Re: [PATCH] gatchat/gatppp.c: Add timeout handler in ppp_enter_phase

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1249 bytes --]

Hi Marko,

On 12/23/2015 02:52 AM, Marko Šulejić wrote:
>   * ppp_enter_phase calls ppp_dead if the PPP phase is PPP_PHASE_DEAD
>     && ppp->sta_pending == FALSE, but fails if sta_pending is never set
>     to FALSE. So a timeout handler was added to handle this case too.
> ---
>   gatchat/gatppp.c | 8 ++++++--
>   1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/gatchat/gatppp.c b/gatchat/gatppp.c
> index 5144084..f84535a 100644
> --- a/gatchat/gatppp.c
> +++ b/gatchat/gatppp.c
> @@ -361,8 +361,12 @@ static inline void ppp_enter_phase(GAtPPP *ppp, enum ppp_phase phase)
>   	DBG(ppp, "%d", phase);
>   	ppp->phase = phase;
>
> -	if (phase == PPP_PHASE_DEAD && ppp->sta_pending == FALSE)
> -		ppp->ppp_dead_source = g_idle_add(ppp_dead, ppp);
> +	if (phase == PPP_PHASE_DEAD) {
> +		if(ppp->sta_pending)
> +			ppp->ppp_dead_source = g_timeout_add(5000, ppp_dead, ppp);
> +		else
> +			ppp->ppp_dead_source = g_idle_add(ppp_dead, ppp);
> +	}

This seems wrong.  sta_pending is set if we are trying to transmit a 
terminate_ack.  If this fails, then the peer is likely severing the 
connection very shortly afterwards.

What is the sequence of events you're observing here?

Regards,
-Denis