patches for DPDK stable branches
 help / color / mirror / Atom feed
From: Jeff Guo <jia.guo@intel.com>
To: "Chen, BoX C" <box.c.chen@intel.com>,
	"Wang, ShougangX" <shougangx.wang@intel.com>,
	"dev@dpdk.org" <dev@dpdk.org>
Cc: "Xing, Beilei" <beilei.xing@intel.com>,
	"stable@dpdk.org" <stable@dpdk.org>
Subject: Re: [dpdk-stable] [dpdk-dev] [PATCH] net/i40e: fix incorrect hash look up table
Date: Sat, 18 Jul 2020 12:07:08 +0800
Message-ID: <1046de40-5938-8dea-b46b-e95cc64a88ba@intel.com> (raw)
In-Reply-To: <BN6PR11MB16674F432499443F98F1172CD97E0@BN6PR11MB1667.namprd11.prod.outlook.com>

hi, shougang

On 7/15/2020 5:07 PM, Chen, BoX C wrote:
> Tested-by: zhang,xi <xix.zhang@intel.com>
>
> Regards,
> Chen Bo
>
>> -----Original Message-----
>> From: dev <dev-bounces@dpdk.org> On Behalf Of Shougang Wang
>> Sent: July 15, 2020 14:35
>> To: dev@dpdk.org
>> Cc: Xing, Beilei <beilei.xing@intel.com>; Guo, Jia <jia.guo@intel.com>; Wang,
>> ShougangX <shougangx.wang@intel.com>; stable@dpdk.org
>> Subject: [dpdk-dev] [PATCH] net/i40e: fix incorrect hash look up table
>>
>> The hash look up table(LUT) will not be initializing when starting testpmd with
>> --disable-rss. So that some queues in LUT will be invalid if rx queues is less
>> than last time. When enable RSS by creating RSS rule, some packets will not
>> be into the valid queues.


What does "So that some queues in LUT will be invalid if rx queues is 
less than last time" means, Could you explain more clear?


>> This patch fixes this issue by initializing the LUT when creating an RSS rule.
>>
>> Fixes: feaae285b342 ("net/i40e: support hash configuration in RSS flow")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Shougang Wang <shougangx.wang@intel.com>
>> ---
>>   drivers/net/i40e/i40e_ethdev.c | 52
>> ++++++++++++++++++++++++++++++++++
>>   1 file changed, 52 insertions(+)
>>
>> diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
>> index 393b5320f..2a92bd8ef 100644
>> --- a/drivers/net/i40e/i40e_ethdev.c
>> +++ b/drivers/net/i40e/i40e_ethdev.c
>> @@ -13070,6 +13070,49 @@ i40e_rss_conf_init(struct
>> i40e_rte_flow_rss_conf *out,
>>   return 0;
>>   }
>>
>> +static int
>> +i40e_rss_init_lut(struct i40e_pf *pf)


I saw that this function is almost the same functional as 
i40e_rss_config_queue_region, could you check if it could be reused and 
merged one.


