DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Eads, Gage" <gage.eads@intel.com>
To: "Van Haaren, Harry" <harry.van.haaren@intel.com>,
	"dev@dpdk.org" <dev@dpdk.org>
Cc: "jerin.jacob@caviumnetworks.com" <jerin.jacob@caviumnetworks.com>
Subject: Re: [dpdk-dev] [PATCH] event/sw: fix credit tracking in port dequeue
Date: Tue, 30 May 2017 14:11:12 +0000	[thread overview]
Message-ID: <9184057F7FC11744A2107296B6B8EB1E01EC4180@FMSMSX108.amr.corp.intel.com> (raw)
In-Reply-To: <1496071015-24906-1-git-send-email-harry.van.haaren@intel.com>

A couple nit-picks on the test, but once resolved:
Acked-by: Gage Eads <gage.eads@intel.com>

>  -----Original Message-----
>  From: Van Haaren, Harry
>  Sent: Monday, May 29, 2017 10:17 AM
>  To: dev@dpdk.org
>  Cc: jerin.jacob@caviumnetworks.com; Eads, Gage <gage.eads@intel.com>;
>  Van Haaren, Harry <harry.van.haaren@intel.com>
>  Subject: [PATCH] event/sw: fix credit tracking in port dequeue
>  
>  This patch targets the next-eventdev tree.
>  
>  Single-link optimized ports previously did not correctly track credits when
>  dequeued, and re-enqueued as a FORWARD type. This could "inflate" the
>  number of credits in the system.
>  
>  A unit test is added to reproduce and verify the issue. The fixed implementation
>  counts FORWARD packets and reduces the number of credits the port has if it is
>  of single-link type.
>  
>  Fixes: 656af9180014 ("event/sw: add worker core functions")
>  
>  Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
>  ---
>   drivers/event/sw/sw_evdev_worker.c |  5 ++++
>   test/test/test_eventdev_sw.c       | 58
>  +++++++++++++++++++++++++++++++++++++-
>   2 files changed, 62 insertions(+), 1 deletion(-)
>  
>  diff --git a/drivers/event/sw/sw_evdev_worker.c
>  b/drivers/event/sw/sw_evdev_worker.c
>  index 9cb6bef..b738506 100644
>  --- a/drivers/event/sw/sw_evdev_worker.c
>  +++ b/drivers/event/sw/sw_evdev_worker.c
>  @@ -87,6 +87,7 @@ sw_event_enqueue_burst(void *port, const struct
>  rte_event ev[], uint16_t num)
>   			return 0;
>   	}
>  
>  +	uint32_t forwards = 0;
>   	for (i = 0; i < num; i++) {
>   		int op = ev[i].op;
>   		int outstanding = p->outstanding_releases > 0; @@ -95,6
>  +96,7 @@ sw_event_enqueue_burst(void *port, const struct rte_event ev[],
>  uint16_t num)
>   		p->inflight_credits -= (op == RTE_EVENT_OP_NEW);
>   		p->inflight_credits += (op == RTE_EVENT_OP_RELEASE) *
>   					outstanding;
>  +		forwards += (op == RTE_EVENT_OP_FORWARD);
>  
>   		new_ops[i] = sw_qe_flag_map[op];
>   		new_ops[i] &= ~(invalid_qid << QE_FLAG_VALID_SHIFT); @@ -
>  113,6 +115,9 @@ sw_event_enqueue_burst(void *port, const struct rte_event
>  ev[], uint16_t num)
>   		}
>   	}
>  
>  +	/* handle directed port forward credits */
>  +	p->inflight_credits -= forwards * p->is_directed;
>  +
>   	/* returns number of events actually enqueued */
>   	uint32_t enq = qe_ring_enqueue_burst_with_ops(p->rx_worker_ring,
>  ev, i,
>   					     new_ops);
>  diff --git a/test/test/test_eventdev_sw.c b/test/test/test_eventdev_sw.c index
>  b187d02..aaa9729 100644
>  --- a/test/test/test_eventdev_sw.c
>  +++ b/test/test/test_eventdev_sw.c
>  @@ -548,6 +548,57 @@ test_single_directed_packet(struct test *t)
>   	return 0;
>   }
>  
>  +static int
>  +test_directed_forward_credits(struct test *t) {
>  +	uint32_t i;
>  +	int32_t err;
>  +
>  +	if (init(t, 1, 1) < 0 ||
>  +			create_ports(t, 1) < 0 ||
>  +			create_directed_qids(t, 1, t->port) < 0)
>  +		return -1;
>  +
>  +	if (rte_event_dev_start(evdev) < 0) {
>  +		printf("%d: Error with start call\n", __LINE__);
>  +		return -1;
>  +	}
>  +
>  +	struct rte_event ev = {
>  +			.op = RTE_EVENT_OP_NEW,
>  +			.queue_id = 0,
>  +	};
>  +
>  +	for (i = 0; i < 1000; i++) {
>  +		err = rte_event_enqueue_burst(evdev, 0, &ev, 1);
>  +		if (err < 0) {
>  +			printf("%d: error failed to enqueue\n", __LINE__);
>  +			return -1;
>  +		}
>  +		rte_event_schedule(evdev);
>  +		struct test_event_dev_stats stats;
>  +		err = test_event_dev_stats_get(evdev, &stats);

Does this stats get call serve a purpose?

>  +		if (err) {
>  +			printf("%d: error failed to get stats\n", __LINE__);
>  +			return -1;
>  +		}
>  +
>  +		uint32_t deq_pkts;
>  +		deq_pkts = rte_event_dequeue_burst(evdev, 0, &ev, 1, 0);
>  +		if (deq_pkts != 1) {
>  +			printf("%d: error failed to deq\n", __LINE__);
>  +			return -1;
>  +		}
>  +
>  +		/* re-write event to be a forward, and continue looping it */
>  +		ev.op = RTE_EVENT_OP_FORWARD;
>  +	}
>  +
>  +	//rte_event_dev_dump(0, stdout);

Comment can be deleted

  reply	other threads:[~2017-05-30 14:11 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-05-29 15:16 Harry van Haaren
2017-05-30 14:11 ` Eads, Gage [this message]
2017-06-01 15:45   ` [dpdk-dev] [PATCH v2] " Harry van Haaren
2017-06-06  3:20     ` Jerin Jacob

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=9184057F7FC11744A2107296B6B8EB1E01EC4180@FMSMSX108.amr.corp.intel.com \
    --to=gage.eads@intel.com \
    --cc=dev@dpdk.org \
    --cc=harry.van.haaren@intel.com \
    --cc=jerin.jacob@caviumnetworks.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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).