On Sat, Dec 22, 2012 at 02:20:56PM -0500, Tom Lane wrote:
> I believe that we'd be better off doing something in postmaster.c to
> positively ensure that each session has a distinct seed value. Notice
> that BackendRun() already takes measures to ensure that's the case for
> the regular libc random() function; it seems like a reasonable extension
> to also worry about OpenSSL's PRNG.
> #ifdef USE_SSL
> if (EnableSSL)
> {
> struct timeval tv;
>
> gettimeofday(&tv, NULL);
> RAND_add(&tv, sizeof(tv), 0);
> }
> #endif
Take the caution one step further and make it independent of EnableSSL. In a
stock installation, a !EnableSSL postmaster will never seed its PRNG, and
there's no vulnerability. Add a shared_preload_libraries module that uses the
OpenSSL PRNG in its _PG_init(), and suddenly you're vulnerable again.
Other than that, looks good.
> We could perhaps also make this conditional on not EXEC_BACKEND, since
> the whole issue is moot if backends are launched by fork/exec.
True.