Re: Kernel broken on processors without performance counters

[Jason Baron <jasonbaron0@gmail.com>]

On 07/10/2015 10:13 AM, Peter Zijlstra wrote:
> On Wed, Jul 08, 2015 at 05:36:43PM -0700, Andy Lutomirski wrote:
>>>> In what universe is "static_key_false" a reasonable name for a
>>>> function that returns true if a static key is true?
>> I think the current naming is almost maximally bad.  The naming would
>> be less critical if it were typesafe, though.
> How about something like so on top? It will allow us to slowly migrate
> existing and new users over to the hopefully saner interface?
>
> ---
>  include/linux/jump_label.h | 67 +++++++++++++++++++++++++++++++++++++++++++++-
>  kernel/sched/core.c        | 18 ++++++-------
>  2 files changed, 74 insertions(+), 11 deletions(-)
>
> diff --git a/include/linux/jump_label.h b/include/linux/jump_label.h
> index f4de473f226b..98ed3c2ec78d 100644
> --- a/include/linux/jump_label.h
> +++ b/include/linux/jump_label.h
> @@ -213,6 +213,71 @@ static inline bool static_key_enabled(struct static_key *key)
>  	return static_key_count(key) > 0;
>  }
>  
> -#endif	/* _LINUX_JUMP_LABEL_H */
> +static inline void static_key_enable(struct static_key *key)
> +{
> +	int count = static_key_count(key);
> +
> +	WARN_ON_ONCE(count < 0 || count > 1);
> +
> +	if (!count)
> +		static_key_slow_inc(key);
> +}
> +
> +static inline void static_key_disable(struct static_key *key)
> +{
> +	int count = static_key_count(key);
> +
> +	WARN_ON_ONCE(count < 0 || count > 1);
> +
> +	if (count)
> +		static_key_slow_dec(key);
> +}

should those be __static_key_enable()/disable() to indicate that we don't
that we don't want ppl using these directly. Similarly for other 'internal'
functions.

> +
> +/* -------------------------------------------------------------------------- */
> +
> +/*
> + * likely -- default enabled, puts the branch body in-line
> + */
> +
> +struct static_likely_key {
> +	struct static_key key;
> +};
> +
> +#define STATIC_LIKELY_KEY_INIT	(struct static_likely_key){ .key = STATIC_KEY_INIT_TRUE, }
> +
> +static inline bool static_likely_branch(struct static_likely_key *key)
> +{
> +	return static_key_true(&key->key);
> +}
> +
> +/*
> + * unlikely -- default disabled, puts the branch body out-of-line
> + */
> +
> +struct static_unlikely_key {
> +	struct static_key key;
> +};
> +
> +#define STATIC_UNLIKELY_KEY_INIT (struct static_unlikely_key){ .key = STATIC_KEY_INIT_FALSE, }
> +
> +static inline bool static_unlikely_branch(struct static_unlikely_key *key)
> +{
> +	return static_key_false(&key->key);
> +}
> +
> +/*
> + * Advanced usage; refcount, branch is enabled when: count != 0
> + */
> +
> +#define static_branch_inc(_k)	static_key_slow_inc(&(_k)->key)
> +#define static_branch_dec(_k)	static_key_slow_dec(&(_k)->key)
> +

I think of these as operations on the 'static_key' so I still
like 'static_key_inc()/dec()' (removing the 'slow' makes them
different still).

> +/*
> + * Normal usage; boolean enable/disable.
> + */
> +
> +#define static_branch_enable(_k)	static_key_enable(&(_k)->key)
> +#define static_branch_disable(_k)	static_key_disable(&(_k)->key)
>  

Same here maybe: static_key_set_true()/false()?

Thanks,

-Jason