[Bug ada/114424] New: gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[Bug ada/114424] New: gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

            Bug ID: 114424
           Summary: gnat: Ada.Calendar.Clock crashes on 32bit
                    architectures with 64bit time_t
           Product: gcc
           Version: 13.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: ada
          Assignee: unassigned at gcc dot gnu.org
          Reporter: doko at gcc dot gnu.org
                CC: dkm at gcc dot gnu.org
  Target Milestone: ---

[forwarded from https://bugs.debian.org/1067453]

Ada packages randomly FTBFS on 32 bit architectures with
gprbuild: raised CONSTRAINT_ERROR : a-calend.adb:371 overflow check failed

The problem originates in the gcc-13 switch to time_t64.
gcc/ada/libgnat/s-os_prim__posix.adb is affected by two apparently
distinct issues.

* s-os_prim.adb allocates 3Long_Integer=3void*=3*32 bits for the
  timeval C struct, while 2*64bits = 2Long_Long_Integer are now needed.

  This issue affects other files, but is easy to find and fix.

* The switch breaks the call from Ada to the C gettimeofday function.

  Can anyone explain this, and ideally provide a real fix instead of
  the ugly work-around below?

cat > mycal.c <<EOF
#include <sys/time.h>
int mygettimeofday(struct timeval *restrict tv,
                   struct timezone *restrict tz) {
  return gettimeofday(tv, tz);
}
EOF

cat > foo.adb <<EOF
with Ada.Text_IO, System.OS_Primitives;
procedure Foo is
   -- On armhf, array(1..3) of Long_Integer gives random usec values.
   type timeval is array (1 .. 2) of Long_Long_Integer;
   procedure timeval_to_duration
     (T    : not null access timeval;
      sec  : not null access Long_Long_Integer;
      usec : not null access Long_Integer);
   pragma Import (C, timeval_to_duration, "__gnat_timeval_to_duration");
   sec    : aliased Long_Long_Integer;
   usec   : aliased Long_Integer;
   TV     : aliased timeval;
   function gettimeofday (Tv : access timeval;
      Tz : System.Address := System.Null_Address) return Integer;
   -- On armhf, a direct import of gettimeofday leads to random Tv values.
   pragma Import (C, gettimeofday, "mygettimeofday");
   Result : constant Integer := gettimeofday (TV'Access, System.Null_Address);
   SOP : constant Duration := System.OS_Primitives.Clock;
begin
   timeval_to_duration (TV'Access, sec'Access, usec'Access);
   Ada.Text_IO.Put_Line ("local               :" & Sec'Img & "." & Usec'Img);
   Ada.Text_IO.Put_Line ("System.OS_Primitives:" & SOP'Img);
end Foo;
EOF

gcc -c mycal.c -o mycal.o
gnatmake -gnat2022 foo -largs mycal.o
echo -n 'Machine             : '; uname -m
date   '+Expected            : %s'
./foo

Machine             : x86_64  (amd64 x86_64-linux-gnu)
Expected            : 1711035183
local               : 1711035183. 324545
System.OS_Primitives: 1711035183.324549000

Machine             : armv7l  (armhf arm-gnu-eabi)
Expected            : 1711035189
local               : 1711035189. 551691
System.OS_Primitives: 1113580361.675821568

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |DUPLICATE

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Basically a dup of bug 114065.

*** This bug has been marked as a duplicate of bug 114065 ***

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

Matthias Klose <doko at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|DUPLICATE                   |---

--- Comment #2 from Matthias Klose <doko at gcc dot gnu.org> ---
I don't see a duplicate here.

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |DUPLICATE
                 CC|                            |ebotcazou at gcc dot gnu.org

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Why don"t you use s-osprim__posix2008.adb if you use s-parame__posix2008.ads?

*** This bug has been marked as a duplicate of bug 114065 ***

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

--- Comment #4 from Matthias Klose <doko at gcc dot gnu.org> ---
are there any other files which should be used?

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

--- Comment #5 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Any file whose name contain "posix2008" should be used in lieu of the simple
"posix" variant, but that apparently makes only two of them.

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

--- Comment #6 from Matthias Klose <doko at gcc dot gnu.org> ---
still seen, when both posix2008 files are used.

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

--- Comment #7 from Matthias Klose <doko at gcc dot gnu.org> ---
something did change with Nicolas example:

date   '+Expected            : %s'; ./foo 
Expected            : 1711306699
local               : 1711306700. 1362
System.OS_Primitives:-6017917000.119493888

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

--- Comment #8 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
But s-osprim__posix2008.adb dies not use gettimeofday, does it ?

[Bug ada/114424] gnat: Ada.Calendar.Clock crashes on 32bit architectures with 64bit time_t

[doko at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114424

--- Comment #9 from Matthias Klose <doko at gcc dot gnu.org> ---
not sure if that's the right forum here, however:

 - in both Debian and Ubuntu, we're currently doing
   the time_t 64bit transition, for gnat that is
   moving from packages built with gnat-12 (32bit time_t)
   to packages built with gnat-13 (64bit time_t),
   on at least armhf (arm-linux-gnueabihf).

 - packages succeeding to build are adasockets,
   libxmlada, libgmpada, gprbuild.

 - next packages failing to build are: ahven, adacgi with that error
   gprbuild: raised CONSTRAINT_ERROR : a-calend.adb:371 overflow check failed

 - libflorist fails with
   posix-c.ads:876:07: error: size for "suseconds_t" too small, minimum allowed
is 64
posix-c.ads:876:07: error: size for "suseconds_t" too small, minimum allowed is
64
posix-c.ads:876:07: error: size for "suseconds_t" too small, minimum allowed is
64
posix-c.ads:876:07: error: size for "suseconds_t" too small, minimum allowed is
64

   compilation of posix_signals.ads failed
   compilation of posix-timers.adb failed
   compilation of posix_process_primitives.ads failed
   compilation of posix-timers-extensions.adb failed

  which is https://sourceware.org/bugzilla/show_bug.cgi?id=31510

 - unrelated, the libtemplates-parser build is timing out

for complete build logs see (enabling "good")
https://ubuntu-archive-team.ubuntu.com/transitions/html/gnat-13.html