[tor-bugs] #23662 [Core Tor/Tor]: prop224: Service edge-case where it re-uploads descriptor with same rev counter

[Tor Bug Tracker & Wiki]

#23662: prop224: Service edge-case where it re-uploads descriptor with same rev
counter
-----------------------------+------------------------------------
 Reporter:  asn              |          Owner:  (none)
     Type:  defect           |         Status:  needs_revision
 Priority:  Medium           |      Milestone:  Tor: 0.3.2.x-final
Component:  Core Tor/Tor     |        Version:
 Severity:  Normal           |     Resolution:
 Keywords:  tor-hs, prop224  |  Actual Points:
Parent ID:                   |         Points:
 Reviewer:                   |        Sponsor:
-----------------------------+------------------------------------
Changes (by asn):

 * status:  needs_review => needs_revision


Comment:

 Hello, thanks for the patch. Here is a review:

 - The refactoring commit is indeed heavy and almost unreviewable if it's
 not split into three commits: one commit to create the stub funcs, one
 commit to just move code and one last commit that finalizes code. Right
 now it's far too hard for a reviewer (either me or nick). In the end we
 can squash all three commits together before merging.

 - Final commit looks nice. A question: Will a single descriptor upload
 time out, trigger an upload to all the HSDirs, or to just the hsdir that
 failed? IIUC the current patch will reupload to all HSDirs, right? Are we
 sure we want that? :X

 - Not a big fan of `tor_assert(dir_conn->hs_ident);`. Let's turn it into a
 BUG for less surprises.

 Given the big amount of changes required for this not-very-relevant bugfix
 (no actual reachability/security issue), I wouldn't be opposed to delaying
 this big fix to 0.3.3, if we have a minimal fix that will silence the
 warns.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23662#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online