[tor-bugs] #3523 [Tor]: Allow controllers to post HS descriptors to the HSDir system

[Tor Bug Tracker & Wiki]

#3523: Allow controllers to post HS descriptors to the HSDir system
-------------------------+-------------------------------------------------
     Reporter:  rransom  |      Owner:
         Type:           |     Status:  needs_review
  enhancement            |  Milestone:  Tor: 0.2.7.x-final
     Priority:  normal   |    Version:  Tor: 0.2.7
    Component:  Tor      |   Keywords:  tor-hs, controller, nickm-review,
   Resolution:           |  027-triaged-1-in, SponsorS, SponsorR
Actual Points:           |  Parent ID:  #8993
       Points:  small-   |
  remaning               |
-------------------------+-------------------------------------------------
Changes (by yawning):

 * status:  needs_revision => needs_review


Comment:

 Because it would be a big shame if this didn't make it into
 0.2.7.1-alpha...

 Branch: https://github.com/Yawning/tor/compare/feature3523_027

 Changes:
  * Rebase/squash/and rewriting the commit log from [comment:23 comment
 23].

 fd11b505ea4fce0a493bcd587b7697c4463b7538

  * Make sure that 'make check-spaces' passes.

 308373e29a52311ccf2e6dd034b105b7ee6eeafa

  * Doesn't the documentation on directory_post_to_hs_dir need to change to
 match its new arguments?

 0b30b698a3449c4ca78a83e8f570868b5095e828

  * Somebody has to write a changes file here.

 b075684f85a3a6a5f721d793f89a99b2873dd8c9

  * Fixed `directory_post_to_hs_dir` to not double-free `hs_dirs`.

 b726a7574534f03e69947c6a3e604df580aef746

 Other things nickm brought up that were not done, and why:
 Replying to [comment:26 nickm]:
 >   * It would be pretty cool if the command parsing part of hspost
 handler were split into a separate function.

 Hmm, I don't think HSPOST is anywhere near as bad as HSFETCH, so I'm not
 going to do this, using the same excuse dgoulet did for HSFETCH.

 >   * Tests would be great, for as much of this as possible.

 See above.  Someone with a better understanding of the HS code and what
 has mock implementations available could do this.

 >   * Shouldn't we be checking the return value of read_escaped_data?

 We could check if it returns 0 I guess?
 `rend_parse_v2_service_descriptor` will do the right thing in that case
 though...

 Tentatively "needs-review"-ing this again.  The events on descriptor
 upload would be *really nice* to have since it will allow people using the
 new `ADD_ONION` stuff to know when their HS is ready.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/3523#comment:30>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online