Instead of a cryptic exception, let the user know if the HTTP target
isn't actually asking for WWW-Authenticate.
There are likely many more opportunities to catch errors, but this is
the most obvious.
Due to the vagaries of various Metasploit module description viewers, we
can't guarantee things like lists and bullet points render right.
Descriptions should avoid using these things.
1.) Now grabs clear-text user hint from XP systems in addition to
Win7/Win8 systems
2.) Changes output so it's no longer inline with hashes as not to
affect copy/paste of hashes output
3.) Adding alternate text in cases when no user hints are available
This solves the looping problem. Since the TCP connection wouldn't
necessarily close, we couldn't count on a 0-length recv, and the
connection was timing out. Changed it so send_tcp can take a block, in
which we do parsing. AXFR responses are sandwiched between SOA answers,
so when the second one is reached, the transfer is done.
This is also cleaner for existing code that uses send_tcp, since if no
block is passed, it just returns the first response and tears down the
connection, just like it used to.
Git commits must always, always, always stick to 50 characters or less
for formatting. This is the default convention followed by many existing
git projects, and many git tools assume these widths for display
formatting.
If new commits don't adhere to this, they should be rejected out of
hand with a brief explination and link to http://r-7.co/OGM3OK . It's
usually not that hard for the committer to either git commit --amend or
to merge commits over to a new branch with a new commit message.
* UserPasswordHint, a key that is used to store the users password
hint, can be easily decoded to clear-text to get the users hint
(Example: "My Favorite Color")
* Added decode_windows_hint() method to perform the decode process
* Added decoded hint output for hashdump.rb and smart_hashdump.rb
This required some changes to send_tcp: it now loops, reading as many
length-prefixed DNS responses as are available before the connection
closes. This shouldn't affect other uses of the function, since most
questions only have one response in answer.
The exploit was not working due to the user's root path causing
the EIP offset to change. To correct this, I was able to get
the server to disclose the root path in an error message (fixed in
5.67). I also radically refactored the exploit due to the feedback
I received from Juan Vazquez.
AXFR support in net-dns is broken. This fixes it, and makes the
requisite modifications to enum_dns module. Basic problem is that AXFR
responses consist of a chain of DNS replies, not a single reply with
multiple answers. Previously, only the first of these replies, the SOA
record, was returned. Also added some exception handling to avoid
problems like #483.
Sometimes a nameserver won't have an A record for its own name. Check
for this and fall back to using the system resolver via
Rex::Socket.gethostbyname. Example:
$ dig +short zonetransfer.me NS
ns12.zoneedit.com.
ns16.zoneedit.com.
$ dig +short @ns12.zoneedit.com ns12.zoneedit.com A
$ dig +short @ns16.zoneedit.com ns12.zoneedit.com A
$ dig +short @ns16.zoneedit.com ns16.zoneedit.com A
$
Also removed an extra A lookup that was unnecessary.
jvazquez-r7
Tod Beardsley
sinn3r
Jonathan Claudius
webstersprodigy
Daniel Miller
saint patrick
Matt Andreko
Tod Beardsley
Raphael Mudge
Samuel Huckins
h0ng10