On Wed, Jul 07, 2021 at 10:35:56PM +0200, Thomas Monjalon wrote:

07/07/2021 22:23, Tyler Retzlaff:

quoted

On Thu, Jun 24, 2021 at 06:29:20PM +0200, Thomas Monjalon wrote:

quoted

24/06/2021 18:02, Tyler Retzlaff:

quoted

On Thu, Jun 24, 2021 at 08:54:49AM +0200, Thomas Monjalon wrote:

quoted

23/06/2021 20:26, Tyler Retzlaff:

quoted

// lib/eal/gcc/rte_toolchain_common.h
#define __rte_noreturn __attribute__((noreturn))

We should keep a macro in rte_common.h which triggers an explicit error

i think that's relatively trivial to do. rte_common.h could after
toolchain specific include do a simple test.

#ifndef __rte_no_return
#error no __rte_no_return defined for toolchain
#endif

No I was thinking of:

/** Doxygen comment for the attribute below */
#define __rte_no_return RTE_ATTR_NOT_SUPPORTED

when you suggested this i thought it was something already established.
i see now that's not the case.  would you mind elaborating a little with
a complete example of the toolchain specific override and the generic
include?  i'd just like to understand completely what you were
proposing so i can try it out.

Sorry I don't have time currently to elaborate,
and not sure how to implement it.
I may look at it next week if that's the way we want to go.

i think the community generally preferences not having obvious
conditional compilation on a per-macro basis so i think it is the way we
want to go.

i'll take a look at the existing uses of include/generic/foo.h and
<arch>/include/foo.h where the arch specific includes generic for
inspiration. but i look forward to seeing what you come up with when you
get a chance to focus on the topic.