Re: [RFC][PATCH] Faster generic_fls

[Falk Hueffner]

Linus Torvalds <[e-mail]> writes:

> On 30 Apr 2003, Alan Cox wrote:
> > 
> > It ought to be basically the same as ffs because if I remember rightly 
> > 
> > ffs(x^(x-1)) == fls(x)
> 
> So did anybody time this? ffs() we have hardware support for on x86,
> and it's even reasonably efficient on some CPU's ..

There appears to be hardware support for fls, too. This is what gcc
generates for

int fls(int x) {
    return x ? 32 - __builtin_clz(x) : 0;
}

fls:
        pushl   %ebp
        xorl    %edx, %edx
        movl    %esp, %ebp
        movl    8(%ebp), %eax
        testl   %eax, %eax
        je      .L3
        bsrl    %eax, %ecx
        movl    $32, %edx
        xorl    $31, %ecx
        subl    %ecx, %edx
.L3:
        popl    %ebp
        movl    %edx, %eax
        ret

-- 
	Falk