Re: [PATCH 02/11] x86: sysfb: remove sysfb when probing real hw

[Ingo Molnar <mingo@kernel.org>]

* David Herrmann <dh.herrmann@gmail.com> wrote:

> Hi
> 
> On Thu, Jan 23, 2014 at 6:14 PM, Ingo Molnar <mingo@kernel.org> wrote:
> >
> > * David Herrmann <dh.herrmann@gmail.com> wrote:
> >
> >> >> +#ifdef CONFIG_X86_SYSFB
> >> >> +#  include <asm/sysfb.h>
> >> >> +#endif
> >> >
> >> > I guess a single space is sufficient?
> >> >
> >> > Better yet, I'd include sysfb.h unconditionally:
> >>
> >> Unconditionally won't work as only x86 has this header. [...]
> >
> > Well, in non-x86 code an #ifdef x86 looks ugly as well - but I guess
> > better than not building.
> >
> >> [...] If there's a way to place a dummy into asm-generic which is
> >> picked if arch/xy/include/asm/ doesn't have the header, let me know.
> >
> > Not that I know of.
> >
> >> But if I include it unconditionally without any fallback, this will
> >> fail on non-x86. And adding the header to all archs seems overkill.
> >
> > So why not drop the x86-ism and rename it to CONFIG_PLATFORM_SYSFB?
> > Some platforms configure it, some don't. Then the prototypes could
> > move into include/linux/sysfb.h or so and would be platform agnostic.
> 
> This is almost exactly what patch #6 does. [...]

Indeed - I never got so far down into the series.

> [...] But it also adds ~400 lines of kernel-doc and ~400 lines of 
> Documentation/. Given your remarks, I guess I will just split this 
> patch into code and docs, so we can just pick it up for stable in 
> case patch #1 does not fix all issues.

I have no objections to this form if it's fixed in a later patch and 
this one is easier to backport. I just missed that aspect.

Thanks,

	Ingo