https://github.com/GrapheneOS/hardened_malloc/issues/89#issuecomment-521759557

> @adrelanos Maybe you can get the man-db maintainers to add `getrandom` to their seccomp-bpf filter since that's the only case of breakage that I know about at the moment. The `getrandom` system call is also used by various other libc hardening, although they might not do anything that would trigger it. The reason this calls `getrandom` after initialization is to reseed the per-size-class CSPRNGs occasionally as a minor security enhancement. If I had a way to detect that it would kill the process, I could just not do it, but it would be best if people just didn't blacklist `getrandom` when they're using `malloc` because hardened allocators should be allowed to use it.