Re: [PATCH v5 29/29] nios2: Build infrastructure

[Thomas Gleixner]

On Thu, 30 Oct 2014, Ley Foon Tan wrote:

> On Tue, Oct 28, 2014 at 9:38 PM, Arnd Bergmann <arnd@xxxxxxxx> wrote:
> > On Friday 24 October 2014 16:24:31 Ley Foon Tan wrote:
> >> +DECLARE_EXPORT(__gcc_bcmp);
> >> +DECLARE_EXPORT(__divdi3);
> >> +DECLARE_EXPORT(__divsi3);
> >> +DECLARE_EXPORT(__moddi3);
> >> +DECLARE_EXPORT(__modsi3);
> >> +DECLARE_EXPORT(__udivdi3);
> >> +DECLARE_EXPORT(__udivmoddi4);
> >> +DECLARE_EXPORT(__udivsi3);
> >> +DECLARE_EXPORT(__umoddi3);
> >> +DECLARE_EXPORT(__umodsi3);
> >> +DECLARE_EXPORT(__muldi3);
> >>
> >
> > I don't remember what all of these are, but at least some of them seem
> > to be concerned with 64-bit division. By convention, we don't provide those
> > in the Linux kernel but instead require all code to use the do_div()
> > macro instead.
> Do you mean kernel doesn't support 64 bit division result for a 32-bit
> architecture? Because do_div() only support 32-bit division result for
> 32-bit architecture.
> Compiler will auto generate call to __divdi3() for the following
> example code. This means loadable module will have compilation error
> if we don't export symbol for __divdi3, but built-in module is fine.
> Is this an expectation? Or we can keep 64-bit division symbols here
> since they are no harm (they are software emulation when without
> hardware divider).
> 
> long long a, b, c;
> a = b / c;

We have proper helper functions for this:

See include/linux/math.h

Thanks,

	tglx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/