Re: Shutdown races...

Hi,

1) it's not a whole patch
2) it's more a hack as solution

Kannel _must_ work out of the box and if there are races then pls 
provide backtrace and debug log so that should be fixed. I need to go 
through my tree to see if I forgot anything to merge...

Thanks,
Alex

Wilfried Goesgens schrieb:
> On Wed, Apr 19, 2006 at 12:37:43PM +0200, Alexander Malysh wrote:
>> Hi,
>>
>> this patch is also wrong. see below...
>>> +        conn = gwlist_get(smsc_list, i);
>>> +		smscconn_wait_shutdown_finished(conn);
>> I don't know any function with such name...
>>
>>> +		info(0,"%li done. %li more to go.", i, j-i);
> 
> sorry, forgot that one. 
> 
> Wilfried G�sgens
> 
> 
> 
> ------------------------------------------------------------------------
> 
> --- a/gateway/gw/smscconn.c	2006-04-18 21:01:30.000000000 +0200
> +++ b/gateway/gw/smscconn.c	2006-04-05 18:16:59.000000000 +0200
> @@ -289,6 +288,19 @@
>  }
>  
>  
> +void smscconn_wait_shutdown_finished(SMSCConn *conn)
> +{
> +    gw_assert(conn != NULL);
> +    mutex_lock(conn->flow_mutex);
> +	while (conn->status != SMSCCONN_DEAD) {
> +		mutex_unlock(conn->flow_mutex);
> +		gwthread_sleep(0.1);
> +		mutex_lock(conn->flow_mutex);
> +    }
> +	mutex_unlock(conn->flow_mutex);
> +	return;
> +}
> +
>  void smscconn_shutdown(SMSCConn *conn, int finish_sending)
>  {