Re: [PATCH 6/6] mm/slab: allocation fastpath without disabling irq

[Christoph Lameter]

On Mon, 5 Jan 2015, Joonsoo Kim wrote:

> index 449fc6b..54656f0 100644
> --- a/mm/slab.c
> +++ b/mm/slab.c
> @@ -168,6 +168,41 @@ typedef unsigned short freelist_idx_t;
>
>  #define SLAB_OBJ_MAX_NUM ((1 << sizeof(freelist_idx_t) * BITS_PER_BYTE) - 1)
>
> +#ifdef CONFIG_PREEMPT
> +/*
> + * Calculate the next globally unique transaction for disambiguiation
> + * during cmpxchg. The transactions start with the cpu number and are then
> + * incremented by CONFIG_NR_CPUS.
> + */
> +#define TID_STEP  roundup_pow_of_two(CONFIG_NR_CPUS)
> +#else
> +/*
> + * No preemption supported therefore also no need to check for
> + * different cpus.
> + */
> +#define TID_STEP 1
> +#endif
> +
> +static inline unsigned long next_tid(unsigned long tid)
> +{
> +	return tid + TID_STEP;
> +}
> +
> +static inline unsigned int tid_to_cpu(unsigned long tid)
> +{
> +	return tid % TID_STEP;
> +}
> +
> +static inline unsigned long tid_to_event(unsigned long tid)
> +{
> +	return tid / TID_STEP;
> +}
> +
> +static inline unsigned int init_tid(int cpu)
> +{
> +	return cpu;
> +}
> +

Ok the above stuff needs to go into the common code. Maybe in mm/slab.h?
And its a significant feature contributed by me so I'd like to have an
attribution here.

>  /*
>   * true if a page was allocated from pfmemalloc reserves for network-based
>   * swap
> @@ -187,7 +222,8 @@ static bool pfmemalloc_active __read_mostly;
>   *
>   */
>  struct array_cache {
> -	unsigned int avail;
> +	unsigned long avail;
> +	unsigned long tid;
>  	unsigned int limit;
>  	unsigned int batchcount;
>  	unsigned int touched;
> @@ -657,7 +693,8 @@ static void start_cpu_timer(int cpu)
>  	}
>  }

This increases the per cpu struct size and should lead to a small
performance penalty.

> -	 */
> -	if (likely(objp)) {
> -		STATS_INC_ALLOCHIT(cachep);
> -		goto out;
> +	objp = ac->entry[avail - 1];
> +	if (unlikely(!this_cpu_cmpxchg_double(
> +		cachep->cpu_cache->avail, cachep->cpu_cache->tid,
> +		avail, tid,
> +		avail - 1, next_tid(tid))))
> +		goto redo;


Hmm... Ok that looks good.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/