin memmove with 32 spaces going to cc.


Ok, then using:

$ gfortran --version
GNU Fortran (GCC) 14.0.0 20230808 (experimental)

$ gfortran -mtune=native -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -g -fdump-tree-original -o test.exe test.f90

$ cat test.f90.005t.original

__attribute__((fn spec (". ")))
void test ()
{
  static logical(kind=4) is_recursive.0 = 0;
  character(kind=1) cc[1:32];

  try
    {
      if (is_recursive.0)
        {
          _gfortran_runtime_error_at (&"At line 1 of file test.f90"[1]{lb: 1 sz: 1}, &"Recursive call to nonrecursive procedure \'test\'"[1]{lb: 1 sz: 1});
        }
      is_recursive.0 = 1;
      __builtin_memmove ((void *) &cc, (void *) &"                                "[1]{lb: 1 sz: 1}, 32);
      {
        struct __st_parameter_dt dt_parm.1;

        dt_parm.1.common.filename = &"test.f90"[1]{lb: 1 sz: 1};
        dt_parm.1.common.line = 7;
        dt_parm.1.common.flags = 128;
        dt_parm.1.common.unit = 6;
        _gfortran_st_write (&dt_parm.1);
        {
          character(kind=1) str.2[38];

          _gfortran_concat_string (38, (character(kind=1)[1:] *) &str.2, 6, &" cc : "[1]{lb: 1 sz: 1}, 32, &cc);
          _gfortran_transfer_character_write (&dt_parm.1, (character(kind=1)[1:] *) &str.2, 38);
        }
        _gfortran_st_write_done (&dt_parm.1);
      }
    }
  finally
    {
      is_recursive.0 = 0;
    }
}


__attribute__((externally_visible))
integer(kind=4) main (integer(kind=4) argc, character(kind=1) * * argv)
{
  static integer(kind=4) options.3[7] = {1282, 1947, 0, 1, 1, 1, 31};

  _gfortran_set_args (argc, argv);
  _gfortran_set_options (7, &options.3[0]);
  test ();
  return 0;
}

$ test.exe 

Program received signal SIGILL: Illegal instruction.

Backtrace for this error:
#0  0x14aded839b1f in ???
#1  0x4011c2 in test
        at /home/jdelia/TESTS/test.f90:6
#2  0x4012ae in main
        at /home/jdelia/TESTS/test.f90:8
Illegal instruction (core dumped)


