On Fri, Feb 14, 2020 at 11:33:50AM +0100, Ruediger Pluem wrote: > On 02/14/2020 10:08 AM, Joe Orton wrote: > > I've been playing with UBSan[1] which catches undefined behaviour, found a > > couple of interesting things so far. > > > > One is with event, I messed with the line numbers but the error is: > > > > event.c:3620:13: runtime error: null pointer passed as argument 2, which is > > declared to never be null > > > > from the memcpy() in this code: > > https://github.com/apache/httpd/blob/trunk/server/mpm/event/event.c#L3619 > > > > new_ptr = (int *)apr_palloc(ap_pglobal, new_max * sizeof(int)); > > memcpy(new_ptr, retained->idle_spawn_rate, > > retained->mpm->num_buckets * sizeof(int)); > > I guess the above only does not crash because retained->mpm->num_buckets is 0 > at the same time. > But this is probably something we should not rely on with all memcpy > implementations.
Ahhhh, yes, that makes more sense now. Thanks a lot. ... > I guess there is no need for two cases here. We should only avoid the call to > memcpy > if retained->idle_spawn_rate is NULL. The initialization happens then in the > block starting > at line 3624. Gotcha. Done in r1874011. Regards, Joe
