Re: [tor-bugs] #2183 [Tor Relay]: router_rebuild_descriptor(): Bug: Couldn't generate extra-info descriptor

["Tor Bug Tracker & Wiki" <torproject-admin@xxxxxxxxxxxxxx>]

#2183: router_rebuild_descriptor(): Bug: Couldn't generate extra-info descriptor
-----------------------+----------------------------------------------------
 Reporter:  Falo       |       Owner:  karsten            
     Type:  defect     |      Status:  needs_review       
 Priority:  normal     |   Milestone:                     
Component:  Tor Relay  |     Version:  Tor: 0.2.2.17-alpha
 Keywords:             |      Parent:                     
-----------------------+----------------------------------------------------

Comment(by rransom):

 Replying to [comment:14 karsten]:
 > Replying to [comment:12 rransom]:
 > > See
 [http://repo.or.cz/w/tor/rransom.git/shortlog/refs/heads/mitigate2183
 mitigate2183] ( !git://repo.or.cz/tor/rransom.git mitigate2183 ) for
 patches to make relays generate (and publish) router descriptors even if
 they cannot generate their extra-info descriptors.
 >
 > I think the approach to publish a router descriptor even if we cannot
 generate an extra-info descriptor is good.
 >
 > A quick comment to your patch: You could add a parameter
 has_extra_info_digest to router_dump_router_to_string() instead of
 iterating over all bytes of extra_info_digest to find out if it contains a
 non-zero value.

 That flag belongs in the signed_descriptor_t record.  I wanted to get a
 patch out quickly to publish the router descriptor even if no extra-info
 descriptor can be published.  Adding a has_extra_info_digest field to
 signed_descriptor_t will take more time both to write and to review.

 > Also, how about using a new variable for the extra-info-digest line
 instead of the "%s%s%s%s" part? Other than that, looks good.

 Using three conditional expressions in a row in the arguments to a printf-
 like function:
  * takes less programmer time,
  * takes less source-code space, and
  * is harder to screw up in a very harmful way (i.e. buffer overflow)
 than allocating a temporary string variable, whether on the stack (as a
 local char-array variable) or in the heap (using `tor_malloc_zero` and the
 associated free function), filling the temporary string variable, and
 feeding the temporary string to printf.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/2183#comment:18>
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