Re: [lwip-users] LCP Termination Request missing

[Prateek Jain]

Hi Ghislain,

 

    Just a suggestion.

PPP thread waits on sio_read. When you call pppCLose, It calls pppMainWakeup which in turn calls sio_read_abort. In case if you have not implemented sio_read_abort, it may give some problems. I feel with this function call, the PPP thread should no longer wait in sio_read. it should come out.

 

Prateek

 

On 11/9/05, Ghislain Mary <address@hidden> wrote:

Thanks for these explanations. Maybe that will help be to solve the
problem I encounter.

Ghislain


David Empson wrote:
> David Haas <address@hidden> wrote:
>
>>LCP is a layer on PPP.
>>
>>I don't know that code. Who is the expert on PPP? How do you tell the
>>link to go up and down?
>
>
> I'm no expert either, but I read the RFC a while ago and have a general
> understanding of it, at least up to the LCP layer.
>
> Assuming a simple PPP implementation (IP only), terminating a connection may
> require an exchange of IPCP packets (Internet Protocol Control Protocol) to
> negotiate termination of the IP connection. This should be followed by an
> LCP "terminate-request". The other side is supposed to respond with LCP
> "terminate-ack", and then both sides can drop the connection. The
> terminate-request should be retried if no terminate-ack is received.
>
> This should all be done in lwIP's PPP driver, which is below the IP layer.
>
>>From a quick glance at the code in src/netif/ppp, particularly fsm.c, lcp.c
> and ppp.c, it looks like the code is trying to send a TERMREQ (with retries)
> and expects to get a TERMACK response. lcp_close() calls fsm_close(), which
> sets the state to CLOSING and sends the first TERMREQ.
>
> According to Ghislain Mary's earlier message, it seemed that the TERMREQ
> wasn't being sent even if lcp_close() was called directly by application
> code.
>
> This suggests a possible problem relating to delays waiting for the LCP
> state to reach CLOSED. Perhaps the physical disconnect is happening
> prematurely.
>
> Looking at pppClose(), it triggers the LCP shutdown via a kill_link flag,
> then waits for LCP to reach PHASE_DEAD before returning, but only if there
> is no link status callback function. If that function has been provided,
> pppClose() will return immediately, and the LCP close won't have finished
> yet. There may be a separate mechanism to poll the state until LCP is DEAD,
> but I haven't looked for it.
>
> I'm not familiar enough with how lwIP's ppp interfaces to the serial driver
> (which must be application specific) but I expect details like physically
> hanging up the modem connection (dropping DTR or sending an ATH command) are
> the responsibility of the serial/modem driver.
>
>
>>Kieran Mansley wrote:
>>
>>
>>>On Tue, 2005-11-08 at 09:07 +0100, Ghislain Mary wrote:
>>>
>>>
>>>>Nobody's having an idea of how to tell lwIP to send an LCP Termination
>>>>Request?
>>>
>>>I don't know what LCP is, but lwIP shouldn't have to do anything unusual
>>>to transmit particular parts of a higher layer protocol - it's all just
>>>a stream of bits as far as lwIP is concerned.
>>>
>>>Kieran
>
>
>
>
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users
>


_______________________________________________
lwip-users mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/lwip-users