DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Kinsella, Ray" <mdr@ashroe.eu>
To: Olivier Matz <olivier.matz@6wind.com>,
	Dharmik Thakkar <dharmik.thakkar@arm.com>
Cc: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>,
	dev@dpdk.org, nd@arm.com, joyce.kong@arm.com
Subject: Re: [dpdk-dev] [PATCH v3 2/2] lib/mempool: distinguish debug counters from cache and pool
Date: Fri, 23 Apr 2021 11:41:39 +0100
Message-ID: <d1e81798-e8b0-c8fb-eb8e-4c54e32f0d23@ashroe.eu> (raw)
In-Reply-To: <20210421162938.GG1726@platinum>



On 21/04/2021 17:29, Olivier Matz wrote:
> Hi Dharmik,
> 
> Please see some comments below.
> 
> On Mon, Apr 19, 2021 at 07:08:00PM -0500, Dharmik Thakkar wrote:
>> From: Joyce Kong <joyce.kong@arm.com>
>>
>> If cache is enabled, objects will be retrieved/put from/to cache,
>> subsequently from/to the common pool. Now the debug stats calculate
>> the objects retrieved/put from/to cache and pool together, it is
>> better to distinguish them.
>>
>> Signed-off-by: Joyce Kong <joyce.kong@arm.com>
>> Signed-off-by: Dharmik Thakkar <dharmik.thakkar@arm.com>
>> Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
>> Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
>> ---
>>  lib/librte_mempool/rte_mempool.c | 24 ++++++++++++++++
>>  lib/librte_mempool/rte_mempool.h | 47 ++++++++++++++++++++++----------
>>  2 files changed, 57 insertions(+), 14 deletions(-)
>>
>> diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
>> index afb1239c8d48..339f14455624 100644
>> --- a/lib/librte_mempool/rte_mempool.c
>> +++ b/lib/librte_mempool/rte_mempool.c
>> @@ -1244,6 +1244,18 @@ rte_mempool_dump(FILE *f, struct rte_mempool *mp)
>>  	for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) {
>>  		sum.put_bulk += mp->stats[lcore_id].put_bulk;
>>  		sum.put_objs += mp->stats[lcore_id].put_objs;
>> +		sum.put_common_pool_bulk +=
>> +			mp->stats[lcore_id].put_common_pool_bulk;
>> +		sum.put_common_pool_objs +=
>> +			mp->stats[lcore_id].put_common_pool_objs;
>> +		sum.put_cache_bulk += mp->stats[lcore_id].put_cache_bulk;
>> +		sum.put_cache_objs += mp->stats[lcore_id].put_cache_objs;
>> +		sum.get_common_pool_bulk +=
>> +			mp->stats[lcore_id].get_common_pool_bulk;
>> +		sum.get_common_pool_objs +=
>> +			mp->stats[lcore_id].get_common_pool_objs;
>> +		sum.get_cache_bulk += mp->stats[lcore_id].get_cache_bulk;
>> +		sum.get_cache_objs += mp->stats[lcore_id].get_cache_objs;
>>  		sum.get_success_bulk += mp->stats[lcore_id].get_success_bulk;
>>  		sum.get_success_objs += mp->stats[lcore_id].get_success_objs;
>>  		sum.get_fail_bulk += mp->stats[lcore_id].get_fail_bulk;
>> @@ -1254,6 +1266,18 @@ rte_mempool_dump(FILE *f, struct rte_mempool *mp)
>>  	fprintf(f, "  stats:\n");
>>  	fprintf(f, "    put_bulk=%"PRIu64"\n", sum.put_bulk);
>>  	fprintf(f, "    put_objs=%"PRIu64"\n", sum.put_objs);
>> +	fprintf(f, "    put_common_pool_bulk=%"PRIu64"\n",
>> +						sum.put_common_pool_bulk);
>> +	fprintf(f, "    put_common_pool_objs=%"PRIu64"\n",
>> +						sum.put_common_pool_objs);
>> +	fprintf(f, "    put_cache_bulk=%"PRIu64"\n", sum.put_cache_bulk);
>> +	fprintf(f, "    put_cache_objs=%"PRIu64"\n", sum.put_cache_objs);
>> +	fprintf(f, "    get_common_pool_bulk=%"PRIu64"\n",
>> +						sum.get_common_pool_bulk);
>> +	fprintf(f, "    get_common_pool_objs=%"PRIu64"\n",
>> +						sum.get_common_pool_objs);
>> +	fprintf(f, "    get_cache_bulk=%"PRIu64"\n", sum.get_cache_bulk);
>> +	fprintf(f, "    get_cache_objs=%"PRIu64"\n", sum.get_cache_objs);
>>  	fprintf(f, "    get_success_bulk=%"PRIu64"\n", sum.get_success_bulk);
>>  	fprintf(f, "    get_success_objs=%"PRIu64"\n", sum.get_success_objs);
>>  	fprintf(f, "    get_fail_bulk=%"PRIu64"\n", sum.get_fail_bulk);
>> diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
>> index 848a19226149..0959f8a3f367 100644
>> --- a/lib/librte_mempool/rte_mempool.h
>> +++ b/lib/librte_mempool/rte_mempool.h
>> @@ -66,12 +66,20 @@ extern "C" {
>>   * A structure that stores the mempool statistics (per-lcore).
>>   */
>>  struct rte_mempool_debug_stats {
>> -	uint64_t put_bulk;         /**< Number of puts. */
>> -	uint64_t put_objs;         /**< Number of objects successfully put. */
>> -	uint64_t get_success_bulk; /**< Successful allocation number. */
>> -	uint64_t get_success_objs; /**< Objects successfully allocated. */
>> -	uint64_t get_fail_bulk;    /**< Failed allocation number. */
>> -	uint64_t get_fail_objs;    /**< Objects that failed to be allocated. */
>> +	uint64_t put_bulk;		  /**< Number of puts. */
>> +	uint64_t put_objs;		  /**< Number of objects successfully put. */
>> +	uint64_t put_common_pool_bulk;	  /**< Number of bulks enqueued in common pool. */
>> +	uint64_t put_common_pool_objs;	  /**< Number of objects enqueued in common pool. */
>> +	uint64_t put_cache_bulk;	  /**< Number of bulks enqueued in cache. */
>> +	uint64_t put_cache_objs;	  /**< Number of objects enqueued in cache. */
>> +	uint64_t get_common_pool_bulk;    /**< Number of bulks dequeued from common pool. */
>> +	uint64_t get_common_pool_objs;	  /**< Number of objects dequeued from common pool. */
>> +	uint64_t get_cache_bulk;	  /**< Number of bulks dequeued from cache. */
>> +	uint64_t get_cache_objs;	  /**< Number of objects dequeued from cache. */
>> +	uint64_t get_success_bulk;	  /**< Successful allocation number. */
>> +	uint64_t get_success_objs;	  /**< Objects successfully allocated. */
>> +	uint64_t get_fail_bulk;		  /**< Failed allocation number. */
>> +	uint64_t get_fail_objs;		  /**< Objects that failed to be allocated. */
> 
> I missed it the first time, but this changes the size of the
> rte_mempool_debug_stats structure. I think we don't care about this ABI
> breakage because this structure is only defined if
> RTE_LIBRTE_MEMPOOL_DEBUG is set. But just in case, adding Ray as Cc.

