Invalid read in mpz_powm_ui
marc.glisse at inria.fr
Thu Dec 6 15:57:22 CET 2012
On Thu, 6 Dec 2012, Torbjorn Granlund wrote:
> Zimmermann Paul <Paul.Zimmermann at loria.fr> writes:
> > PS: aren't the nightly builds of GMP tested against valgrind?
> > Nope. Perhaps we should. I ran it now, I and found 4 errors (including
> > yours). Two of these have not yet been analysed, but one of these
> > unanalysed problems is related to code not yet used by any documented
> > interface. The 4th error was harmless and due to how we do mpn_incr_u.
> > I implemented a simple workaround to silence valgrind.
> I fully agree you should!
> I gave it a shot, but neither on a AMD FX nor on an ARM v7 does it
> complete GMP tests in a default build. It hits unsupported instructions
> like vmovsd (for AMD64) and umaal (for ARM).
On a non-VEX amd64, all tests (default build with --enable-cxx) pass
inside valgrind. VEX-encoded instructions do seem to be new in
valgrind-3.8. -mno-sse2avx might help until then, or a -march for a
slightly older cpu.
> I'll wait for valgrind 3.8 to arrive to FreeBSD's ports or GNU/Linux's
> various package systems.
3.8.1 is in debian unstable, for instance.
More information about the gmp-bugs