Re: [RFC PATCH v3 11/12] powerpc: Remove unreachable() from WARN_ON()

On Mon, Jul 04, 2022 at 12:34:08PM +0000, Christophe Leroy wrote:

Le 04/07/2022 à 13:45, Peter Zijlstra a écrit :

quoted

I'm somewhat confused; how is an empty STT_FUNC a valid construct on
Power?

So am I. It is likely not a valid construct, but that's what GCC seems 
to generate when you call annotate_unreachable().

It is a valid construct on (almost) all targets.  If the user chooses to
have executable code terminate in limbo, that is what the compiler will
do (and this can result in a code symbol with size 0).  Compare this to
data symbols with no size, the situation is quite similar.


Segher