[Bug fortran/50924] New: Attempt to allocate negative amount of memory. Possible integer overflow

[Bug fortran/50924] New: Attempt to allocate negative amount of memory. Possible integer overflow

[fwi at inducks dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50924

             Bug #: 50924
           Summary: Attempt to allocate negative amount of memory.
                    Possible integer overflow
    Classification: Unclassified
           Product: gcc
           Version: 4.4.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: fwi@inducks.org


When allocating a large array the progrem stops with:

Fortran runtime error: Attempt to allocate negative amount of memory. Possible
integer overflow

Test program:

  PROGRAM MAIN
  DOUBLE PRECISION, ALLOCATABLE :: E(:,:,:)
  INTEGER :: N
  N=645
  ALLOCATE(E(N,N,N))
  DEALLOCATE(E)
  END PROGRAM MAIN

With N=646:
$ gfortran -fbacktrace -g test.f90 && ./a.out 
Fortran runtime error: Attempt to allocate negative amount of memory. Possible
integer overflow

Backtrace for this error:
  + /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe6) [0xe2dbd6]

With N=645, the program works fine and takes 68.5% of the memory, so there
should be enough room for a 646^3 array (no other program is using a
significant amount of memory).

I believe the problem is that sizeof(double)=8 so the array size is:

8*(645^3) = 2 146 689 000     
8*(646^3) = 2 156 689 088

and: 2^31 = 2 147 483 648

So my guess is that the total amount of memory needed for the array is saved in
some integer, which overflows and become negative.

If that's of any use:

$ cat /proc/cpuinfo 
processor    : 0
vendor_id    : GenuineIntel
cpu family    : 6
model        : 23
model name    : Intel(R) Celeron(R) CPU        E3300  @ 2.50GHz
stepping    : 10
cpu MHz        : 2493.807
cache size    : 1024 KB
physical id    : 0
siblings    : 2
core id        : 0
cpu cores    : 2
apicid        : 0
initial apicid    : 0
fdiv_bug    : no
hlt_bug        : no
f00f_bug    : no
coma_bug    : no
fpu        : yes
fpu_exception    : yes
cpuid level    : 13
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat
pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc
arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl vmx est tm2 ssse3
cx16 xtpr pdcm xsave lahf_lm tpr_shadow vnmi flexpriority
bogomips    : 4987.61
clflush size    : 64
cache_alignment    : 64
address sizes    : 36 bits physical, 48 bits virtual
power management:

processor    : 1
vendor_id    : GenuineIntel
cpu family    : 6
model        : 23
model name    : Intel(R) Celeron(R) CPU        E3300  @ 2.50GHz
stepping    : 10
cpu MHz        : 2493.807
cache size    : 1024 KB
physical id    : 0
siblings    : 2
core id        : 1
cpu cores    : 2
apicid        : 1
initial apicid    : 1
fdiv_bug    : no
hlt_bug        : no
f00f_bug    : no
coma_bug    : no
fpu        : yes
fpu_exception    : yes
cpuid level    : 13
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat
pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc
arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl vmx est tm2 ssse3
cx16 xtpr pdcm xsave lahf_lm tpr_shadow vnmi flexpriority
bogomips    : 4987.44
clflush size    : 64
cache_alignment    : 64
address sizes    : 36 bits physical, 48 bits virtual
power management:

$ cat /etc/lsb-release 
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu 10.04.3 LTS"

Thanks.

[Bug fortran/50924] Attempt to allocate negative amount of memory. Possible integer overflow

[fwi at inducks dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50924

fwi at inducks dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fwi at inducks dot org

--- Comment #1 from fwi at inducks dot org 2011-10-30 18:52:10 UTC ---
NB: my system is 32-bit

[Bug fortran/50924] Attempt to allocate negative amount of memory. Possible integer overflow

[kargl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50924

kargl at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |kargl at gcc dot gnu.org

--- Comment #2 from kargl at gcc dot gnu.org 2011-10-30 18:55:09 UTC ---
(In reply to comment #1)
> NB: my system is 32-bit

What happens if you use 'INTEGER(8) N'?

You may simply be hitting an OS problem.  Your code
fails on i686-*-freebsd and runs fine on x86_64-*-freebsd.

The moral seems to be if you want to grab large chunks
of memory, get hardware that will it.

[Bug fortran/50924] Attempt to allocate negative amount of memory. Possible integer overflow

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50924

Richard Guenther <rguenth at gcc dot gnu.org> changed:

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

--- Comment #3 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-10-30 21:55:06 UTC ---
The maximum object size that can be allocated is half of the virtual
address-space.  Your object is too large (this is a middle-end restriction,
documented
in the C implementation defined section of the manual - the same restriction
applies to fortran).

[Bug fortran/50924] Attempt to allocate negative amount of memory. Possible integer overflow

[fwi at inducks dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50924

--- Comment #4 from fwi at inducks dot org 2011-10-31 08:43:23 UTC ---
Actually contrary to what I wrote my system is 64 bit (I clicked on "edit" on
the top right and somehow the system sent a message that I was starting to
write). 


Still when I run my program on a 64-bit machine with 16 Gb of RAM (under
CentOS), I could fill up 96% of the RAM with just one big array. I don't really
understand why such restriction is enforced.