<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<br>
<br>
<div class="moz-cite-prefix">On 7/23/15 1:57 PM, Cantor, Scott
wrote:<br>
</div>
<blockquote cite="mid:A660435A-596D-4EEB-8205-2C3D6BA533A1@osu.edu"
type="cite">
As far as people patching, there's the ability to set the flag in
Java to turn off the bug, whether they intended that to happen or
not.<br>
</blockquote>
<br>
Maybe. Hopefully that works for most people. On further
reflection, however: if the hostname -> IP addr conversion is
really unintentional and buried somewhere, then it may be that
setting the flag doesn't just give you the original hostname. It
might be doing a 2nd (reverse) resolution of the IP addr ->
hostname. If that is the case, and there are multiple PTR records
for the IP, there's no guarantee that the (single) hostname it picks
for getPeerHost() will be the original one. The double resolution
would be lossy, basically.<br>
<br>
I thought of that because we actually have several web servers with
that exact situation, due to name-based virtual hosting. And not
all of the domains have an https listener, so the shared
multi-domain cert doesn't have all of them as subject alt names.
This would break there. (Before someone asks: no SNI here b/c the
Apache/mod_ssl is too old).<br>
</body>
</html>