[Hamlib-developer] threaded thoughts !!

[Frank S. <vk...@ix...>]

Stephane Fillod wrote:

> > --- ft747.c   2000/11/25 21:49:34     1.21
> >    *
> > !  * 1. Rentrant code, handle static stuff  <--- ie: get rid of this stuff
> >    * 2. rationalise code, more helper functions.
> >
> > --- 262,317 ----
> >
> >   int ft747_set_freq(RIG *rig, freq_t freq) {
> > +   struct rig_state *rig_s;
> > +   struct ft747_priv_data *p;
> > +
> > +   static unsigned char bcd[] = { 0,0,0,0 }; /* set freq */
> 
>   static local variables (kindof "restricted" global variables) were
>   not well suited for reentrant code. And the p_cmd are not either, since
>   the struct would be shared in the data section. Most of the time,
>   reentrant mutexless solutions would make use of variables on the stack
>   (eventually alloca) or dynamic memory allocated on a per thread basis.
>   Anyway, reentrancy is not the primary focus right now :)

True, as my comments suggest, I need to handle this static 
stuff properly. ie: Remove it.

Can we not have instance data in priv_ and mutex locks
on API function level.

eg:

pthread_mutex_lock(&shared.mutex);
....do stuff
pthread_mutex_unlock(&shared.mutex);

Do we consider function call level (eg: rig_set..() ) the right
size of thread "granularity" ?

Its probably a good starting point.

Anyway, not an immediate problem, but
something to think about :-()

/Frank :-)

> _______________________________________________
> Hamlib-developer mailing list
> Ham...@li...
> http://lists.sourceforge.net/mailman/listinfo/hamlib-developer