Re: Socket troubles: Too many open files

[David Fifield <david () bamsoftware com>]

On Thu, Sep 20, 2012 at 10:08:33PM -0400, Patrick Donnelly wrote:
> On Thu, Sep 20, 2012 at 9:23 PM, David Fifield <david () bamsoftware com> wrote:
> > The descriptors created in the first four calls would be lost, and only
> > the last would be reclaimed by the close. I think this is a bug in Nsock
> > or in the NSE binding to Nsock.
>
> I think the problem was introduced when the nsock binding moved socket
> iod allocation from new_socket to connect (which was a good thing). I
> don't know if throwing a `if (nu->nsiod == NULL) nu->nsiod =
> nsi_new(nsp, NULL);' is the right fix...

Good thought, but I get this error:

NSOCK (0.2220s) TCP connection requested to 127.0.0.1:1000 (IOD #1) EID 8
NSOCK (0.2720s) Callback: CONNECT TIMEOUT for EID 8 [127.0.0.1:1000]
NSE: TCP 127.0.0.1:36140 > 127.0.0.1:1000 | CONNECT
NSOCK (0.2720s) TCP connection requested to 127.0.0.1:1000 (IOD #1) EID 16
Unable to update events for IOD #1: No such file or directory
QUITTING!
NSE: TCP 127.0.0.1:36141 > 127.0.0.1:1000 | CLOSE
NSOCK (0.2720s) nsi_delete (IOD #1)
Trying to delete NSI, but could not find 1 of the purportedly pending events on that IOD.

QUITTING!

What do you think about this patch? It moves some common code to a
close_internal function, then calls close_internal when connect is
called again on an already initialized socket. We don't call initialize
again, because that discards socket data like the timeout.

David Fifield

Attachment: nse-nsock-reconnect-close.patch
Description:

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