Re: GMP kernel and stack abuse

[Bill Allombert <allomber@math.u-bordeaux.fr>]

On Mon, Apr 07, 2003 at 04:54:51PM -0700, Ilya Zakharevich wrote:
> On Mon, Apr 07, 2003 at 11:50:51PM +0200, Bill Allombert wrote:
> > > >   Using the GMP kernel, my session dies on a huge division!
> > > > 
> > > >    This is a general problem in GMP, which may allocate a huge amount of
> > > >    memory in the process stack space (using alloca). The best solution so far
> > > >    is to increase the maximum size of the process stack segment before
> > > >    starting gp from the shell. Alternatively, you might wish to configure GMP
> > > >    with
> > > > 
> > > >      configure --enable-alloca=malloc-noreentrant
> > > > 
> > > >    but this will slow down GMP.
> > > 
> > > We *know* the stack limit.  Why not use this knowledge to substitute
> > > our alloca()?
> > 
> > Does not it require recompiling libgmp ?
> 
> So do it.  ;-)

... then I am afraid we cannot really avoid pointing at the FAQ. Most
people just use whatever libgmp is provided by their software
distribution/sysadmin and the default is to use the real alloca.

The FAQ already suggest to recompile libgmp with
configure --enable-alloca=malloc-noreentrant

We could provide a way to recompile GMP with a TMP_ALLOC interface
that use the PARI stack instead of the system stack, but it will
be a burden  for most people to have to recompile a modified GMP version.

> > The stack overflow here is of the order of tens of megabytes.
> 
> So it is not hard to detect.  ;-)

PARI is able to catch small stack overflow and output a deep recursion
error. Is it really possible to catch large stack overflow?

Cheers,
Bill..