Re: [PATCH v6 1/3] x86/platform/p2sb: New Primary to Sideband bridge support driver for Intel SOC's

[Paul Gortmaker]

[RE: [PATCH v6 1/3] x86/platform/p2sb: New Primary to Sideband bridge support driver for Intel SOC's] On 18/07/2016 (Mon 03:35) Tan, Jui Nee wrote:

[...]

> > > +
> > > +       return 0;
> > > +}
> > > +EXPORT_SYMBOL(p2sb_bar);
> > > +
> > > +MODULE_LICENSE("GPL");
> > 
> > ...the above is the only modular "use" that I can find.  So is the
> > tristate bogus?   Without a module_init and a module_exit I am
> > confused....
> > 
> > I just finished an audit of arch/x86 for bogus uses of module.h so I'd like to
> > ensure we don't add more.
> > 
> > Thanks,
> > Paul.
> > --
> > 
> P2SB could be "bool" instead of tristate. 
> My concern is if LPC_ICH built as module and not loaded, P2SB might
> consume memory when P2SB is not being used.
> What do you think? If that's ok for you, my next patch will be something like
> this:
> ...
> config P2SB
> 	bool
> 	depends on PCI

After seeing that this module isn't using an init/exit to allocate and
free resources, I'd initially thought it was linked into part of a
bigger module file, in which case we could just delete the token
MODULE_LICENSE tag (since it adds zero value) and then delete the
module.h include too (but probably use export.h instead).

However, re-examining the Makefile, I see this does look to be used as a
single file module, (but more like a library vs. an active driver), so
the MODULE_LICENSE will be required to keep the kernel happy -- i.e. it
is fine as it was and my apologies for adding confusion.

Thanks,
Paul.
--

> ...
> In p2sb.c, module.h header file will be removed as well.
> Hi Andy, please provide your comments and/or concerns if any. Thanks.
> > > --
> > > 1.9.1
> > >