Re: SegFault in 2**111111111 calculation

[Kevin Ryde]

Pearu Peterson <address@hidden> writes:
>
>   mpz_pow_ui(b,b,EX);
> ...

Sylvain Pion <address@hidden> writes:
>
> I think it's a stack size problem.  If you set "limit stacksize
> 512M" in your shell, then it works.  Default is 8MB on linux, and I
> think I remember seeing in the GMP doc that it sometimes stores
> temporaries on the stack.

Yes, that looks like it.  mpz_pow_ui doesn't do anything special for
base==2, and will allocate a lot of memory with TMP_ALLOC, which is
alloca unless using --disable-alloca.  ulimit -s will reveal the
current stack size limit.

mpz_mul_2exp is the right function for 2**n, though perhaps pow_ui
should have a quick test for a power of 2 base.