Re: [PATCH v1] mfd: intel-lpss: Move linux/pm.h to the local header

[Lee Jones]

On Fri, 08 Feb 2019, Andy Shevchenko wrote:

> On Fri, Feb 01, 2019 at 03:08:17PM +0000, Lee Jones wrote:
> > On Fri, 01 Feb 2019, Andy Shevchenko wrote:
> > > On Fri, Feb 1, 2019 at 11:50 AM Lee Jones <lee.jones@xxxxxxxxxx> wrote:
> > > > On Thu, 24 Jan 2019, Andy Shevchenko wrote:
> > > >
> > > > > We now using a common macro for PM operations in Intel LPSS driver,
> > > > > and, since that macro relies on the definition and macro from linux/pm.h
> > > > > header file, it's logical to include it directly in intel-lpss.h.
> > > > > Otherwise it's a bit fragile and requires a proper ordering
> > > > > of header inclusion in C files.
> > > >
> > > > I don't agree with this.  File which use various headers should
> > > > explicitly include them.  Inheriting header files is non-optimal.
> > > >
> > > 
> > > intel-lpss.h _is_ using pm.h.
> > > I don't see a contradiction here.
> > 
> > Then it should be including in there *also*.
> 
> Why?
> 
> intel-lpss-*.c are not direct users of this header.

They're not?  That's is where the miscommunication lies then.

If a C-file isn't using the offerings of a headerfile, obviously there
is no need for the C-file to include it.

> > My point is that if drivers/mfd/intel-lpss-{acpi,pci}.c use the header
> > file, it should include it explicitly.
> 
> They are using it indirectly.

Indirectly is fine.

> As far as I know we don't, for example, include "asm/*.h" to each of our C-file
> because they are in _indirect_ use of.

-- 
Lee Jones [æçæ]
Linaro Services Technical Lead
Linaro.org â Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog