What should confdefs.h do if CONFIGURE_INIT is not defined?

[Sebastian Huber]

On 17/12/2019 15:13, Joel Sherrill wrote:
> 
>     My approach would be to place all the define evaluations into a #ifdef
>     CONFIGURE_INIT guard, so that #include <rtems/confdefs.h> provides
>     nothing to the C compiler if CONFIGURE_INIT is not defined.
> 
> 
> Originally, the intent was that user code could use CONFIGURE_MAXIMUM_xxx
> if CONFIGURE_INIT was not defined. We likely don't do this much in the 
> tests but
> you could have a header file which did all the CONFIGURE_xxx defines, then
> include <rtems/confdefs.h>. Every application file could use the 
> CONFIGURE_xxx
> macros if you did an "appconf.h" type header. Then a "appconf.c" could 
> define
> CONFIGURE_INIT and include rtems/confdefs.h.

Ok, this is an interesting use case. I am not sure if it is a good 
application design if the global limits are widely visible and used.

I guess exposing the header includes is not necessary for this use case.

For this the general pattern would be:

#ifndef XXX
#define XXX default
#endif

#ifdef CONFIGURE_INIT
#if XXX needs configured data structures
#include <xxx.h>
XXX xxx_config;
#endif
#endif

> 
> Do we have all the rtems_xxx which fetch configuration data documented? 
> Those
> may largely cover the use case I described.

Yes, but you have to make a trade-off between compile-time and link-time 
constants.

> Now I generally recommend putting RTEMS configuration and an init task
> to start RTEMS services in a dedicated file and invoking main() to start
> the application. There are some of the rtems-examples which follow this 
> pattern.

Yes, this is in line with what I have seen in applications.

-- 
Sebastian Huber, embedded brains GmbH

Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone   : +49 89 189 47 41-16
Fax     : +49 89 189 47 41-09
E-Mail  : sebastian.huber at embedded-brains.de
PGP     : Public key available on request.

Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.