On Fri, Jul 1, 2011 at 5:43 PM, Amit Nagal <helloin.amit@xxxxxxxxx> wrote:
> instead why not prepare_to_wait() be used once only like finish_wait()
> above and after schedule() call returns ,
> set the current state = TASK_INTERRUPTIBLE (before continue statement
> ) something like below :
>
> #define __wait_event_interruptible(wq, condition, ret) \
> do { \
> DEFINE_WAIT(__wait); \
> prepare_to_wait(&wq, &__wait, TASK_INTERRUPTIBLE); \
> \
> for (;;) { \
>
> if (condition) \
> break; \
> if (!signal_pending(current)) { \
> schedule(); \
> set_current_state(TASK_INTERRUPTIBLE) \
> continue; \
> } \
> ret = -ERESTARTSYS; \
> break; \
> } \
> finish_wait(&wq, &__wait); \
> } while (0)
>
> kindly share facts related to above .
>
> Regards
> Amit Nagal
>
Reframing the above question , what i want to ask is while waiting for
wake up events or signals in a loop ,
do we need to call prepare_to_wait in every iteration of the loop (case 1 ) or
we should call prepare_to_wait once only before entring the loop
and after schedule call returns set the current state =
TASK_INTERRUPTIBE (case 2) .
both cases are detailed below :
case 1 :
for(;;){
prepare_to_wait( &wq , &wait , TASK_INTERRUPTIBLE )
if(condition) break ;
if(! signal_pending ) {
schedule();
continue ;
}
ret = -ERESTARTSYS;
break;
}
finish_to_wait(&wq , &wait ) ;
case 2 :
prepare_to_wait( &wq , &wait , TASK_INTERRUPTIBLE )
for(;;){
if(condition) break ;
if(! signal_pending ) {
schedule();
set_current_state(TASK_INTERRUPTIBLE);
continue ;
}
ret = -ERESTARTSYS;
break;
}
finish_to_wait(&wq , &wait ) ;
please let me know whether implementing wait as in case 2 is right approach ?
Regards
Amit Nagal
_______________________________________________
Kernelnewbies mailing list
Kernelnewbies@xxxxxxxxxxxxxxxxx
http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
