signal SIGSEGV: Segmentation fault - invalid memory reference

[Nat]

Hi,

I execute a program with gfortran in a MAC OS.

The program opens a file which contains 24 columns and 40000 lines, reads the file and writes in another file the sum of the 24 columns for each line.

For the first five lines the program works properly and then stops returning the error "Program received signal SIGSEGV: Segmentation fault - invalid memory reference"

Is there any way to solve this?

Nat

[mecej4]

To answer, we need to see at least the READ and Format statements, if
used, and the declarations of the variables that are read into. Have you
noted that for doing what you described the memory need only have space
for 25 REAL or INTEGER variables?

--
mecej...@nospam.invalid
(Replace four by 4, nospam by operamail, invalid by com,
and remove all underscores)

[Richard Maine]

Yes. I'm sure there is.

Oh. If you perhaps wanted more details than that for an answer, then
you'll have to give more pertinent details of the problem. Like actually
showing the code in question. Knowing the number of lines and columns in
the file seems about as useful as knowing what day of the week you ran
the program on.

--
Richard Maine
email: last name at domain . net
domain: summer-triangle

[Nat]

OK. Well, at fist I thought if I had to include any flag in the makefile. You can find below the code.Note that neither the output file is written properly. Thank you!!

integer :: ncellmax,ncell,write_ncell
real :: h1,h2,h3,h4,h5,h6,h7,h8,h9,h10,h11,h12,h13,h14,h15,h16,h17,h18,h19,h20,h21,h22,h23,h24
real, allocatable, dimension(:) :: sum_hours, vct_h1,vct_h2,vct_h3,vct_h4 ......etc

allocate (sum_hours(ncellmax))
allocate (vct_h1(ncellmax))......

open(100,file=fname)
do ncell=1, ncellmax
read(100,'(24e10.2)') h1,h2,h3,h4,h5,h6,h7,h8,h9,h10,h11,h12,h13,h14,h15,h16,h17,h18, &
h19,h20,h21,h22,h23,h24
vct_h1(ncell)=h1
vct_h2(ncell)=h2
.....etc to vct_h24

sum_hours(ncell)=vct_h1(ncell)+vct_h2(cell)+....

write(50,'(1I1,A,e10.2)') (write_ncell,tab,sum_hours(write_ncell),write_ncell=1,ncellmax)

[mecej4]

Your code snippets do not show any assignment of a value to 'ncellmax'.

[Nat]

sorry I forgot to include this
ncellmax=41400

[Gordon Sande]

The trivial solution is to only write correct programs. But I expect
your question
is more on how to find errors in existing programs.

The usual step one is to turn on all the debugging aids your compiler offers.
Segmentation errors suggest bad subscripts so the subscript checking aids
are certainly suggested.

To get help you need to show the code. Otherwise you will get whimsical
answers that are about as useless as your question.

[Nat]

I have also included the code. Please find below

integer :: ncellmax,ncell,write_ncell
real :: h1,h2,h3,h4,h5,h6,h7,h8,h9,h10,h11,h12,h13,h14,h15,h16,h17,h18,h19,h20,h21,h22,h23,h24
real, allocatable, dimension(:) :: sum_hours, vct_h1,vct_h2,vct_h3,vct_h4 ......etc

allocate (sum_hours(ncellmax))
allocate (vct_h1(ncellmax))......

open(100,file=fname)
do ncell=1, ncellmax
read(100,'(24e10.2)') h1,h2,h3,h4,h5,h6,h7,h8,h9,h10,h11,h12,h13,h14,h15,h16,h17,h18, &
h19,h20,h21,h22,h23,h24
vct_h1(ncell)=h1
vct_h2(ncell)=h2
.....etc to vct_h24

sum_hours(ncell)=vct_h1(ncell)+vct_h2(cell)+....

write(50,'(1I6,A,e10.2)') (write_ncell,tab,sum_hours(write_ncell),write_ncell=1,ncellmax)

[Nat]

On Friday, January 22, 2016 at 7:35:21 PM UTC+2, mecej4 wrote:

Aslo, I tried to include the flags -fbacktrace -fcheck=all and returned the error

Fortran runtime error: Index '1' of dimension 1 of array 'vct_h1' above upper bound of 0
This is referred to the line : vct_h1(ncell)=h1

I cannot understand the error. Any help with this?

[steve kargl]

The above is only *a part* of the code. It can not be compiled by any Fortran compiler.
Expecting others to fill in the missing code seems to be rather odd. As pointed out
by others, ncellmax has no value.

--
steve

[James Van Buskirk]

"Nat" wrote in message
news:387860c6-6941-4fd1...@googlegroups.com...

> Aslo, I tried to include the flags -fbacktrace -fcheck=all and returned
> the error

> Fortran runtime error: Index '1' of dimension 1 of array 'vct_h1' above
> upper bound of 0
> This is referred to the line : vct_h1(ncell)=h1

> I cannot understand the error. Any help with this?

Probably means you allocated vct_h1 before you set the value of
ncellmax.

[Nat]

Yes you are right, I had set the value ncellmax after the allocation statement.Thank you very much!

[baf]

Also, typo in the following assignment statement

sum_hours(ncell)=vct_h1(ncell)+vct_h2(cell)+....

where you reference "cell" rather than ncell.

It is hopeless for us to try and help you without the actual code being
given.

[Walt Brainerd]

Without really looking at the code very carefully, I can suggest
that if you need variables h1, h2, ..., h24, you probably
should be thinking of an array. Same with vct_h1, vect_h2, etc.

And look up the SUM intrinsic function. It should help you.

[robin....@gmail.com]

Yes, by showing the code.

[robin....@gmail.com]

cell? should be ncell.

But this is still not the code.
It doesn't even show the entire loop.

[robin....@gmail.com]

You need to show the code.

[campbel...@gmail.com]

This is a possible code that would improve things a bit.
Also what version of gFortran, as line 6 is OPEN which fails with some versions of gFortran 5.1.0


character :: fname*128 = 'name_of_data_file.txt'
character :: tab = char(9)
integer :: ncellmax, ncell, write_ncell, iostat
real :: h(24)
real, allocatable :: sum_hours(:), vct_h(:,:)
!
ncellmax = 41400
allocate (sum_hours(ncellmax))
allocate (vct_h(ncellmax,24))
!
open (100,file=fname)
do ncell = 1,ncellmax
read (100,'(24e10.2)',iostat=iostat) h
if ( iostat/=0 ) exit
vct_h(ncell,:) = h
sum_hours(ncell)= sum (h)
end do
ncellmax = ncell-1
!
! open (50 ??
write (50,'(I0,A,es10.2)') (write_ncell,tab,sum_hours(write_ncell),write_ncell=1,ncellmax)
!
end