> Regards
> Paul
> 
> On Fri, 11 Aug 2023 at 17:21, Jerry D via Fortran  wrote:
>>
>> On 8/11/23 5:34 AM, Jorge D'Elia via Fortran wrote:
>> > Dear GFortran developers,
>> >
>> > With the beta gfortran 14.x.y versions we are noticing some runtime
>> > errors in a production code.
>> >
>> > One type of runtime errors is related to the concurrent use of the
>> > intrinsic "repeat" when the source code is compiled with the
>> > flag -march=native, please, see below:
>> >
>> > $ cat test.f90
>> > program test
>> >    implicit none
>> >    integer      , parameter :: iin = kind (1)
>> >    integer (iin), parameter :: pp = 32
>> >    character (len=pp) :: cc
>> >    cc (1:pp) = repeat (" ",pp)
>> >    write (*,*)" cc : " // cc
>> > end program test
>> >
>> > $ gfortran --version
>> > GNU Fortran (GCC) 14.0.0 20230808 (experimental)
>> > Copyright (C) 2023 Free Software Foundation, Inc.
>> > This is free software; see the source for copying conditions.  There is NO
>> > warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
>> >
>> > a) Compiling with the flag -mtune=native only, without -march=native, the test
>> > is ok:
>> >
>> > $ gfortran -mtune=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o
>> > test.exe test.f90
>> > $ test.exe
>> >    cc :
>> >
>> > b) However, compiling with the flag -march=native:
>> >
>> > $ gfortran -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -Og -o
>> > test.exe test.f90
>> > $ test.exe
>> >
>> > Program received signal SIGILL: Illegal instruction.
>> >
>> > Backtrace for this error:
>> > #0  0x14fae277fb1f in ???
>> > #1  0x4011ad in ???
>> > #2  0x401272 in ???
>> > #3  0x14fae276a50f in ???
>> > #4  0x14fae276a5c8 in ???
>> > #5  0x4010c4 in ???
>> > #6  0xffffffffffffffff in ???
>> > Illegal instruction (core dumped)
>> >
>> > On the other hand, compiling with the system version (GNU Fortran (GCC)
>> > 12.3.1 20230508 (Red Hat 12.3.1-1)) or replacing the intrinsic repeat
>> > with:
>> >
>> > do kk = 1, pp
>> >    cc (kk:kk) = " "
>> > end do
>> >
>> > both tests are ok. The error occurs on any of Intel or AMD computers,
>> > e.g. in the present case:
>> >
>> > $ lscpu
>> > Architecture:           x86_64
>> >    CPU op-mode(s):       32-bit, 64-bit
>> >    Address sizes:        46 bits physical, 48 bits virtual
>> >    Byte Order:           Little Endian
>> > CPU(s):                 6
>> >    On-line CPU(s) list:  0-5
>> > Vendor ID:              GenuineIntel
>> >    Model name:           Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
>> >      CPU family:         6
>> >      Model:              45
>> >      Thread(s) per core: 1
>> >      Core(s) per socket: 6
>> >      Socket(s):          1
>> >      Stepping:           7
>> >      BogoMIPS:           6400.22
>> > Caches (sum of all):
>> >    L1d:                  192 KiB (6 instances)
>> >    L1i:                  192 KiB (6 instances)
>> >    L2:                   1.5 MiB (6 instances)
>> >    L3:                   12 MiB (1 instance)
>> > NUMA:
>> >    NUMA node(s):         1
>> >    NUMA node0 CPU(s):    0-5
>> >
>> >
>> > By the way, I do not know if the -march=native flag or the intrinsic
>> > repeat would be of deprecated use (or not)...
>>
>> Regardless, we should never segfault.  

Ok.

>> Thanks for the code example. We need to get a bug report opened on this.  
>> I am on travel this morning, but if I have time i will do so this afternnon 
>> if someone else does not beat me to it.

Ok. Thanks.

>> There were some recent patches in this area IIRC.

Ok, although to avoid this problem, we are also using older beta versions.

>> Jerry

Regards,
Jorge.
--

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Paul Richard Thomas]

Hi Jorge,

> There were some recent patches in this area IIRC.
>
> Jerry

The tree dump is identical to mine, obtained with GNU Fortran (GCC)
14.0.0 20230809 (experimental), so I doubt that any recent patches are
responsible.

Being unable to reproduce the error, there is not much that I can do.

Sorry

Paul

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Jorge D'Elia]

Hi Paul,

----- Mensaje original -----
> De: "Paul Richard Thomas" <paul.richard.thomas@gmail.com>
> Para: "Jerry" <jvdelisle2@gmail.com>
> CC: "Jorge D'Elia" <jdelia@cimec.unl.edu.ar>, "Gfortran List" <fortran@gcc.gnu.org>, "Jorge D'Elia"
> <jdelia@intec.unl.edu.ar>
> Enviado: Sábado, 12 de Agosto 2023 3:56:44
> Asunto: Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL
>
> Hi Jorge,
> 
>> There were some recent patches in this area IIRC.
>>
>> Jerry
> 
> The tree dump is identical to mine, obtained with GNU Fortran (GCC)
> 14.0.0 20230809 (experimental), so I doubt that any recent patches are
> responsible.

It seems so. Furthermore, I have the feeling that the problem is outside 
GFortran, between the use of -mtune=native versus -march=native. For example,

$ gfortran --version
GNU Fortran (GCC) 14.0.0 20230808 (experimental)

$ ls -all
total 4
drwxr-xr-x. 1 jdelia jdelia   16 Aug 12 10:06 .
drwxr-xr-x. 1 jdelia jdelia 2782 Aug 12 10:05 ..
-rwxr-xr-x. 1 jdelia jdelia  223 Aug 11 08:28 test.f90

$ gfortran -mtune=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -g -fdump-tree-original -o test-mtune.exe test.f90
$ ls -all
total 36
drwxr-xr-x. 1 jdelia jdelia   110 Aug 12 10:17 .
drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
-rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
-rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
-rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune-test.f90.005t.original

