[tor-bugs] #23588 [Core Tor/Tor]: Write fascist_firewall_choose_address_ls() and use it in hs_get_extend_info_from_lspecs()

[Tor Bug Tracker & Wiki]

#23588: Write fascist_firewall_choose_address_ls() and use it in
hs_get_extend_info_from_lspecs()
-------------------------------------------------+-------------------------
 Reporter:  teor                                 |          Owner:  neel
     Type:  defect                               |         Status:
                                                 |  needs_revision
 Priority:  Medium                               |      Milestone:  Tor:
                                                 |  0.3.5.x-final
Component:  Core Tor/Tor                         |        Version:
 Severity:  Normal                               |     Resolution:
 Keywords:  prop224, tor-hs, single-onion,       |  Actual Points:
  ipv6, 034-triage-20180328,                     |
  034-removed-20180328                           |
Parent ID:  #23493                               |         Points:  1
 Reviewer:  teor                                 |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by teor):

 We try not to duplicate code. I think your new patch is much simpler than
 fascist_firewall_choose_address_base(), because it doesn't actually choose
 addresses using the firewall rules on the client.

 Here's what I suggest:
 * If fascist_firewall_choose_address_base() doesn't work, then we should
 fix it.
 * If fascist_firewall_choose_address_ls() is calling
 fascist_firewall_choose_address_base() wrong, then we should fix the call.

 Also, I don't understand what you mean by:

     We compare
     addresses manually here as fascist_firewall_choose_address_base()
     doesn't always return addresses given from lspecs.

 Can you explain what behaviour you need from
 fascist_firewall_choose_address_base(), and what is happening instead?

 I think you might be expecting fascist_firewall_choose_address_base() to
 always return a valid address. But sometimes, there won't be any reachable
 addresses. If there aren't any reachable addresses, we need the 3-hop
 fallback code, which hasn't been written yet.

 But localhost (127.0.0.1 and ::1) should be reachable from the clients in
 the chutney networks. So I'm not sure why we're seeing this bug in
 chutney.

 Maybe some unit tests could help track it down.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23588#comment:30>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online