segmentation fault occurred while declaring fourth element of an array :-S

[Giampiero_Sindoni]

Hello all!

I am having a really strange issue with the following code: pratically if I declare a value of elements from REC(4) to REC(7) different from zero I have segmentation fault

....................

[bash]program exat
implicit none
real*8 start_value, stop_t, start_sec, stop_sec, interval, sat_id
real*8 theta_0, theta_punto, psi_0,  psi_punto,  phi_0, phi_punto
real*8 theta, psi, phi
real*8 zero, uno
real*8 delta_T
real*8 q1, q2,q3,q4
real*8 pigreco
real*8 REC(9), REC1(9)


pigreco=3.141593
zero=0.00
uno=1.50[/bash]

[bash]
close(10) 
close(11)
open(10,file='file1',status='unknown',form='binary')
open(11,file='file1.txt',status='unknown',form='formatted')
REC(1)=pigreco   
REC(2)=pigreco
REC(3)=pigreco
REC(4)=zero
REC(5)=zero
REC(6)=zero
REC(7)= zero
REC(8)=pigreco
REC(9)=pigreco
write(*,*) REC[/bash]

....This Work

but If I do something like:

[bash]REC(1)=pigreco   
REC(2)=pigreco
REC(3)=pigreco
REC(4)=zero
REC(5)=zero
REC(6)=zero
REC(7)= pigreco
REC(8)=pigreco
REC(9)=pigreco[/bash]

[bash]
[/bash]

[bash]I receive a segmentation fault during execution,[/bash]

[bash]This is really strange and it is driving me out of head.[/bash]

[bash]
[/bash]

[bash]Thanks in advance for your suggests[/bash]

[bash]
[/bash]

[Izaak_Beekman]

Is this the entire contents of your program? What hardware are you running on? What compiler version are you using?

[jimdempseyatthecove]

What is the name of your source file (in particular the file type/extension)?
x.for, x.f are by defaultcompiled as fixed form files. i.e. spaces on left of line are important
x.f90, x.f95 are free form by default.
You can change fixed/free form by way of option switch.

The program you copied to the forum has all the lines squished to the left margin.
In fixed form, this will cause problems.
Normally you will receive many error messages when this happens.
Sometimes you do not (but have problems with running of program).

Jim Dempsey

[Steven_L_Intel1]

Intel Fortran does not recognize .f95 as a Fortran file unless you ask it to. .f90 is the recommended and most widely supported file type for free-form source.

[Giampiero_Sindoni]

Thank you so much for all you suggestions,

this is the compiler version I have:

Intel Fortran Intel 64 Compiler Professional for applications running on Intel 64, Version 11.1 Build 20091130 Package ID: m_cprof_p_11.1.080

running on a

10.8.0 Darwin Kernel Version 10.8.0: Tue Jun 7 16:32:41 PDT 2011; root:xnu-1504.15.3~1/RELEASE_X86_64 x86_64 a Xserver with snow leopard

I compiled it usingifort -r8 -c EXAT.f90 ; ifort -oEXAT.x EXAT.o

I realized that I don't have the error while I assign value to the vector but when I try to write the vector in a binary file, here the program:

[bash]program exat
implicit none
real*8 start_value, stop_t, start_sec, stop_sec, interval, sat_id
real*8 theta_0, theta_punto, psi_0,  psi_punto,  phi_0, phi_punto
real*8 theta, psi, phi
real*8 pigreco, zero, uno
real*8 delta_T
real*8 q1, q2,q3,q4

real*8, dimension(9):: REC, REC1

start_value=00.00
stop_t=02.00
start_sec=00.00
stop_sec=00.00
interval=0
sat_id=1111111.00

pigreco=3.141593
theta_0= 103.783*180/pigreco
theta_punto=-0.005*180/pigreco
psi_0=-44.910*180/pigreco
psi_punto=0.0108*180/pigreco
phi_0=-159.108*180/pigreco
phi_punto=-0.254*180/pigreco
zero=0.00
uno=1.50


