DPDK patches and discussions
 help / color / mirror / Atom feed
From: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
To: Joyce Kong <Joyce.Kong@arm.com>,
	"thomas@monjalon.net" <thomas@monjalon.net>,
	"david.marchand@redhat.com" <david.marchand@redhat.com>,
	Ruifeng Wang <Ruifeng.Wang@arm.com>,
	"konstantin.ananyev@intel.com" <konstantin.ananyev@intel.com>,
	"rsanford@akamai.com" <rsanford@akamai.com>,
	"erik.g.carrillo@intel.com" <erik.g.carrillo@intel.com>,
	"olivier.matz@6wind.com" <olivier.matz@6wind.com>,
	"yipeng1.wang@intel.com" <yipeng1.wang@intel.com>,
	"sameh.gobriel@intel.com" <sameh.gobriel@intel.com>,
	"bruce.richardson@intel.com" <bruce.richardson@intel.com>,
	"vladimir.medvedkin@intel.com" <vladimir.medvedkin@intel.com>,
	"anatoly.burakov@intel.com" <anatoly.burakov@intel.com>,
	"andrew.rybchenko@oktetlabs.ru" <andrew.rybchenko@oktetlabs.ru>,
	"jerinj@marvell.com" <jerinj@marvell.com>,
	"declan.doherty@intel.com" <declan.doherty@intel.com>,
	"ciara.power@intel.com" <ciara.power@intel.com>,
	"xiaoyun.li@intel.com" <xiaoyun.li@intel.com>,
	"nicolas.chautru@intel.com" <nicolas.chautru@intel.com>,
	"maryam.tahhan@intel.com" <maryam.tahhan@intel.com>,
	"reshma.pattan@intel.com" <reshma.pattan@intel.com>,
	"cristian.dumitrescu@intel.com" <cristian.dumitrescu@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>, nd <nd@arm.com>,
	Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>,
	nd <nd@arm.com>
Subject: Re: [dpdk-dev] [PATCH v1 01/12] test/pmd_perf: use compiler atomic builtins for polling sync
Date: Mon, 8 Nov 2021 22:50:32 +0000	[thread overview]
Message-ID: <DBAPR08MB5814015E30BBCBAC702C8B9598919@DBAPR08MB5814.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <20210802101847.3462-2-joyce.kong@arm.com>

<snip>

> 
> Convert rte_atomic usages to compiler atomic built-ins for polling sync in
> pmd_perf test cases.
> 
> Signed-off-by: Joyce Kong <joyce.kong@arm.com>
> Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
> ---
>  app/test/test_pmd_perf.c | 12 +++++-------
>  1 file changed, 5 insertions(+), 7 deletions(-)
> 
> diff --git a/app/test/test_pmd_perf.c b/app/test/test_pmd_perf.c index
> 3a248d512c..41ef001b90 100644
> --- a/app/test/test_pmd_perf.c
> +++ b/app/test/test_pmd_perf.c
> @@ -10,7 +10,6 @@
>  #include <rte_cycles.h>
>  #include <rte_ethdev.h>
>  #include <rte_byteorder.h>
> -#include <rte_atomic.h>
>  #include <rte_malloc.h>
>  #include "packet_burst_generator.h"
>  #include "test.h"
> @@ -526,7 +525,7 @@ main_loop(__rte_unused void *args)
>  	return 0;
>  }
> 
> -static rte_atomic64_t start;
> +static uint64_t start;
> 
>  static inline int
>  poll_burst(void *args)
> @@ -564,8 +563,7 @@ poll_burst(void *args)
>  		num[portid] = pkt_per_port;
>  	}
> 
> -	while (!rte_atomic64_read(&start))
> -		;
> +	rte_wait_until_equal_64(&start, 1, __ATOMIC_RELAXED);
This needs to be ACQUIRE. Please see comments below for function 'exec_burst'.

> 
>  	cur_tsc = rte_rdtsc();
>  	while (total) {
> @@ -617,7 +615,7 @@ exec_burst(uint32_t flags, int lcore)
>  	pkt_per_port = MAX_TRAFFIC_BURST;
>  	num = pkt_per_port * conf->nb_ports;
> 
> -	rte_atomic64_init(&start);
> +	__atomic_store_n(&start, 0, __ATOMIC_RELAXED);
> 
>  	/* start polling thread, but not actually poll yet */
>  	rte_eal_remote_launch(poll_burst,
> @@ -625,7 +623,7 @@ exec_burst(uint32_t flags, int lcore)
> 
>  	/* Only when polling first */
>  	if (flags == SC_BURST_POLL_FIRST)
> -		rte_atomic64_set(&start, 1);
> +		__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
This line can be combined with initialization of 'start'. i.e.
if (flags == SC_BURST_POLL_FIRST) {
	__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
} else {
	__atomic_store_n(&start, 0, __ATOMIC_RELAXED);
}
It is ok to use RELAXED here as it is part of initialization and 'rte_eal_remote_launch' will ensure that the store to 'start' is visible to the worker thread.