$ mv test-mtune-test.f90.005t.original test-mtune.f90.005t.original

$ ls -all
total 36
drwxr-xr-x. 1 jdelia jdelia   100 Aug 12 10:19 .
drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
-rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
-rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
-rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune.f90.005t.original

$ gfortran -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -g -fdump-tree-original -o test-march.exe test.f90
$ ls -all
total 68
drwxr-xr-x. 1 jdelia jdelia   194 Aug 12 10:19 .
drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
-rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
-rwxr-xr-x. 1 jdelia jdelia 28064 Aug 12 10:19 test-march.exe
-rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:19 test-march-test.f90.005t.original
-rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
-rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune.f90.005t.original

$ mv test-march-test.f90.005t.original test-march.f90.005t.original

$ ls -all
total 68
drwxr-xr-x. 1 jdelia jdelia   184 Aug 12 10:19 .
drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
-rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
-rwxr-xr-x. 1 jdelia jdelia 28064 Aug 12 10:19 test-march.exe
-rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:19 test-march.f90.005t.original
-rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
-rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune.f90.005t.original

$ diff test-mtune.f90.005t.original test-march.f90.005t.original 
# (nothing is shown, thus both tree dumps ares identical).

$ test-mtune.exe 
  cc :                                 

$ test-march.exe 

Program received signal SIGILL: Illegal instruction.

Backtrace for this error:
#0  0x15268845fb1f in ???
#1  0x4011c2 in test
	at /home/jdelia/TEST/test.f90:6
#2  0x4012ae in main
	at /home/jdelia/TEST/test.f90:8
Illegal instruction (core dumped)

> Being unable to reproduce the error, there is not much that I can do.
> Sorry

Non problem. To avoid this error, for now we exclude the use 
of -march=native in our makefiles, and just use -mtune=native.

> Paul

Thanks for your time.

Regards.
Jorge.
--

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Steve Kargl]

On Sat, Aug 12, 2023 at 10:41:10AM -0300, Jorge D'Elia via Fortran wrote:
> 
> $ test-march.exe 
> 
> Program received signal SIGILL: Illegal instruction.
> 

This, to me, suggests that the backend is emitting a wrong
assembly language code.  I suspect that you'll need to 
compare the generated assembly code between a working 
gfortran and the bad gfortran.  It is also possible to get
a listing of the actually options that -march=native activates.
For example,

% gfcx -o z -O -march=native --verbose a.f90

It is the options that begin with -m that matter.  For my hardware
I see 108 such options.  With patients you can determine if one 
of these triggers the issues.

One can also scan gcc/gcc/ChangeLog for changes to config/i386/i386.md.
For example 

2023-08-07  Jan Beulich  <jbeulich@suse.com>

        * config/i386/i386.md (sse4_1_round<mode>2): Make
        "length_immediate" uniformly 1.
        * config/i386/mmx.md (mmx_pblendvb_v8qi): Likewise.
        (mmx_pblendvb_<mode>): Likewise.

There are at least 10 commits on 2023-08-07 to this file.

Finally, as a non-git user, you can always try git's bisection
feature to identify where things go left.

-- 
Steve

Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL

[Jerry D]