REC(1)=zero
REC(2)=sat_id
REC(3)=zero
REC(4)=zero
REC(5)=start_value
REC(6)=start_sec
REC(7)=stop_t
REC(8)=stop_sec
REC(9)=interval
write(*,*) REC

close(10) 
close(11)
open(10,file='file.bin',status='unknown',form='binary')
open(11,file='file.txt',status='unknown',form='formatted')
write(10)  REC
write(11, *)  rec

do delta_T=0,1,1

theta=theta_0 + theta_punto*delta_T
psi=psi_0+ psi_punto*delta_T
phi=phi_0+ phi_punto*delta_T

q1=cos(theta/2)*cos(psi/2)*cos(phi/2)+ sin(theta/2)*sin(psi/2)*sin(phi/2)

q2=sin(theta/2)*cos(psi/2)*cos(phi/2)- cos(theta/2)*sin(psi/2)*sin(phi/2)

q3=cos(theta/2)*sin(psi/2)*cos(phi/2)+ sin(theta/2)*cos(psi/2)*sin(phi/2)

q4=cos(theta/2)*cos(psi/2)*sin(phi/2)-sin(theta/2)*sin(psi/2)*cos(phi/2)

end do

close(10) 
close(11)

end
[/bash]

Now, if I declare the vector as

[bash]REC(1)=start_value
REC(2)=sat_id
REC(3)=zero
REC(4)= zero
REC(5)=zero
REC(6)=zero
REC(7)=zero
REC(8)=zero
REC(9)=interval[/bash]

The program runs smooth, but if I use something like:

[bash]REC(1)=zero
REC(2)=zero
REC(3)=zero
REC(4)= zero
REC(5)=sat_id
REC(6)=zero
REC(7)=zero
REC(8)=zero
REC(9)=zero[/bash]

Everything zero exept the 5th element I have segmentation error

I don't understand

Thank you so much

[jimdempseyatthecove]

An update might correct the problem you are observing.

Out of curiosity, with the failing set of initialization data for REC, following "REC(9) = zero", add "REC1(1) = sat_it" (any non-zero initializer).

Then see if the failure occurs.

Jim Dempsey

[Giampiero_Sindoni]

Quoting jimdempseyatthecove

An update might correct the problem you are observing.

Out of curiosity, with the failing set of initialization data for REC, following "REC(9) = zero", add "REC1(1) = sat_it" (any non-zero initializer).

Then see if the failure occurs.

Jim Dempsey

I tried your suggestion but nothing changed, I don't know how to update and I don't think it is feasable, I just want to create a 64bit array and store it in a binary file, it doesn't work for some error of me, the task doesn't sound so hard in general to hide a bug inside the compiler :)

Thanks

Ciao, Giampiero

[jimdempseyatthecove]

Giampiero,

In testing here:

Windows XP x32 (P4)IVF 11.0.066 Release and Debug builds no error (with either REC initializations)

Windows 7 x64 (Core i7 2600K) IVF 12.0.4.196 as x32 app Release and Debug no error (with either REC initializations), as x64 app Release and Debug no error (with either REC initializations).

IOW non reproducable.

One potential problem area comes to mind. If you compiled without processor detection and with declaring and advanced SSE/AVX requirement (one not supported by your hardware), then you might observe an error due to alignment or missing instruction.

What is your processor and what are your compilation options?

Jim Dempsey

[jimdempseyatthecove]

I noticed you are running on a Darwin kernel - i.e. Apple (I am on Windows).

Apple versions of compilers tend to produce PIC code (Position Independent Code). I am not sure if IVF does this when compiling on Mac (Darwin). You might search the ISN FORTRAN forums for PIC related problems. This may be an issue already discussed. Hopefully this is a problem that can be "fixed" with the appropriate compiler option and/or linker option.

Jim Dempsey