> 
>  	/* start xmit */
>  	i = 0;
> @@ -642,7 +640,7 @@ exec_burst(uint32_t flags, int lcore)
> 
>  	/* only when polling second  */
>  	if (flags == SC_BURST_XMIT_FIRST)
> -		rte_atomic64_set(&start, 1);
> +		__atomic_store_n(&start, 1, __ATOMIC_RELAXED);
This store should happen after transmitting the packets. It is better to use RELEASE here.

> 
>  	/* wait for polling finished */
>  	diff_tsc = rte_eal_wait_lcore(lcore);
> --
> 2.17.1


  reply	other threads:[~2021-11-08 22:51 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-02 10:18 [dpdk-dev] [PATCH v1 00/12] use compiler atomic builtins for app Joyce Kong
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 01/12] test/pmd_perf: use compiler atomic builtins for polling sync Joyce Kong
2021-11-08 22:50   ` Honnappa Nagarahalli [this message]
2021-11-10  6:10     ` Joyce Kong
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 02/12] test/ring_perf: use compiler atomic builtins for lcores sync Joyce Kong
2021-11-09  5:43   ` Honnappa Nagarahalli
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 03/12] test/timer: use compiler atomic builtins for sync Joyce Kong
2021-11-09 20:59   ` Honnappa Nagarahalli
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 04/12] test/stack_perf: use compiler atomics for lcore sync Joyce Kong
2021-11-09 21:12   ` Honnappa Nagarahalli
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 05/12] test/bpf: use compiler atomics for calculation Joyce Kong
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 06/12] test/func_reentrancy: use compiler atomic for data sync Joyce Kong
2021-11-09 21:54   ` Honnappa Nagarahalli
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 07/12] app/eventdev: use compiler atomic builtins for packets sync Joyce Kong
2021-11-10 23:19   ` Honnappa Nagarahalli
2021-11-11  7:27     ` Joyce Kong
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 08/12] app/crypto: use compiler atomic builtins for display sync Joyce Kong
2021-11-09 22:11   ` Honnappa Nagarahalli
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 09/12] app/compress: " Joyce Kong
2021-11-09 22:59   ` Honnappa Nagarahalli
2021-11-11  8:13     ` Joyce Kong
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 10/12] app/testpmd: use compiler atomic builtins for port sync Joyce Kong
2021-11-09 23:14   ` Honnappa Nagarahalli
2021-11-11  8:51     ` Joyce Kong
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 11/12] app/bbdev: use compiler atomics for thread sync Joyce Kong
2021-11-10 21:25   ` Honnappa Nagarahalli
2021-08-02 10:18 ` [dpdk-dev] [PATCH v1 12/12] app: remove unnecessary include of atomic Joyce Kong
2021-10-21  6:35 ` [dpdk-dev] [PATCH v1 00/12] use compiler atomic builtins for app Joyce Kong

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=DBAPR08MB5814015E30BBCBAC702C8B9598919@DBAPR08MB5814.eurprd08.prod.outlook.com \
    --to=honnappa.nagarahalli@arm.com \
    --cc=Joyce.Kong@arm.com \
    --cc=Ruifeng.Wang@arm.com \
    --cc=anatoly.burakov@intel.com \
    --cc=andrew.rybchenko@oktetlabs.ru \
    --cc=bruce.richardson@intel.com \
    --cc=ciara.power@intel.com \
    --cc=cristian.dumitrescu@intel.com \
    --cc=david.marchand@redhat.com \
    --cc=declan.doherty@intel.com \
    --cc=dev@dpdk.org \
    --cc=erik.g.carrillo@intel.com \
    --cc=jerinj@marvell.com \
    --cc=konstantin.ananyev@intel.com \
    --cc=maryam.tahhan@intel.com \
    --cc=nd@arm.com \
    --cc=nicolas.chautru@intel.com \
    --cc=olivier.matz@6wind.com \
    --cc=reshma.pattan@intel.com \
    --cc=rsanford@akamai.com \
    --cc=sameh.gobriel@intel.com \
    --cc=thomas@monjalon.net \
    --cc=vladimir.medvedkin@intel.com \
    --cc=xiaoyun.li@intel.com \
    --cc=yipeng1.wang@intel.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).