Agreed, if it is just a debugging non-default feature. 

> About the field themselves, I'm not certain that there is an added value
> to have stats for cache gets and puts. My feeling is that the important
> stat to monitor is the access to common pool, because it is the one that
> highlights a possible performance impact (contention). The cache stats
> are more or less equal to "success + fail - common". Moreover, it will
> simplify the patch and avoid risks of mistakes.
> 
> What do you think?
> 
>>  	/** Successful allocation number of contiguous blocks. */
>>  	uint64_t get_success_blks;
>>  	/** Failed allocation number of contiguous blocks. */
>> @@ -699,10 +707,18 @@ rte_mempool_ops_dequeue_bulk(struct rte_mempool *mp,
>>  		void **obj_table, unsigned n)
>>  {
>>  	struct rte_mempool_ops *ops;
>> +	int ret;
>>  
>>  	rte_mempool_trace_ops_dequeue_bulk(mp, obj_table, n);
>>  	ops = rte_mempool_get_ops(mp->ops_index);
>> -	return ops->dequeue(mp, obj_table, n);
>> +	ret = ops->dequeue(mp, obj_table, n);
>> +	if (ret == 0) {
>> +		__MEMPOOL_STAT_ADD(mp, get_common_pool_bulk, 1);
>> +		__MEMPOOL_STAT_ADD(mp, get_common_pool_objs, n);
>> +		__MEMPOOL_STAT_ADD(mp, get_success_bulk, 1);
>> +		__MEMPOOL_STAT_ADD(mp, get_success_objs, n);
>> +	}
>> +	return ret;
>>  }
>>  
>>  /**
>> @@ -749,6 +765,8 @@ rte_mempool_ops_enqueue_bulk(struct rte_mempool *mp, void * const *obj_table,
>>  {
>>  	struct rte_mempool_ops *ops;
>>  
>> +	__MEMPOOL_STAT_ADD(mp, put_common_pool_bulk, 1);
>> +	__MEMPOOL_STAT_ADD(mp, put_common_pool_objs, n);
>>  	rte_mempool_trace_ops_enqueue_bulk(mp, obj_table, n);
>>  	ops = rte_mempool_get_ops(mp->ops_index);
>>  	return ops->enqueue(mp, obj_table, n);
>> @@ -1297,14 +1315,18 @@ __mempool_generic_put(struct rte_mempool *mp, void * const *obj_table,
>>  
>>  	/* Add elements back into the cache */
>>  	rte_memcpy(&cache_objs[0], obj_table, sizeof(void *) * n);
>> -
>>  	cache->len += n;
>>  
>> +	__MEMPOOL_STAT_ADD(mp, put_cache_bulk, 1);
>> +
>>  	if (cache->len >= cache->flushthresh) {
>> +		__MEMPOOL_STAT_ADD(mp, put_cache_objs,
>> +				   n - (cache->len - cache->size));
>>  		rte_mempool_ops_enqueue_bulk(mp, &cache->objs[cache->size],
>>  				cache->len - cache->size);
>>  		cache->len = cache->size;
>> -	}
>> +	} else
>> +		__MEMPOOL_STAT_ADD(mp, put_cache_objs, n);
>>  
> 
> In case we keep cache stats, I'd add {} after the else to be consistent
> with the if().
> 
>>  	return;
>>  
>> @@ -1438,8 +1460,8 @@ __mempool_generic_get(struct rte_mempool *mp, void **obj_table,
>>  
>>  	cache->len -= n;
>>  
>> -	__MEMPOOL_STAT_ADD(mp, get_success_bulk, 1);
>> -	__MEMPOOL_STAT_ADD(mp, get_success_objs, n);
>> +	__MEMPOOL_STAT_ADD(mp, get_cache_bulk, 1);
>> +	__MEMPOOL_STAT_ADD(mp, get_cache_objs, n);
> 
> In case we keep cache stats, I don't think we should remove get_success
> stats increment. Else, the success stats will never be incremented when
> retrieving objects from the cache.
> 
> 
>>  
>>  	return 0;
>>  
>> @@ -1451,9 +1473,6 @@ __mempool_generic_get(struct rte_mempool *mp, void **obj_table,
>>  	if (ret < 0) {
>>  		__MEMPOOL_STAT_ADD(mp, get_fail_bulk, 1);
>>  		__MEMPOOL_STAT_ADD(mp, get_fail_objs, n);
>> -	} else {
>> -		__MEMPOOL_STAT_ADD(mp, get_success_bulk, 1);
>> -		__MEMPOOL_STAT_ADD(mp, get_success_objs, n);
>>  	}
>>  
>>  	return ret;
>> -- 
>> 2.17.1
>>

