Re: [Bug-glpk] Segmentation fault (core dumped) in table statement

[Andrew Makhorin]

Hi Xypron,

> the following code leads to a segmentation error in glpk-4.31:

>     set S;
>     param p {s in S} symbolic;
>     set P;
>     table plants IN 'iODBC' 'DSN=glpk;UID=glpk;PWD=gnu'
>       p['PLANTS'] : 
>       S <- [LOC1];
>     solve;
>     data;
>     param p :=
>       [PLANTS] 'SELECT PLANT, CAPA AS CAPACITY FROM transp_capa';
>     end;

> 11 lines were read
> Reading plants...
> error.mod:5: no value for S
> Model processing error
> Segmentation fault (core dumped)

>     set S;
>     param p {s in S} symbolic;
>     set P;
>     display p['PLANTS'];
>     solve;
>     data;
>     param p :=
>       [PLANTS] 'SELECT PLANT, CAPA AS CAPACITY FROM transp_capa';
>     end;

> gives

> 10 lines were read
> Display statement at line 4
> error.mod:4: no value for S
> Model processing error

> I would have expected successful error treatment for the table
> statement, too.

Thank you. I could reproduce the bug.

It appears in the following fragment:

         if (dca->arg != NULL)
         {  for (k = 1; k <= dca->na; k++)
               xfree(dca->arg[k]);
            xfree(dca->arg);
         }

(in the internal routine free_dca (glpmpl03.c), which is called
from mpl_terminate in case of error), because some dca->arg[k] is
not initialized (contains 0x3f3f3f3f written by xmalloc/xcalloc).

I need a time to fix the bug.


Andrew Makhorin