[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #7571 [Tor]: Make AutomapHostsOnResolve work with IPv6
#7571: Make AutomapHostsOnResolve work with IPv6
-------------------------+--------------------------------------------------
Reporter: nickm | Owner:
Type: enhancement | Status: needs_review
Priority: normal | Milestone: Tor: 0.2.4.x-final
Component: Tor | Version:
Keywords: | Parent:
Points: | Actualpoints:
-------------------------+--------------------------------------------------
Comment(by andrea):
Other concern: letting 'attempts' be UINT32_MAX for picking random IPv6
addresses. Assuming almost all addresses are free is plausible given the
huge address space of IPv6, but to be sure the expectation value of the
number of attempts needed is small (i.e., O(log(n)) in the address space
size at worst), you also need to assume the random number generator is
well behaved, especially in terms of correlations between successive
bytes. With a linear congruential generator in particular, this can fail
horribly due to Marsaglia's Theorem, making the effective reachable
address space much smaller and hence much more densely occupied.
Now, I know lots and lots of other things would break if crypto_rand()
were that bad and I sure as hell hope it's never an LCG, but is
crypto_rand() actually *under our control*? I think if for some horrible,
messed-up reason it turns out to be a really terrible not-at-all-crypto-
grade PRNG, the manifestation of this probably should not be
addressmap_get_virtual_address() inexplicably hanging after an apparently-
small number of addresses have been mapped or something bizarre like that.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/7571#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs