Comment 412 for bug 417757
Revision history for this message
|
|
#412 |
© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
73416ba
(Get the code!)
(In reply to Tom Horsley from comment #77)
> (In reply to Phil Oester from comment #9)
> > But the question remains, WHY did the behavior change? Originally, glibc
> > DID use unique ports for the AAAA and A queries. From a "predictability"
> > perspective, that is a more secure approach, no? Similar to how ISNs are
> > now randomized in TCP.
> >
> > It seems many people's problems would be solved by going back to the
> > (arguably more secure) method of using distinct ports for the A and AAAA
> > queries.
>
> Since Ulrich is no longer around to defend to the death indefensible
> decisions, maybe it is time to just go ahead and put back the separate
> ports, the elimination of which caused all the problems in the first place.
The glibc community is consensus driven. Someone needs to write up a plan and drive it forward. The glibc team can do this, but this particular issue is lower on the overall priority list for stub resolver fixes. Principally we have no way to test this easily, so we're trying to build out our testing infrastructure to get coverage. In the past this was all tested by hand, and we can see how badly that turned out.