From: "Xie, Huawei" <huawei.xie@intel.com>
To: "Chen, Jing D" <jing.d.chen@intel.com>, "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH 2/2] examples/vhost: use factorized default Rx/Tx configuration
Date: Fri, 14 Nov 2014 02:17:32 +0000 [thread overview]
Message-ID: <C37D651A908B024F974696C65296B57B0F302796@SHSMSX101.ccr.corp.intel.com> (raw)
In-Reply-To: <4341B239C0EFF9468EE453F9E9F4604D0161CAD6@shsmsx102.ccr.corp.intel.com>
> -----Original Message-----
> From: Chen, Jing D
> Sent: Wednesday, November 12, 2014 11:02 PM
> To: Xie, Huawei; dev@dpdk.org
> Subject: RE: [PATCH 2/2] examples/vhost: use factorized default Rx/Tx
> configuration
>
> Hi,
>
> > -----Original Message-----
> > From: Xie, Huawei
> > Sent: Thursday, November 13, 2014 6:34 AM
> > To: dev@dpdk.org
> > Cc: Chen, Jing D; Xie, Huawei
> > Subject: [PATCH 2/2] examples/vhost: use factorized default Rx/Tx
> > configuration
> >
> > Refer to Pablo's commit:
> > "use factorized default Rx/Tx configuration
> >
> > For apps that were using default rte_eth_rxconf and rte_eth_txconf
> > structures, these have been removed and now they are obtained by
> > calling rte_eth_dev_info_get, just before setting up RX/TX queues."
> >
> > move zero copy's deferred start set up ahead.
> >
> > Signed-off-by: Huawei Xie <huawei.xie@intel.com>
> > ---
> > examples/vhost/main.c | 78 +++++++++++++++----------------------------------
> > --
> > 1 file changed, 22 insertions(+), 56 deletions(-)
> >
> > diff --git a/examples/vhost/main.c b/examples/vhost/main.c
> > index 2b1bf02..fa36913 100644
> > --- a/examples/vhost/main.c
> > +++ b/examples/vhost/main.c
> > @@ -79,25 +79,6 @@
> > + RTE_PKTMBUF_HEADROOM)
> > #define MBUF_CACHE_SIZE_ZCP 0
> >
> > -/*
> > - * RX and TX Prefetch, Host, and Write-back threshold values should be
> > - * carefully set for optimal performance. Consult the network
> > - * controller's datasheet and supporting DPDK documentation for guidance
> > - * on how these parameters should be set.
> > - */
> > -#define RX_PTHRESH 8 /* Default values of RX prefetch threshold reg. */
> > -#define RX_HTHRESH 8 /* Default values of RX host threshold reg. */
> > -#define RX_WTHRESH 4 /* Default values of RX write-back threshold reg. */
> > -
> > -/*
> > - * These default values are optimized for use with the Intel(R) 82599 10 GbE
> > - * Controller and the DPDK ixgbe PMD. Consider using other values for other
> > - * network controllers and/or network drivers.
> > - */
> > -#define TX_PTHRESH 36 /* Default values of TX prefetch threshold reg. */
> > -#define TX_HTHRESH 0 /* Default values of TX host threshold reg. */
> > -#define TX_WTHRESH 0 /* Default values of TX write-back threshold reg. */
> > -
> > #define MAX_PKT_BURST 32 /* Max burst size for RX/TX */
> > #define BURST_TX_DRAIN_US 100 /* TX drain every ~100us */
> >
> > @@ -217,32 +198,6 @@ static uint32_t burst_rx_retry_num =
> > BURST_RX_RETRIES;
> > /* Character device basename. Can be set by user. */
> > static char dev_basename[MAX_BASENAME_SZ] = "vhost-net";
> >
> > -
> > -/* Default configuration for rx and tx thresholds etc. */
> > -static struct rte_eth_rxconf rx_conf_default = {
> > - .rx_thresh = {
> > - .pthresh = RX_PTHRESH,
> > - .hthresh = RX_HTHRESH,
> > - .wthresh = RX_WTHRESH,
> > - },
> > - .rx_drop_en = 1,
> > -};
> > -
> > -/*
> > - * These default values are optimized for use with the Intel(R) 82599 10 GbE
> > - * Controller and the DPDK ixgbe/igb PMD. Consider using other values for
> > other
> > - * network controllers and/or network drivers.
