How to send dynamic vector?

[dopiotr]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

Thanks!

 

[santosh]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

Without further code I can't really tell what your problem is. It may
not be a C problem but an incorrect use of MPI functions. You might do
better to ask in a forum dedicated to your MPI implementation.

<http://en.wikipedia.org/wiki/Message_Passing_Interface>
<http://www.mpi-forum.org/docs/>
<http://www.open-mpi.org/>
<http://www.personal.leeds.ac.uk/~bgy1mm/MPITutorial/MPIHome.html>

 

[CryptiqueGuy]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

Did u check the return value of malloc?
It might be NULL, due to failed memory allocation.

 

[dopiotr]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

Without further code I can't really tell what your problem is. It may
not be a C problem but an incorrect use of MPI functions. You might do
better to ask in a forum dedicated to your MPI implementation.

<http://en.wikipedia.org/wiki/Message_Passing_Interface>
<http://www.mpi-forum.org/docs/>
<http://www.open-mpi.org/>
<http://www.personal.leeds.ac.uk/~bgy1mm/MPITutorial/MPIHome.html>

Working wersion:
#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;

void main(int c, char *v[])
{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);
s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;
}
MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);
if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();
}

Why it is not working:
#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;
int main(int c, char *v[])
{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;
}
s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference
MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);
if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();
}

And why value of s_tmp[0] & s_tmp[1] is incorrect (in working
version).

 

[santosh]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

Without further code I can't really tell what your problem is. It may
not be a C problem but an incorrect use of MPI functions. You might
do better to ask in a forum dedicated to your MPI implementation.

<http://en.wikipedia.org/wiki/Message_Passing_Interface>
<http://www.mpi-forum.org/docs/>
<http://www.open-mpi.org/>
<http://www.personal.leeds.ac.uk/~bgy1mm/MPITutorial/MPIHome.html>

Working wersion:
#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;

void main(int c, char *v[])

Use int main

{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);
s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

In C the cast of malloc's return value is not needed. I'd write your
statements as:

suma_sigm = malloc(sizeof *suma_sigm * n_sigm);
s_tmp = malloc(sizeof *s_tmp * n_sigm);

Also be sure to check that malloc actually succeeded.

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;
}
MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);
if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();

Return a value here.

}

Why it is not working:
#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;
int main(int c, char *v[])
{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;

Well, here s_tmp is not yet pointing to any valid memory. You're writing
to memory you don't own or should not be writing to. Initialise s_tmp
to a valid object or an array of objects before using it.

}
s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

/Now/ you're doing it, but it's too late, unless rank was not zero.

MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);
if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();
}

And why value of s_tmp[0] & s_tmp[1] is incorrect (in working
version).

How should I know? What value did you expect? The obvious bug waiting to
happen in your second version is the use of the pointer s_tmp without
making it point to anything valid beforehand.

 

[Barry Schwarz]

snip

dopiotr wrote:

Hello Group,
int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.
But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);
It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

snip

Working wersion:

This is not a working version. It invokes undefined behavior and you
state it produces incorrect results.

#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;

void main(int c, char *v[])

main should return int.

{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);

Do not cast the return from malloc. It can never help and the only
effect would be to suppress a diagnostic you really want to see if it
applies to your code.

s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

You should always check that malloc succeeds.

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;
}

At this point, the elements of suma_sigm or s_tmp (but not both) have
been assigned values. The elements of the other array are
indeterminate (uninitialized).

MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);

Since you don't pass rank to this function, how is it supposed to
determine which of the arrays has valid values and which doesn't. Any
attempt by the function to evaluate an indeterminate element will
invoke undefined behavior.

if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();

}

Why it is not working:
#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;
int main(int c, char *v[])
{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;

At this point, s_tmp is still indeterminate; it does not yet point to
any memory you can access. s_tmp[j] does not exist. This invokes
undefined behavior.

}
s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

And if the previous undefined behavior were not sufficient, this would
replace the address that s_tmp was pointing to with the address of the
newly allocated memory making all the elements of the array
indeterminate again.

MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);
if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();

}

And why value of s_tmp[0] & s_tmp[1] is incorrect (in working
version).

By definition, a working version would produce correct results. What
is the vlaue of rank when you see these incorrect values? What is the
function MPI_Reduce supposed to do with the two arrays you pass?

 

[Malcolm McLean]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

Should be fine. If the call sends a static array it should work exactly the
same way with a dynamic buffer allocated with malloc(). This is a general
feature of C functions, not just MPI.

However you have to know the message length, and allocate another buffer, to
receive the message at the other end.

 

[Tor Rustad]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

Have you included <stdlib.h>? We recommend dropping the cast, and write
this:

int *Msg = malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

MPI related questions are off-topic here, try

comp.parallel.mpi

 

[Richard Heathfield]

Tor Rustad said:

Have you included <stdlib.h>? We recommend dropping the cast, and write
this:

int *Msg = malloc(sizeof(int)*size);

Who's this "we"? Dropping the cast, sure, but repeating the type? Blech.