>> +{
>> +struct i40e_hw *hw = I40E_PF_TO_HW(pf);
>> +uint32_t lut = 0;
>> +uint16_t j, num;
>> +uint32_t i;
>> +
>> +/*
>> + * If both VMDQ and RSS enabled, not all of PF queues are configured.
>> + * It's necessary to calculate the actual PF queues that are configured.
>> + */
>> +if (pf->dev_data->dev_conf.rxmode.mq_mode &
>> ETH_MQ_RX_VMDQ_FLAG)
>> +num = i40e_pf_calc_configured_queues_num(pf);
>> +else
>> +num = pf->dev_data->nb_rx_queues;
>> +
>> +num = RTE_MIN(num, I40E_MAX_Q_PER_TC);
>> +PMD_INIT_LOG(INFO, "Max of contiguous %u PF queues are
>> configured",
>> +     num);
>> +
>> +if (num == 0) {
>> +PMD_INIT_LOG(ERR,
>> +"No PF queues are configured to enable RSS for
>> port %u",
>> +pf->dev_data->port_id);
>> +return -ENOTSUP;
>> +}
>> +
>> +if (pf->adapter->rss_reta_updated == 0) {
>> +for (i = 0, j = 0; i < hw->func_caps.rss_table_size; i++, j++) {
>> +if (j == num)
>> +j = 0;
>> +lut = (lut << 8) | (j & ((0x1 <<
>> +hw->func_caps.rss_table_entry_width) - 1));
>> +if ((i & 3) == 3)
>> +I40E_WRITE_REG(hw, I40E_PFQF_HLUT(i >>
>> 2),
>> +       rte_bswap32(lut));
>> +}
>> +}
>> +
>> +return 0;
>> +}
>> +
>>   /* Write HENA register to enable hash */  static int  i40e_rss_hash_set(struct
>> i40e_pf *pf, struct i40e_rte_flow_rss_conf *rss_conf) @@ -13318,12
>> +13361,21 @@ static int  i40e_rss_enable_hash(struct i40e_pf *pf,
>>   struct i40e_rte_flow_rss_conf *conf)
>>   {
>> +enum rte_eth_rx_mq_mode mq_mode =
>> +pf->dev_data->dev_conf.rxmode.mq_mode;
>>   struct i40e_rte_flow_rss_conf *rss_info = &pf->rss_info;
>>   struct i40e_rte_flow_rss_conf rss_conf;
>>
>>   if (!(conf->conf.types & pf->adapter->flow_types_mask))
>>   return -ENOTSUP;
>>
>> +/*
>> + * If the RSS is disabled before this, the LUT is uninitialized.
>> + * So it is necessary to initialize it here.
>> + */


Please don't use an empty /* line, use /* Comment....


>> +if (!(mq_mode & ETH_MQ_RX_RSS_FLAG) && !pf-
>>> rss_info.conf.queue_num)
>> +if (i40e_rss_init_lut(pf))
>> +return -ENOTSUP;


How can it know it is not support in function's caller,  better to use 
variable to restore the calling return and return the variable.


>> +
>>   memset(&rss_conf, 0, sizeof(rss_conf));
>>   rte_memcpy(&rss_conf, conf, sizeof(rss_conf));
>>
>> --
>> 2.17.1

  reply	other threads:[~2020-07-18  4:07 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-15  6:35 [dpdk-stable] " Shougang Wang
2020-07-15  9:07 ` [dpdk-stable] [dpdk-dev] " Chen, BoX C
2020-07-18  4:07   ` Jeff Guo [this message]
2020-07-21  5:49 ` [dpdk-stable] [PATCH v2] " Shougang Wang
2020-07-21  6:41   ` [dpdk-stable] [dpdk-dev] " Xie, WeiX
2020-07-22  5:50     ` Jeff Guo
2020-07-22  6:39       ` Wang, ShougangX
2020-07-22  5:31   ` Zhang, Qi Z
2020-07-22  6:20     ` Wang, ShougangX
2020-07-22  8:15 ` [dpdk-stable] [PATCH v3] " Shougang Wang
2020-07-23  1:57   ` [dpdk-stable] [dpdk-dev] " Yang, Qiming
2020-07-23 12:53     ` Zhang, Qi Z
2020-07-24  2:34       ` Wang, ShougangX
2020-07-24  2:47 ` [dpdk-stable] [PATCH v4] " Shougang Wang
2020-07-24  3:57   ` Jeff Guo
2020-07-24  4:49     ` Wang, ShougangX
2020-07-24  5:07   ` [dpdk-stable] [dpdk-dev] " Yang, Qiming
2020-07-24  5:42     ` Wang, ShougangX
2020-07-24  8:12 ` [dpdk-stable] [PATCH v5] " Shougang Wang
2020-07-24  9:00   ` [dpdk-stable] [dpdk-dev] " Xie, WeiX
2020-07-24  9:38 ` [dpdk-stable] [PATCH v6] " Shougang Wang
2020-07-24  9:54   ` Jeff Guo
2020-07-24  9:58     ` [dpdk-stable] [dpdk-dev] " 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=1046de40-5938-8dea-b46b-e95cc64a88ba@intel.com \
    --to=jia.guo@intel.com \
    --cc=beilei.xing@intel.com \
    --cc=box.c.chen@intel.com \
    --cc=dev@dpdk.org \
    --cc=shougangx.wang@intel.com \
    --cc=stable@dpdk.org \
    /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

patches for DPDK stable branches

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/stable/0 stable/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 stable stable/ https://inbox.dpdk.org/stable \
		stable@dpdk.org
	public-inbox-index stable

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


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