> > - */
> > -static struct rte_eth_txconf tx_conf_default = {
> > - .tx_thresh = {
> > - .pthresh = TX_PTHRESH,
> > - .hthresh = TX_HTHRESH,
> > - .wthresh = TX_WTHRESH,
> > - },
> > - .tx_free_thresh = 0, /* Use PMD default values */
> > - .tx_rs_thresh = 0, /* Use PMD default values */
> > -};
> > -
> > /* empty vmdq configuration structure. Filled in programatically */
> > static struct rte_eth_conf vmdq_conf_default = {
> > .rxmode = {
> > @@ -410,7 +365,9 @@ port_init(uint8_t port)
> > {
> > struct rte_eth_dev_info dev_info;
> > struct rte_eth_conf port_conf;
> > - uint16_t rx_rings, tx_rings;
> > + struct rte_eth_rxconf *rxconf;
> > + struct rte_eth_txconf *txconf;
> > + int16_t rx_rings, tx_rings;
> > uint16_t rx_ring_size, tx_ring_size;
> > int retval;
> > uint16_t q;
> > @@ -418,6 +375,21 @@ port_init(uint8_t port)
> > /* The max pool number from dev_info will be used to validate the
> > pool number specified in cmd line */
> > rte_eth_dev_info_get (port, &dev_info);
> >
> > + rxconf = &dev_info.default_rxconf;
> > + txconf = &dev_info.default_txconf;
> > + rxconf->rx_drop_en = 1;
> > +
> > + /*
> > + * Zero copy defers queue RX/TX start to the time when guest
> > + * finishes its startup and packet buffers from that guest are
> > + * available.
> > + */
> > + if (zero_copy) {
> > + rxconf->rx_deferred_start = 1;
> > + rxconf->rx_drop_en = 0;
> > + txconf->tx_deferred_start = 1;
> > + }
> > +
>
> May I know why 'rx_drop_en' is cleared after 'zero_copy' set?
:), first of all, this is not related to this patch. This patch inheritate old rx_drop_en behavior, and apply pablo's change.
Secondly, this is due to rx ring in zero copy case has very limited descriptor number. Clearing this setting will cache the packets
when there are not enough descriptors
>
> > /*configure the number of supported virtio devices based on VMDQ
> > limits */
> > num_devices = dev_info.max_vmdq_pools;
> >
> > @@ -460,14 +432,16 @@ port_init(uint8_t port)
> > /* Setup the queues. */
> > for (q = 0; q < rx_rings; q ++) {
> > retval = rte_eth_rx_queue_setup(port, q, rx_ring_size,
> > - rte_eth_dev_socket_id(port),
> > &rx_conf_default,
> > + rte_eth_dev_socket_id(port),
> > + rxconf,
> > vpool_array[q].pool);
> > if (retval < 0)
> > return retval;
> > }
> > for (q = 0; q < tx_rings; q ++) {
> > retval = rte_eth_tx_queue_setup(port, q, tx_ring_size,
> > - rte_eth_dev_socket_id(port),
> > &tx_conf_default);
> > + rte_eth_dev_socket_id(port),
> > + txconf);
> > if (retval < 0)
> > return retval;
> > }
> > @@ -2920,14 +2894,6 @@ MAIN(int argc, char *argv[])
> > char pool_name[RTE_MEMPOOL_NAMESIZE];
> > char ring_name[RTE_MEMPOOL_NAMESIZE];
> >
> > - /*
> > - * Zero copy defers queue RX/TX start to the time when
> > guest
> > - * finishes its startup and packet buffers from that guest are
> > - * available.
> > - */
> > - rx_conf_default.rx_deferred_start = (uint8_t)zero_copy;
> > - rx_conf_default.rx_drop_en = 0;
> > - tx_conf_default.tx_deferred_start = (uint8_t)zero_copy;
> > nb_mbuf = num_rx_descriptor
> > + num_switching_cores * MBUF_CACHE_SIZE_ZCP
> > + num_switching_cores * MAX_PKT_BURST;
> > --
> > 1.8.1.4
next prev parent reply other threads:[~2014-11-14 2:08 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-11-12 22:34 [dpdk-dev] [PATCH 0/2] examples/vhost: support new VMDQ api and new nic i40e in vhost example Huawei Xie
2014-11-12 22:34 ` [dpdk-dev] [PATCH 1/2] examples/vhost: support new VMDQ API and new nic i40e Huawei Xie
2014-11-13 0:49 ` Ouyang, Changchun
2014-11-13 1:20 ` Xie, Huawei
2014-11-13 5:58 ` Chen, Jing D
2014-11-14 6:30 ` Xie, Huawei
2014-11-14 7:24 ` Chen, Jing D
2014-11-12 22:34 ` [dpdk-dev] [PATCH 2/2] examples/vhost: use factorized default Rx/Tx configuration Huawei Xie
2014-11-13 6:02 ` Chen, Jing D
2014-11-14 2:17 ` Xie, Huawei [this message]
2014-11-12 22:52 ` [dpdk-dev] [PATCH 0/2] examples/vhost: support new VMDQ api and new nic i40e in vhost example Xie, Huawei
2014-12-06 10:16 ` Thomas Monjalon
2014-12-05 10:51 ` Fu, JingguoX
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=C37D651A908B024F974696C65296B57B0F302796@SHSMSX101.ccr.corp.intel.com \
--to=huawei.xie@intel.com \
--cc=dev@dpdk.org \
--cc=jing.d.chen@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).