Re: [PATCH v2 04/16] KVM: MMU: return bool in __rmap_write_protect

[Avi Kivity <avi@redhat.com>]

On 04/16/2012 05:14 PM, Takuya Yoshikawa wrote:
> On Sun, 15 Apr 2012 14:25:30 +0300
> Avi Kivity <avi@redhat.com> wrote:
>
> > > > @@ -1689,7 +1690,7 @@ static void mmu_sync_children(struct kvm_vcpu *vcpu,
> > > > 
> > > >  	kvm_mmu_pages_init(parent, &parents, &pages);
> > > >  	while (mmu_unsync_walk(parent, &pages)) {
> > > > -		int protected = 0;
> > > > +		bool protected = false;
> > > > 
> > > >  		for_each_sp(pages, sp, parents, i)
> > > >  			protected |= rmap_write_protect(vcpu->kvm, sp->gfn);
> > >
> > > Isn't this the reason we prefer int to bool?
> > >
> > > Not sure people like to use |= with boolean.
> > >
> > 
> > Why not?
> > 
>
> The code "bitwise OR assignment" is assuming the internal representations
> of true and false: true=1, false=0.

No, it doesn't.  |= converts the result back to bool.

In fact it's better than

   int x;
   ...
   x |= some_value() & MASK;

   since MASK might be of type longer than int, and the result can be
truncated.  With bool |=, it cannot.

Disassembly of section .text:

0000000000000000 <f>:

static bool x;

void f(long y)
{
    x |= y;
   0:   0f b6 05 00 00 00 00    movzbl 0x0(%rip),%eax        # 7 <f+0x7>
                        3: R_X86_64_PC32        .bss-0x4
   7:   48 09 c7                or     %rax,%rdi
   a:   0f 95 05 00 00 00 00    setne  0x0(%rip)        # 11 <f+0x11>
                        d: R_X86_64_PC32        .bss-0x4
}
  11:   c3                      retq  

The corresponding code with 'int x' would just store the truncated
result back into x.


> I might have seen some point if it had been
> 	protected = protected || rmap_...
>
>
> But the real question is whether there is any point in re-writing completely
> correct C code: there are tons of int like this in the kernel code.
>
> __rmap_write_protect() was introduced recently, so if this conversion is
> really worthwhile, I should have been told to use bool at that time, no?

It's up to developer and maintainer preference.  I like bool since it
documents the usage and is safer, but sometimes I miss it on review.

-- 
error compiling committee.c: too many arguments to function