Sergei Trofimovich slyfox@gentoo.org writes:
Overlap happens at a call of
ecc_modp_mul (ecc, y3, B, z1);which is basically
mpn_mul_n (y3, B, z1, m->size),
I'm adding asserts to these functions, and then I can easily reproduce.
I think linking with a gmp configured with --enable-assert should be enough to catch the problem in a platform-independent way?
Nikos, is it easy to add that gmp config to one of the gitlab ci jobs? Alternatively, we could add the no-overlap asserts to mini-gmp, to catch it in mini-gmp builds.
/* NOTE: mul and sqr needs 2*m->size limbs at rp */ void ecc_mod_mul (const struct ecc_modulo *m, mp_limb_t *rp, const mp_limb_t *ap, const mp_limb_t *bp) {
- /* NOTE: mpn_mul_n does not work correctly when source and destination overlap */
- assert (no_overlap (rp, 2 * m->size, ap, m->size));
- assert (no_overlap (rp, 2 * m->size, bp, m->size)); mpn_mul_n (rp, ap, bp, m->size); m->reduce (m, rp);
}
As you observed, this function doesn't allow in-place operation. It's less clear what the interface of the ecc_add_* and ecc_dup_* functions is. It needs to be determined if they should or shouldn't allow in-place operation, and if not, back that up with asserts.
Thanks for the bug report, /Niels