Re: get_max_open_descriptors() is more generous than Nsock.

["Luis MartinGarcia." <luis.mgarc () gmail com>]

On 12/19/2011 03:10 PM, Henri Doreau wrote:
> 2011/12/19 Luis MartinGarcia. <luis.mgarc () gmail com>:
> > Hi Henri,
> >
> > How complicated would it be to implement something like
> > libnetutil::get_max_open_descriptors() in Nsock? (that function is
> > basically just a wrapper for getrlimit()) Personally, I think it would
> > be useful to be able to determine the maximum number of descriptors that
> > Nsock can handle at runtime.
> >
> > I am not an expert on Nsocks internals but If that is easy to do, then I
> > think it is a much cleaner approach than providing getters to determine
> > the underlying engine and then try to guess the max number of
> > descriptors. What do you think?
> >
> > Best regards,
> >
> > Luis.
> Hi,
>
> rlimit gives limitations for the whole process and nsock has
> limitations (or not) per nsock_pool. Callers can of course open
> several pools in parallel (though I think it's not the case in nmap
> and its companion tools).
>
> A get_max_open_descriptors()-like function in nsock could return
> min(<engine limit>, <rlimit>) and let the users deal with multiple
> pools if they have some. That would be easy to implement. The fallback
> engine (select-based) will be the only one with such a limitation.
>
> I agree that functions directly returning the information would be
> much cleaner, that's what I was thinking about.

That sounds great, Henri.

I'm going to add a new item to Nping's to-do list so I can look into
this again the day the nsock-engines branch gets merged.

Thanks a lot! Best regards,

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