[R] Problems with .C - [Ff]ree()?

[Prof Brian Ripley]

On Mon, 14 Jan 2002, Thomas Lumley wrote:

> On Tue, 15 Jan 2002, Mark White wrote:
> > It was indeed 1.3.something.  I agree that it sounds like a
> > memory leak in my code, but... I just tried compiling the
> > following C library routine (based on the arguments of a
> > file-reading function that's causing problems):
> >
> >     void R_test (char **filename, int *dim, Rcomplex *matrix)
> >     {
> >         double *ptr;
> >         ptr = (double*) malloc(sizeof(double) * 25600);
> >         free(ptr);
> >     }
> >
> > which, when called like this from within my R function
> >
> >     .C("R_test",as.character(filename),as.integer(adimi),
> >         as.complex(im))
> >
> > produces a segfault in Rf_duplicate soon after returning for
> > the somethingth time.  If I make replace the last argument
> > with something like complex(5) - rather smaller than im  - I
> > don't get an crash.  Likewise I don't get a crash if I just
> > repeatedly call R_test from a simple command-line for(){}.
> >
> > Is there something horribly wrong with my use of .C above?
> >
>
> No.  I tried
> -----complete program: bar.c ----
> #include <R.h>
> #include <stdlib.h>
>
>   void R_test (char **filename, int *dim, Rcomplex *matrix)
>     {
>         double *ptr;
>         ptr = (double *) malloc(sizeof(double) * 25600);
>         free(ptr);
>     }
> -----------------------
> then
>   R SHLIB bar.c
> to compile and then in R
>   dyn.load("bar.so")
>   repeat({
>    print(.C("R_test",as.character("filename"), integer(100000),
>     complex(100000))[[1]])
>   })
>
> with no problems. (I also tried without <stdlib.h> and without the
> unnecessary cast to double * on the malloc, for the sake of variety).
>
> I also tried it with
>  .C("R_test", as.character("filename"),as.integer(rpois(10000,10)),
>      log(as.complex(rnorm(10000))))[[1]]
> in case the values mattered. Still no problems.
>
> I could see this happening if you somehow had two different malloc()s
> around, but I can't see how you could.

It's a common symptom on some systems of writing off the end of an
allocated block, and that is normally R-version-sensitive.  OTOH, I use
Calloc/Free all the time and on correct code never see this.  I was just
wondering about the use of Rcomplex.  That's a structure and unusual in R.
.C is copying arguments to and from the call to the C code, and I'm
wondering if the compiler used here is handling arrays of structures
correctly.  There are lot of broken gcc's about in Linux distributions
....

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._