[Devel] register_script_cb problems?

Bogdan-Andrei Iancu bogdan at voice-system.ro
Mon Jun 20 09:52:53 CEST 2005


Hi Ingo,

indeed there is small bug in the code - in second block, the "if" must 
test against "RPL_TYPE_CB" flag - looks like copy and paste error :(.

regarding the error message you get: note that in head version the 
callbacks were split in two type: callback for request and for replies 
(before all were fixed and called for both types of  messages).
When you register a callback you have to specify two flags: (REQ_TYPE_CB 
or RPL_TYPE_CB - both can pe set) and (PRE_SCRIPT_CB or POST_SCRIPT_CB - 
only one can e set).

most probably your module doesn't set the xxx_TYPE_CB flag.

thanks for reporting, the bug is fixed on cvs head.

regards,
bogdan

Ingo Wolfsberger wrote:

>Hi
>
>When I start my HEAD openser proxy, I get:
>BUG:register_script_cb: REQUEST or REPLY type not specified
>
>Function modules/sl/sl.c calls it wrong?
>register_script_cb( sl_filter_ACK, PRE_SCRIPT_CB, 0 );
>
>And in script_cb.c:
>register_scrbt_cb function:
>        if (type&REQ_TYPE_CB) {
>                /* callback for request script */
>                if (type&PRE_SCRIPT_CB) {
>                        if (add_callback( &pre_req_cb, f, param)<0)
>                                goto add_error;
>                } else if (type&POST_SCRIPT_CB) {
>                        if (add_callback( &post_req_cb, f, param)<0)
>                                goto add_error;
>                }
>        }
>        if (type&REQ_TYPE_CB) {
>                /* callback (also) for reply script */
>                if (type&PRE_SCRIPT_CB) {
>                        if (add_callback( &pre_rpl_cb, f, param)<0)
>                                goto add_error;
>                } else if (type&POST_SCRIPT_CB) {
>                        if (add_callback( &post_rpl_cb, f, param)<0)
>                                goto add_error;
>                }
>        }
>
>That are 2 times the same if?
>Or should the second be:
>if(type&RPL_TYPE_CB) {
>?
>
>bye,
>   Ingo
>
>  
>




More information about the Devel mailing list