Re: [PATCH] rt-numa: Correct the comment of numa_initialize()

[John Kacur <jkacur@redhat.com>]

On Thu, 23 Dec 2021, Sebastian Andrzej Siewior wrote:

> On 2021-12-23 22:01:07 [+0800], Oscar Shiang wrote:
> > --- a/src/lib/rt-numa.c
> > +++ b/src/lib/rt-numa.c
> > @@ -15,7 +15,7 @@
> >  
> >  /*
> >   * numa_available() must be called before any other calls to the numa library
> > - * returns 0 if numa is available, or 1 if numa is not available
> > + * returns 1 if numa is available, or 0 if numa is not available
> >   */
> 
> To quote the man page:
> 
>        Before  any  other  calls  in this library can be used numa_available()
>        must be called. If it returns -1, all other functions in  this  library
>        are undefined.
> 
> Based on that, neither 0 nor 1 is defined.
> 
> Sebastian
> 

Right, but the numa_initialize function is meant to wrap that in such a 
way that we only call numa_available once, and then subsequent calls will 
return 1 (or true) for numa is available or 0 (false) if it is not 
available. This wrapper could probably be omitted, but it's supposed 
to make the code more readable.

This still isn't entirely cleaned up after this functionality 
was removed during the JSON stuff that mistakenly assumed numa is always 
available at runtime. (it might not be for example on some embedded 
platforms). In some case there might still be paths through the code that 
don't call non-numa versions of the functions when numa is not available.
It's on my list to fix.

John