Re: can not compile recent svn nmap on windows vc 2010

[Henri Doreau <henri.doreau () gmail com>]

2013/5/12 Rob Nicholls <robert () robnicholls co uk>:
> Hi Henri,
>
> Thanks, I wasn't expecting such a quick response on a Sunday! My comments
> are also inline.
>
> > -----Original Message-----
> > From: Henri Doreau [mailto:henri.doreau () gmail com]
> > Sent: 12 May 2013 18:44
> > To: Rob Nicholls
> > Cc: Nmap dev; jaebum shin
> > Subject: Re: can not compile recent svn nmap on windows vc 2010
> >
> > Hi Rob,
> >
> > thanks for the report. My comments inline.
> >
> > I don't fully understand this error. What does the compiler complain
> > about? It
> > doesn't seem to me that any structure is re-defined... How did you fix it?
>
> Visual C++ seems to be complaining that line 71 of proxy_http.c contains the
> line:
>
> const struct proxy_spec ProxySpecHttp;
>
> But further down (from line 213) in the same file it's redefined as:
>
> const struct proxy_spec ProxySpecHttp = {
>   "http://";,
>   PROXY_TYPE_HTTP,
>   &ProxyOpsHttp,
> };
>
> I tried moving the redefined version further up to replace what's on line
> 71, but then it's defined before ProxyOptsHttp. I can't move the definition
> of ProxyOptsHttp up there as that contains references to
> proxy_http_node_new, proxy_http_node_delete and  proxy_http_handler, which
> aren't defined until afterwards. I can't define proxy_http_node_new before
> ProxyOptsHttp as that contains a reference to ProxySpecHttp (and I ended up
> going around in a circle, hence giving up - sorry!).
>
> It's a virtually identical issue for Socks4, which isn't a surprise as IIRC
> the Socks4 file was based on the HTTP proxy file.
>
> My "fix" was basically to revert all of the changes related to the nsock
> directory until before April 22nd (to remove all the proxy related commits,
> removing the offending code and all references to it).
Thanks, I see. I guess ranting against the compiler won't help... I'll
do what I did for the nsock engines:
  - declare functions prototypes
  - define the structure
  - define the (actual) functions.

It's not as terse and neat as the current way but it should fix it.
I'll try to do so tomorrow.

> > > That compiled version also appears to run okay, albeit with lots of
> > > SO_BROADCAST related errors with NSE:
> > >
> > > NSOCK ERROR [96.3770s] nsock_make_socket(): Setting of SO_BROADCAST
> > > failed (IOD #1): No error
> > >
> > > But I think that's a separate issue to the proxy related issues.
> > >
> > > Rob
> > I think so too. At least I've pushed a change (r30885) so that the error
> codes
> > are also properly reported on windows. Can you give it a try? IIRC the
> failure
> > of setsockopt(SO_BROADCAST) used to be ignored but I don't know why, I
> > should check and maybe re-add this behavior...
>
> 've manually edited the version of nsock_connect.c as I couldn't cleanly
> apply a patch based on your recent commit due to differences when I reverted
> the nsock changes so Nmap would compile (one line wasn't an exact match, so
> I wrapped the socket_errno() call with socket_strerror() like you've done
> for the rest). In particular, my SO_BROADCAST error line now looks like
> this:
>
>       nsock_log_error(ms, "Setting of SO_BROADCAST failed (IOD #%li): %s",
>                       iod->id, socket_strerror(socket_errno()));
>
> I'm not sure what I'm expected to see with this change, I'm guessing
> something informative if it works; but a quick test came back with the same
> as before, a few dozen lines saying "No error". E.g.:
>
> NSOCK ERROR [96.5160s] nsock_make_socket(): Setting of SO_BROADCAST failed
> (IOD #2): No error
You're right, the patch aimed to get an error message. Unfortunately I
can't test and from what I can read on MSDN the call looks correct...
Does a windows expert here know what's going on? Any idea why
setsockopt() returns -1 (SOCKET_ERROR is -1, right?) without setting
any error message?

Regards

--
Henri
_______________________________________________
Sent through the dev mailing list
http://nmap.org/mailman/listinfo/dev
Archived at http://seclists.org/nmap-dev/