From: "Wang, Haiyue" <haiyue.wang@intel.com>
To: "Yang, Qiming" <qiming.yang@intel.com>,
"dev@dpdk.org" <dev@dpdk.org>,
"Zhang, Qi Z" <qi.z.zhang@intel.com>,
"Wu, Jingjing" <jingjing.wu@intel.com>,
"Lu, Wenzhuo" <wenzhuo.lu@intel.com>
Subject: Re: [dpdk-dev] [PATCH v1] net/ice: update the RSS RETA size with support values
Date: Tue, 21 May 2019 06:17:10 +0000 [thread overview]
Message-ID: <E3B9F2FDCB65864C82CD632F23D8AB877335E616@SHSMSX101.ccr.corp.intel.com> (raw)
In-Reply-To: <F5DF4F0E3AFEF648ADC1C3C33AD4DBF17A4BF1B9@SHSMSX101.ccr.corp.intel.com>
> -----Original Message-----
> From: Yang, Qiming
> Sent: Tuesday, May 21, 2019 13:50
> To: Wang, Haiyue <haiyue.wang@intel.com>; dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Wu,
> Jingjing <jingjing.wu@intel.com>; Lu, Wenzhuo <wenzhuo.lu@intel.com>
> Subject: RE: [PATCH v1] net/ice: update the RSS RETA size with support values
>
> Little comments.
>
> > -----Original Message-----
> > From: Wang, Haiyue
> > Sent: Monday, May 20, 2019 4:15 PM
> > To: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Yang, Qiming
> > <qiming.yang@intel.com>; Wu, Jingjing <jingjing.wu@intel.com>; Lu,
> > Wenzhuo <wenzhuo.lu@intel.com>
> > Cc: Wang, Haiyue <haiyue.wang@intel.com>
> > Subject: [PATCH v1] net/ice: update the RSS RETA size with support values
> >
> > Since ice can support 128, 512, 2K RSS RETA size value, change the update
> > API to set it to resize the RSS RETA table. And by default, use 512 to sync
> > with ETH_RSS_RETA_SIZE_x maximum value definition.
> > Also the flag ICE_FLAG_RSS_AQ_CAPABLE is missed to set.
> >
> > Fixes: 690175ee51bf ("net/ice: support getting device information")
> > Fixes: ff963bfa7cb1 ("net/ice: support RSS")
> >
> > Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> > ---
> > drivers/net/ice/ice_ethdev.c | 41 +++++++++++++++++++++++++++--------------
> > 1 file changed, 27 insertions(+), 14 deletions(-)
> >
> > diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c index
> > bbaa7cf..c4ea09f 100644
> > --- a/drivers/net/ice/ice_ethdev.c
> > +++ b/drivers/net/ice/ice_ethdev.c
> > @@ -1149,6 +1149,12 @@ ice_setup_vsi(struct ice_pf *pf, enum ice_vsi_type
> > type)
> > TAILQ_INIT(&vsi->mac_list);
> > TAILQ_INIT(&vsi->vlan_list);
> >
> > + /* Be sync with ETH_RSS_RETA_SIZE_x maximum value definition */
>
> I think this stats is not clear, can't understand what's the ETH_RSS_RETA_SIZE_x maximum value, why
> don't say the default max
> Value is 512?
>
The original plan was to add ETH_RSS_RETA_SIZE_2048, but found that testpmd use 512 as maximum in many
places, so just write down some comment here to sync with testpmd. And it doesn't say 512 is default,
just because it is used in many places. If ETH_RSS_RETA_SIZE_2048 is added, we can sync with it later.
> > + pf->hash_lut_size = hw->func_caps.common_cap.rss_table_size >
> > + ETH_RSS_RETA_SIZE_512 ? ETH_RSS_RETA_SIZE_512 :
> > + hw->func_caps.common_cap.rss_table_size;
> > + pf->flags |= ICE_FLAG_RSS_AQ_CAPABLE;
> > +
> > memset(&vsi_ctx, 0, sizeof(vsi_ctx));
> > /* base_queue in used in queue mapping of VSI add/update
> > command.