  parent reply	other threads:[~2021-04-23 10:41 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-18 11:20 [dpdk-dev] [PATCH v2] " Joyce Kong
2021-04-07 14:28 ` Olivier Matz
2021-04-20  0:31   ` Dharmik Thakkar
2021-04-20  0:07 ` [dpdk-dev] [PATCH v3 0/2] lib/mempool: add debug stats Dharmik Thakkar
2021-04-20  0:07   ` [dpdk-dev] [PATCH v3 1/2] lib/mempool: make stats macro generic Dharmik Thakkar
2021-04-21 16:09     ` Olivier Matz
2021-04-20  0:08   ` [dpdk-dev] [PATCH v3 2/2] lib/mempool: distinguish debug counters from cache and pool Dharmik Thakkar
2021-04-21 16:29     ` Olivier Matz
2021-04-22 21:27       ` Dharmik Thakkar
2021-04-22 21:47         ` Honnappa Nagarahalli
2021-04-23 10:41       ` Kinsella, Ray [this message]
2021-04-23  1:29   ` [dpdk-dev] [PATCH v4 0/2] lib/mempool: add debug stats Dharmik Thakkar
2021-04-23  1:29     ` [dpdk-dev] [PATCH v4 1/2] lib/mempool: make stats macro generic Dharmik Thakkar
2021-04-23  1:29     ` [dpdk-dev] [PATCH v4 2/2] lib/mempool: distinguish debug counters from cache and pool Dharmik Thakkar
2021-04-23 20:29       ` Dharmik Thakkar
2021-04-27 12:18       ` Olivier Matz
2021-04-27 12:28     ` [dpdk-dev] [PATCH v4 0/2] lib/mempool: add debug stats Olivier Matz
2021-04-27 16:01     ` [dpdk-dev] [PATCH v5 0/2] mempool: " Dharmik Thakkar
2021-04-27 16:01       ` [dpdk-dev] [PATCH v5 1/2] mempool: make stats macro generic Dharmik Thakkar
2021-04-27 16:01       ` [dpdk-dev] [PATCH v5 2/2] mempool: distinguish debug counters from cache and pool Dharmik Thakkar
2021-05-04  6:54         ` Olivier Matz
2021-05-04  7:02       ` [dpdk-dev] [PATCH v5 0/2] mempool: add debug stats David Marchand

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=d1e81798-e8b0-c8fb-eb8e-4c54e32f0d23@ashroe.eu \
    --to=mdr@ashroe.eu \
    --cc=andrew.rybchenko@oktetlabs.ru \
    --cc=dev@dpdk.org \
    --cc=dharmik.thakkar@arm.com \
    --cc=joyce.kong@arm.com \
    --cc=nd@arm.com \
    --cc=olivier.matz@6wind.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git