patches for DPDK stable branches
 help / color / mirror / Atom feed
From: "Jiawen Wu" <jiawenwu@trustnetic.com>
To: "'Ferruh Yigit'" <ferruh.yigit@intel.com>, <dev@dpdk.org>
Cc: <stable@dpdk.org>
Subject: Re: [dpdk-stable] [PATCH v2 1/5] net/txgbe: fix RSS in QINQ
Date: Thu, 6 May 2021 11:33:17 +0800
Message-ID: <002601d74228$8dfc5b90$a9f512b0$@trustnetic.com> (raw)
In-Reply-To: <1fd58ff8-5028-2216-eec7-1437630de896@intel.com>

On April 29, 2021 10:31 PM, Ferruh Yigit wrote:
> On 4/29/2021 11:33 AM, Jiawen Wu wrote:
> > Support to enable and disable QINQ hardware strip, when configure vlan
> > offload with QINQ strip mask, to avoid RSS does not work for QINQ
> > packets.
> >
> 
> Hi Jiawen,
> 
> What was not working and fixed here?
> Is it RSS hash calculation is wrong when packet has double VLAN tag? Is it
> failing to calculate hash for VLAN field of the packet or calculation for any fields
> are wrong? Can't device detect/parse QinQ fields?
> 
> And how enabling QinQ strip is solving the issue? Should user enable QinQ strip
> before configuring the RSS? What happens if user don't, should driver has
> checks to cover this, like fail to enable RSS if QinQ strip is not enable etc?
> 
> Can you please provide more details?
> 

Hi Ferruh,

I think I might be mistaken something.
At first I fixed this bug because RSS did not take effect when packet has QINQ tag.
Hardware does not support to calculate VLAN field, but the insertion of QINQ causes five-tuple filed changes.
The root cause is that QINQ is not properly stripped. So I add the functions to handle QINQ stripping when ETH_QINQ_STRIP_MASK set.
And users should enable QINQ strip before configuring the RSS.

This patch should be renamed to "fix to strip QINQ", is this more appropriate?

> 
> > Fixes: 220b0e49bc47 ("net/txgbe: support VLAN")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
> > ---
> >  drivers/net/txgbe/txgbe_ethdev.c | 39
> > +++++++++++++++++++++++++++-----
> >  1 file changed, 33 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/net/txgbe/txgbe_ethdev.c
> > b/drivers/net/txgbe/txgbe_ethdev.c
> > index 97796f040b..3d6d356102 100644
> > --- a/drivers/net/txgbe/txgbe_ethdev.c
> > +++ b/drivers/net/txgbe/txgbe_ethdev.c
> > @@ -1209,7 +1209,6 @@ txgbe_vlan_hw_extend_disable(struct
> rte_eth_dev
> > *dev)
> >
> >  	ctrl = rd32(hw, TXGBE_PORTCTL);
> >  	ctrl &= ~TXGBE_PORTCTL_VLANEXT;
> > -	ctrl &= ~TXGBE_PORTCTL_QINQ;
> >  	wr32(hw, TXGBE_PORTCTL, ctrl);
> >  }
> >
> > @@ -1217,17 +1216,38 @@ static void
> >  txgbe_vlan_hw_extend_enable(struct rte_eth_dev *dev)  {
> >  	struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
> > -	struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
> > -	struct rte_eth_txmode *txmode = &dev->data->dev_conf.txmode;
> >  	uint32_t ctrl;
> >
> >  	PMD_INIT_FUNC_TRACE();
> >
> >  	ctrl  = rd32(hw, TXGBE_PORTCTL);
> >  	ctrl |= TXGBE_PORTCTL_VLANEXT;
> > -	if (rxmode->offloads & DEV_RX_OFFLOAD_QINQ_STRIP ||
> > -	    txmode->offloads & DEV_TX_OFFLOAD_QINQ_INSERT)
> > -		ctrl |= TXGBE_PORTCTL_QINQ;
> > +	wr32(hw, TXGBE_PORTCTL, ctrl);
> > +}
> > +
> > +static void
> > +txgbe_qinq_hw_strip_disable(struct rte_eth_dev *dev) {
> > +	struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
> > +	uint32_t ctrl;
> > +
> > +	PMD_INIT_FUNC_TRACE();
> > +
> > +	ctrl = rd32(hw, TXGBE_PORTCTL);
> > +	ctrl &= ~TXGBE_PORTCTL_QINQ;
> > +	wr32(hw, TXGBE_PORTCTL, ctrl);
> > +}
> > +
> > +static void
> > +txgbe_qinq_hw_strip_enable(struct rte_eth_dev *dev) {
> > +	struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
> > +	uint32_t ctrl;
> > +
> > +	PMD_INIT_FUNC_TRACE();
> > +
> > +	ctrl  = rd32(hw, TXGBE_PORTCTL);
> > +	ctrl |= TXGBE_PORTCTL_QINQ | TXGBE_PORTCTL_VLANEXT;
> >  	wr32(hw, TXGBE_PORTCTL, ctrl);
> >  }
> >
> > @@ -1294,6 +1314,13 @@ txgbe_vlan_offload_config(struct rte_eth_dev
> *dev, int mask)
> >  			txgbe_vlan_hw_extend_disable(dev);
> >  	}
> >
> > +	if (mask & ETH_QINQ_STRIP_MASK) {
> > +		if (rxmode->offloads & DEV_RX_OFFLOAD_QINQ_STRIP)
> > +			txgbe_qinq_hw_strip_enable(dev);
> > +		else
> > +			txgbe_qinq_hw_strip_disable(dev);
> > +	}
> > +
> >  	return 0;
> >  }
> >
> >





  reply	other threads:[~2021-05-06  3:33 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20210429103335.23060-1-jiawenwu@trustnetic.com>
2021-04-29 10:33 ` Jiawen Wu
2021-04-29 14:31   ` Ferruh Yigit
2021-05-06  3:33     ` Jiawen Wu [this message]
2021-04-29 10:33 ` [dpdk-stable] [PATCH v2 2/5] net/txgbe: fix VF MTU limit setting Jiawen Wu
2021-04-29 14:59   ` Ferruh Yigit
2021-05-06  1:43     ` Jiawen Wu

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='002601d74228$8dfc5b90$a9f512b0$@trustnetic.com' \
    --to=jiawenwu@trustnetic.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@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