Re: [lwip-users] Configuring PPP Options in Version 2.0.0 Beta1

[Sylvain Rochet]

Hi Greg,


On Sat, May 07, 2016 at 02:27:15AM +0200, Sylvain Rochet wrote:
> On Wed, May 04, 2016 at 08:10:41PM +0000, Greg Smith wrote:
> > Hello.
> > 
> > I have been looking forward to the PPP enhancements made in lwIP 2.0.0 
> > Beta1 compared to 1.4.1.  I've added the new version to my project 
> > with success and have a few questions about configuring PPP.  I 
> > appreciate any help you can give.
> > 
> > 1) How can I setup a static IP address when _not_ running in PPP 
> > Server mode?  (I have been successful when setting the IP address as 
> > second argument of pppapi_listen() when running as a server.) However, 
> > when running as a "client" using pppapi_connect(), the only thing I've 
> > found to work is something like: pppapi_connect(pppPcb, 0); 
> > pppPcb->ipcp_wantoptions.ouraddr = usartPpp.ip_addr.addr;
> > 
> > That seems kludgy and wrong.  And, it does not work if the delay time 
> > (second argument of pppapi_connect()) is set >0.  I would like to be 
> > able to set a static IP address and have a delay time.
> > So how do I get something like:
> > 
> > // <-- function call to set static IP address here --> //
> > pppapi_connect(pppPcb, 10);
> 
> Well, you are right, you can't. You are actually the first one to ask. I 
> can't remember that far away but that wasn't possible in 1.4.1 as well 
> isn't it ?
> 
> 
> > 2) In lcp_init(), several PPP options are hard-coded, such as 
> > negotiating the async map and compression.  Is there a programmatic 
> > way to set these options at either compile-time or run-time without 
> > directly modifying the lcp_init() function?  If so, how?
> 
> Well, you can't, this is the same issue as 1).
> 
> 
> What worry me about both of them is that it is not possible to add them 
> without an API change, and we are in beta… that would have been possible 
> months ago but it's a bit too late now, you should have tried the Git 
> master branch :-)
> 
> The only way to achieve that is to add a timeframe when the user can 
> change the default configuration after PPP initialisation, this way:
> 
> ppp_clear(ppp);
> ... user configuration changes ...
> pppXX_connect(ppp_pcb *ppp, ...);
> 
> It changes the API because the user will be then responsible of clearing 
> the PPP machine state before re-connecting, which in my non-perfect 
> world means that most of them won't do it properly.
> 
> 
> The only configurable settings, using functions or macro if they exist 
> are currently our ppp_settings structure. That's actually the only way 
> to add your 1) and 2) without breaking the API, but we can't do that for 
> all options provided by the PPP stack, there are just too many 
> configurable values everywhere.
> 
> 
> What you could do, for now, is to remove ppp_clear(ppp); calls from 
> pppos_connect() and pppos_listen(), then don't forget to call it before 
> calling pppos_connect() and pppos_listen(). This way it adds the 
> necessary timeframe you need to manually change settings into all 
> protocols structures.

After thinking about it again, this is probably one proper way to 
achieve that. Nothing prevent us to reset the PPP state machine at 
session termination instead of session opening. This way we can get rid 
of ppp_clear() calls in pppXX_connect|listen functions and you will be 
then allowed to change the PPP configuration before calling 
pppXX_connect|listen. Then we only need to add thousands of macros to 
change PPP protocols configuration and it should do the job without 
changing the API and, and this is very important, without adding a lot 
of more code which cannot be compiled-out.


Sylvain

signature.asc
Description: Digital signature