 

[Richard]

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

Did u check the return value of malloc?
It might be NULL, due to failed memory allocation.

I wonder how many times people have come across this? I must admit that
malloc has never failed me under Linux or Windows...

 

[Tor Rustad]

Tor Rustad said:


Who's this "we"?

The FAQ

Dropping the cast, sure, but repeating the type? Blech.

Agreed, I didn't notice the sizeof(type).


int *msg = malloc(sizeof(*msg) * size);

 

[Tor Rustad]

[...]

Did u check the return value of malloc?
It might be NULL, due to failed memory allocation.

I wonder how many times people have come across this?

I can remember, doing my first matrix computations under DOS. The C
compiler had a size limit at 64 kb for objects. Not only could malloc
fail, but the max matrix size, was rather useless for doing scientific
double precision calculations.

> I must admit that malloc has never failed me under Linux or Windows...

With the default setting in Linux, malloc should never fail.

 

[Richard Heathfield]

Tor Rustad said:

The FAQ

Then the FAQ needs fixing. The above is poor style, and it is evident that
you agree...

Agreed, I didn't notice the sizeof(type).


int *msg = malloc(sizeof(*msg) * size);

Close enough. Personally, I'd use: int *msg = malloc(size * sizeof *msg);

 

[blmblm]

snip

dopiotr wrote:

Hello Group,

int Msg[size];
MPI_Send(Msg, size, MPI_INT, DEST, TAG, MPI_COMM_WORLD);
this is working fine.

But how to send Msg declared like this:
int *Msg=(int*)malloc(sizeof(int)*size);

It gives me
p4_error: interrupt SIGSEGV: 11
not_send: could not write to fd=4, errno = 32

snip

Working wersion:

This is not a working version. It invokes undefined behavior and you
state it produces incorrect results.

First -- agreed that this is off-topic and that comp.parallel.mpi
would be a more appropriate newsgroup. A couple of points for the
record, one here, one later:

When I correct the problems mentioned by others here, the OP's
program produces what appear to me to be correct results (using
OpenMPI on a Fedora 7 system).

#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;

void main(int c, char *v[])

main should return int.

{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);

Do not cast the return from malloc. It can never help and the only
effect would be to suppress a diagnostic you really want to see if it
applies to your code.

s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

You should always check that malloc succeeds.

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;
}

At this point, the elements of suma_sigm or s_tmp (but not both) have
been assigned values. The elements of the other array are
indeterminate (uninitialized).

MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);

Since you don't pass rank to this function, how is it supposed to
determine which of the arrays has valid values and which doesn't. Any
attempt by the function to evaluate an indeterminate element will
invoke undefined behavior.

Well, this *is* a library function, part of the same library that
initialized the local "rank" variable in the first place (via
the call to MPI_Comm_rank). Agreed that this isn't necessarily
obvious to readers of comp.lang.c, of course.

if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();

}

Why it is not working:
#include "mpi.h"
#include<stdio.h>
#include<stdlib.h>
uint n_sigm;
float *suma_sigm;
int main(int c, char *v[])
{
int size,rank;
int j;
float *s_tmp;
MPI_Init(&c,&v);
MPI_Comm_size(MPI_COMM_WORLD,&size);
MPI_Comm_rank(MPI_COMM_WORLD,&rank);
n_sigm=1000;

suma_sigm=(float *)malloc(sizeof(float)*n_sigm);

if(rank != 0)
{
for(j=0;j<n_sigm;j++)
suma_sigm[j]=rank*1000+j;
}
else
{
for(j=0;j<n_sigm;j++)
s_tmp[j]=0;

At this point, s_tmp is still indeterminate; it does not yet point to
any memory you can access. s_tmp[j] does not exist. This invokes
undefined behavior.

}
s_tmp=(float *)malloc(sizeof(float)*n_sigm);//difference

And if the previous undefined behavior were not sufficient, this would
replace the address that s_tmp was pointing to with the address of the
newly allocated memory making all the elements of the array
indeterminate again.

MPI_Reduce(suma_sigm,s_tmp ,n_sigm,MPI_FLOAT,MPI_SUM,
0,MPI_COMM_WORLD);
if( rank == 0 )
for(j=0;j<n_sigm;j++)
printf("%i %f\n",j,s_tmp[j]);
free(suma_sigm);
free(s_tmp);
MPI_Finalize();

}

And why value of s_tmp[0] & s_tmp[1] is incorrect (in working
version).

By definition, a working version would produce correct results. What
is the vlaue of rank when you see these incorrect values? What is the
function MPI_Reduce supposed to do with the two arrays you pass?

 

[Charlie Gordon]

"Technical skill is mastery of complexity, while creativity is mastery of
simplicity"

What a striking quote! (attributed to Erik Christopher Zeeman)