Re: [PATCH -next] evm: Use IS_ENABLED to initialize .enabled

[xiujianfeng]

在 2022/6/27 0:13, Mimi Zohar 写道:
> On Tue, 2022-06-21 at 19:17 -0700, Eric Biggers wrote:
> > On Tue, Jun 21, 2022 at 10:03:39AM -0400, Mimi Zohar wrote:
> > > On Tue, 2022-06-21 at 18:58 +0800, xiujianfeng wrote:
> > > > Hi, Ahmad
> > > >
> > > > 在 2022/6/7 14:06, Ahmad Fatoum 写道:
> > > > > On 06.06.22 12:10, Xiu Jianfeng wrote:
> > > > > > Use IS_ENABLED(CONFIG_XXX) instead of #ifdef/#endif statements to
> > > > > > initialize .enabled, minor simplicity improvement.
> > > The difference between using ifdef's and IS_ENABLED is when the
> > > decision is made - build time, run time.   Please update the patch
> > > description providing an explanation for needing to make the decision
> > > at run time.
> > >
> > > thanks,
> > IS_ENABLED() is a compile time constant.  So the patch looks fine to me.
> Thanks, Eric, for the clarification.
>
> As LSMs are only builtin, why the need for using IS_ENABLED as opposed
> to IS_BUILTIN?
>
> #define IS_ENABLED(option) __or(IS_BUILTIN(option), IS_MODULE(option))
>
> thanks,

I think IS_ENALBED() is a bit more generic, maybe. here is another 
example in rcutorture.c

which uses IS_ENABLED() to initialize the member in structure:

static struct rcu_torture_ops rcu_ops = {

...

.can_boost = IS_ENABLED(CONFIG_RCU_BOOST),

...

};

Do you want me to change IS_ENABLED() to IS_BUILTIN()?

> Mimi
>
> .