On Mon, Mar 03, 2008 at 10:31:03AM +0100, Sam Ravnborg wrote:
> On Mon, Mar 03, 2008 at 10:17:14AM +0100, Ingo Molnar wrote:
> >
> > * Adrian Bunk <bunk@xxxxxxxxxx> wrote:
> >
> > > On Mon, Mar 03, 2008 at 09:57:20AM +0100, Ingo Molnar wrote:
> > > >
> > > > * Adrian Bunk <bunk@xxxxxxxxxx> wrote:
> > > >
> > > > > On Sat, Mar 01, 2008 at 09:26:41PM +0300, Alexey Starikovskiy wrote:
> > > > > > May I keep them inline?
> > > > >
> > > > > The problem with such manual inlines is that we force gcc to always
> > > > > inline them - and history has shown that functions grow without the
> > > > > "inline" being removed.
> > > >
> > > > what do you mean by "we force gcc to always inline them"?
> > >
> > > #define inline inline __attribute__((always_inline))
> > >
> > > > gcc is free to decide whether to inline or to not inline.
> > >
> > > Not with __attribute__((always_inline)).
> >
> > but that wasnt used in the code you patched:
> >
> > -inline int acpi_battery_present(struct acpi_battery *battery)
> > +static int acpi_battery_present(struct acpi_battery *battery)
>
> >From compiler-gcc.h:
>
> #define inline inline __attribute__((always_inline))
>
> So unless I am missing something obvious then each time we
> say inline to a function we require gcc to inline the function.
Exactly.
> It is my impression that today we only say inline if really needed
> and otherwise let gcc decide. So in almost all cases inlise should
> just be nuked?
The rule is:
- all static functions in headers should be marked inline
- no functions in .c files should be marked inline
For the latter there are a _few_ exceptions in hotpaths where doing so
brings measurable advantages.
But generally (and especially long-term) it's best to let gcc decide
which static functions in .c files should be inlined.
> Sam
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
