Re: [PATCH v4 3/6] clk: fractional-divider: keep mwidth and nwidth internally

[Stephen Boyd <sboyd@codeaurora.org>]

On 07/22, Andy Shevchenko wrote:
> On Tue, 2015-07-21 at 17:41 -0700, Stephen Boyd wrote:
> > On 07/14, Andy Shevchenko wrote:
> > > diff --git a/drivers/clk/clk-fractional-divider.c b/drivers/clk/clk
> > > -fractional-divider.c
> > > index 7cfcc56..16f42ae 100644
> > > --- a/drivers/clk/clk-fractional-divider.c
> > > +++ b/drivers/clk/clk-fractional-divider.c
> > > @@ -128,9 +128,11 @@ struct clk 
> > > *clk_register_fractional_divider(struct device *dev,
> > >  
> > >  	fd->reg = reg;
> > >  	fd->mshift = mshift;
> > > -	fd->mmask = (BIT(mwidth) - 1) << mshift;
> > > +	fd->mwidth = mwidth;
> > > +	fd->mmask = GENMASK(mwidth - 1, 0) << mshift;
> > >  	fd->nshift = nshift;
> > > -	fd->nmask = (BIT(nwidth) - 1) << nshift;
> > > +	fd->nwidth = nwidth;
> > > +	fd->nmask = GENMASK(nwidth - 1, 0) << nshift;
> > 
> > Please do the shifts in the GENMASK.
> 
> It's not optimal. Waste of performance.
> 
> 32-bit case on 32-bit machine (similar to other cases on x86).
> 
> a) GENMASK(x - 1, 0) << y
> 
>         movl    $32, %ecx
>         subb    4(%esp), %cl
>         movl    $-1, %eax
>         shrl    %cl, %eax
>         movzbl  8(%esp), %ecx
>         sall    %cl, %eax
>         ret
> 
> b) GENMASK(x + y - 1, y)
> 
>         pushl   %esi
>         pushl   %ebx
>         movl    $1, %edx
>         movzbl  12(%esp), %eax
>         movzbl  16(%esp), %esi
>         movl    $-1, %ebx
>         subl    %eax, %edx
>         movl    %ebx, %eax
>         subl    %esi, %edx
>         leal    31(%edx), %ecx
>         shrl    %cl, %eax
>         movl    %esi, %ecx
>         sall    %cl, %ebx
>         andl    %ebx, %eax
>         popl    %ebx
>         popl    %esi
>         ret
> 

This is not hotpath code, so why do we care?

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project