On 8/12/23 6:41 AM, Jorge D'Elia wrote:
> Hi Paul,
> 
> ----- Mensaje original -----
>> De: "Paul Richard Thomas" <paul.richard.thomas@gmail.com>
>> Para: "Jerry" <jvdelisle2@gmail.com>
>> CC: "Jorge D'Elia" <jdelia@cimec.unl.edu.ar>, "Gfortran List" <fortran@gcc.gnu.org>, "Jorge D'Elia"
>> <jdelia@intec.unl.edu.ar>
>> Enviado: Sábado, 12 de Agosto 2023 3:56:44
>> Asunto: Re: beta gfortran 14.x.y (using -march=native and intrinsic repeat) gives a signal SIGILL
>>
>> Hi Jorge,
>>
>>> There were some recent patches in this area IIRC.
>>>
>>> Jerry
>>
>> The tree dump is identical to mine, obtained with GNU Fortran (GCC)
>> 14.0.0 20230809 (experimental), so I doubt that any recent patches are
>> responsible.
> 
> It seems so. Furthermore, I have the feeling that the problem is outside
> GFortran, between the use of -mtune=native versus -march=native. For example,
> 
> $ gfortran --version
> GNU Fortran (GCC) 14.0.0 20230808 (experimental)
> 
> $ ls -all
> total 4
> drwxr-xr-x. 1 jdelia jdelia   16 Aug 12 10:06 .
> drwxr-xr-x. 1 jdelia jdelia 2782 Aug 12 10:05 ..
> -rwxr-xr-x. 1 jdelia jdelia  223 Aug 11 08:28 test.f90
> 
> $ gfortran -mtune=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -g -fdump-tree-original -o test-mtune.exe test.f90
> $ ls -all
> total 36
> drwxr-xr-x. 1 jdelia jdelia   110 Aug 12 10:17 .
> drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
> -rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
> -rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
> -rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune-test.f90.005t.original
> 
> $ mv test-mtune-test.f90.005t.original test-mtune.f90.005t.original
> 
> $ ls -all
> total 36
> drwxr-xr-x. 1 jdelia jdelia   100 Aug 12 10:19 .
> drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
> -rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
> -rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
> -rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune.f90.005t.original
> 
> $ gfortran -march=native -fcheck=all -std=f2018 -Wall -Werror -Wextra -g -fdump-tree-original -o test-march.exe test.f90
> $ ls -all
> total 68
> drwxr-xr-x. 1 jdelia jdelia   194 Aug 12 10:19 .
> drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
> -rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
> -rwxr-xr-x. 1 jdelia jdelia 28064 Aug 12 10:19 test-march.exe
> -rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:19 test-march-test.f90.005t.original
> -rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
> -rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune.f90.005t.original
> 
> $ mv test-march-test.f90.005t.original test-march.f90.005t.original
> 
> $ ls -all
> total 68
> drwxr-xr-x. 1 jdelia jdelia   184 Aug 12 10:19 .
> drwxr-xr-x. 1 jdelia jdelia  2782 Aug 12 10:05 ..
> -rwxr-xr-x. 1 jdelia jdelia   223 Aug 11 08:28 test.f90
> -rwxr-xr-x. 1 jdelia jdelia 28064 Aug 12 10:19 test-march.exe
> -rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:19 test-march.f90.005t.original
> -rwxr-xr-x. 1 jdelia jdelia 26856 Aug 12 10:17 test-mtune.exe
> -rw-r--r--. 1 jdelia jdelia  1405 Aug 12 10:17 test-mtune.f90.005t.original
> 
> $ diff test-mtune.f90.005t.original test-march.f90.005t.original
> # (nothing is shown, thus both tree dumps ares identical).
> 
> $ test-mtune.exe
>    cc :
> 
> $ test-march.exe
> 
> Program received signal SIGILL: Illegal instruction.
> 
> Backtrace for this error:
> #0  0x15268845fb1f in ???
> #1  0x4011c2 in test
> 	at /home/jdelia/TEST/test.f90:6
> #2  0x4012ae in main
> 	at /home/jdelia/TEST/test.f90:8
> Illegal instruction (core dumped)
> 
>> Being unable to reproduce the error, there is not much that I can do.
>> Sorry
> 
> Non problem. To avoid this error, for now we exclude the use
> of -march=native in our makefiles, and just use -mtune=native.
> 
>> Paul
> 
> Thanks for your time.
> 
> Regards.
> Jorge.
> --

I have not been able to reproduce it here.  I tried on:

gcc version 14.0.0 20230804 (experimental) (GCC)

and

gcc version 14.0.0 20230812 (experimental) (GCC)

I wonder if you caught a version in between changes happening?

$ lscpu
Architecture:            x86_64
   CPU op-mode(s):        32-bit, 64-bit
   Address sizes:         48 bits physical, 48 bits virtual
   Byte Order:            Little Endian

I started to draft a PR this morning, not sure.

I know it is a pain to do, can you build a later version of gfortran and 
see if your results are different?

Also, I have seen times where the libgfortran files were out of sync 
with the build because LD_LIBRARY_PATH was not pointing in the correct 
place.  So check these things.

Regards,

Jerry