Re: Yet more ARM breakage in linux-next

Re: Yet more ARM breakage in linux-next

[Geert Uytterhoeven]

On Fri, 5 Dec 2008, Rusty Russell wrote:
> On Thursday 04 December 2008 19:48:06 Geert Uytterhoeven wrote:
> > Also broke m68k (a bit more hidden due to other build failures).
> 
> And m68knommu has similar issues.
> 
> Here's what I have for m68k for tomorrow's linux-next.  Please steal.
> 
> m68k: define __fls
> 
> Like fls, but can't be handed 0 and returns the bit number.
> 
> (I broke this arch in linux-next by using __fls in generic code).
> 
> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
> 
> diff --git a/include/asm-m68k/bitops.h b/include/asm-m68k/bitops.h
> --- a/include/asm-m68k/bitops.h
> +++ b/include/asm-m68k/bitops.h
> @@ -315,6 +315,11 @@ static inline int fls(int x)
>  	return 32 - cnt;
>  }
>  
> +static inline int __fls(int x)
                 ^^^       ^^^
Other implementations take `unsigned long' and return `unsigned long'...

> +{
> +	return fls(x) - 1;
> +}
> +

On Thu, 4 Dec 2008, Rusty Russell wrote:
> Like fls, but can't be handed 0 and returns the bit number.
> 
> (I added find_last_bit() to bitmap.h which broke archs which didn't
>  define this)
> 
> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
> ---
>  arch/avr32/include/asm/bitops.h |    5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/arch/avr32/include/asm/bitops.h b/arch/avr32/include/asm/bitops.h
> --- a/arch/avr32/include/asm/bitops.h
> +++ b/arch/avr32/include/asm/bitops.h
> @@ -263,6 +263,11 @@ static inline int fls(unsigned long word
>  	return 32 - result;
>  }
>  
> +static inline int __fls(unsigned long word)
                 ^^^       ^^^^^^^^^^^^^
> +{
> +	return fls(word) - 1;
> +}
> +

... but this one uses mixed types?

What are the official semantics of __fls()?

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Re: Yet more ARM breakage in linux-next

[Rusty Russell]

On Monday 08 December 2008 08:20:59 Geert Uytterhoeven wrote:
> On Fri, 5 Dec 2008, Rusty Russell wrote: 
> > +static inline int __fls(int x)
>                  ^^^       ^^^
> Other implementations take `unsigned long' and return `unsigned long'...

It's all over the place, actually.  32 bit archs are especially loose.

I've been toying with the idea of a boottime testsuite for all the
bitops to see who gets them wrong.

> > +static inline int __fls(unsigned long word)
>                  ^^^       ^^^^^^^^^^^^^
> > +{
> > +	return fls(word) - 1;
> > +}
> > +
> 
> ... but this one uses mixed types?

I cut and pasted.  I thought you were 32 bit, so doesn't matter?

> What are the official semantics of __fls()?

Find last bit set in the word, undefined if word is 0.  Returns 0
to BITS_PER_LONG-1.

Cheers,
Rusty.