> > * Suppose vsi->base_queue is 0 now, don't consider SRIOV, VMDQ
> > @@ -1627,7 +1633,7 @@ static int ice_init_rss(struct ice_pf *pf)
> > rss_conf = &dev->data->dev_conf.rx_adv_conf.rss_conf;
> > nb_q = dev->data->nb_rx_queues;
> > vsi->rss_key_size = ICE_AQC_GET_SET_RSS_KEY_DATA_RSS_KEY_SIZE;
> > - vsi->rss_lut_size = hw->func_caps.common_cap.rss_table_size;
> > + vsi->rss_lut_size = pf->hash_lut_size;
> >
> > if (is_safe_mode) {
> > PMD_DRV_LOG(WARNING, "RSS is not supported in safe
> > mode\n"); @@ -2033,7 +2039,7 @@ ice_dev_info_get(struct rte_eth_dev
> > *dev, struct rte_eth_dev_info *dev_info)
> > dev_info->rx_queue_offload_capa = 0;
> > dev_info->tx_queue_offload_capa = 0;
> >
> > - dev_info->reta_size = hw->func_caps.common_cap.rss_table_size;
> > + dev_info->reta_size = pf->hash_lut_size;
> > dev_info->hash_key_size = (VSIQF_HKEY_MAX_INDEX + 1) *
> > sizeof(uint32_t);
> >
> > dev_info->default_rxconf = (struct rte_eth_rxconf) { @@ -2605,28
> > +2611,31 @@ ice_rss_reta_update(struct rte_eth_dev *dev,
> > uint16_t reta_size)
> > {
> > struct ice_pf *pf = ICE_DEV_PRIVATE_TO_PF(dev->data->dev_private);
> > - struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data-
> > >dev_private);
> > - uint16_t i, lut_size = hw->func_caps.common_cap.rss_table_size;
> > + uint16_t i, lut_size = pf->hash_lut_size;
> > uint16_t idx, shift;
> > uint8_t *lut;
> > int ret;
> >
> > - if (reta_size != lut_size ||
> > - reta_size > ETH_RSS_RETA_SIZE_512) {
> > + if (reta_size != ICE_AQC_GSET_RSS_LUT_TABLE_SIZE_128 &&
> > + reta_size != ICE_AQC_GSET_RSS_LUT_TABLE_SIZE_512 &&
> > + reta_size != ICE_AQC_GSET_RSS_LUT_TABLE_SIZE_2K) {
> > PMD_DRV_LOG(ERR,
> > "The size of hash lookup table configured (%d)"
> > "doesn't match the number hardware can "
> > - "supported (%d)",
> > - reta_size, lut_size);
> > + "supported (128, 512, 2048)",
> > + reta_size);
> > return -EINVAL;
> > }
> >
> > - lut = rte_zmalloc(NULL, reta_size, 0);
> > + /* It MUST use the current LUT size to get the RSS lookup table,
> > + * otherwise if will fail with -100 error code.
> > + */
>
> I think it's no need too detail explain. /* use current size to create lookup table*/
>
In fact, this RTE_MAX 'lut' memory content is used to get the original lookup table
firstly with original size, then use part of them to set new. Normally, people may
think use the new size can get the part of the original table contents, just record
the debug result here to make it clean why we change like this.
> > + lut = rte_zmalloc(NULL, RTE_MAX(reta_size, lut_size), 0);
> > if (!lut) {
> > PMD_DRV_LOG(ERR, "No memory can be allocated");
> > return -ENOMEM;
> > }
> > - ret = ice_get_rss_lut(pf->main_vsi, lut, reta_size);
> > + ret = ice_get_rss_lut(pf->main_vsi, lut, lut_size);
> > if (ret)
> > goto out;
> >
> > @@ -2637,6 +2646,12 @@ ice_rss_reta_update(struct rte_eth_dev *dev,
> > lut[i] = reta_conf[idx].reta[shift];
> > }
> > ret = ice_set_rss_lut(pf->main_vsi, lut, reta_size);
> > + if (ret == 0 && lut_size != reta_size) {
> > + PMD_DRV_LOG(INFO,
> > + "The size of hash lookup table is changed from (%d)
> > to (%d)",
> > + lut_size, reta_size);
> > + pf->hash_lut_size = reta_size;
> > + }
> >
> > out:
> > rte_free(lut);
> > @@ -2650,14 +2665,12 @@ ice_rss_reta_query(struct rte_eth_dev *dev,
> > uint16_t reta_size)
> > {
> > struct ice_pf *pf = ICE_DEV_PRIVATE_TO_PF(dev->data->dev_private);
> > - struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data-
> > >dev_private);
> > - uint16_t i, lut_size = hw->func_caps.common_cap.rss_table_size;
> > + uint16_t i, lut_size = pf->hash_lut_size;
> > uint16_t idx, shift;
> > uint8_t *lut;
> > int ret;
> >
> > - if (reta_size != lut_size ||
> > - reta_size > ETH_RSS_RETA_SIZE_512) {
> > + if (reta_size != lut_size) {
> > PMD_DRV_LOG(ERR,
> > "The size of hash lookup table configured (%d)"
> > "doesn't match the number hardware can "
> > --
> > 2.7.4
next prev parent reply other threads:[~2019-05-21 6:17 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-20 8:14 Haiyue Wang
2019-05-21 5:50 ` Yang, Qiming
2019-05-21 6:17 ` Wang, Haiyue [this message]
2019-05-21 6:45 ` Yang, Qiming
2019-05-21 9:00 ` Zhang, Qi Z
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=E3B9F2FDCB65864C82CD632F23D8AB877335E616@SHSMSX101.ccr.corp.intel.com \
--to=haiyue.wang@intel.com \
--cc=dev@dpdk.org \
--cc=jingjing.wu@intel.com \
--cc=qi.z.zhang@intel.com \
--cc=qiming.yang@intel.com \
--cc=wenzhuo.lu@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).