DPDK patches and discussions
 help / color / mirror / Atom feed
From: Tiwei Bie <tiwei.bie@intel.com>
To: Maxime Coquelin <maxime.coquelin@redhat.com>
Cc: zhihong.wang@intel.com, dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH 4/4] net/virtio: improve offload check performance
Date: Mon, 4 Jun 2018 19:55:07 +0800	[thread overview]
Message-ID: <20180604115507.GA21406@debian> (raw)
In-Reply-To: <20180601124758.22652-5-maxime.coquelin@redhat.com>

On Fri, Jun 01, 2018 at 02:47:58PM +0200, Maxime Coquelin wrote:
> Instead of checking the multiple Virtio features bits for
> every packet, let's do the check once at configure time and
> store it in virtio_hw struct.
> 
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> ---
>  drivers/net/virtio/virtio_ethdev.c | 19 +++++++++++++++++++
>  drivers/net/virtio/virtio_pci.h    |  2 ++
>  drivers/net/virtio/virtio_rxtx.c   | 29 ++++++-----------------------
>  3 files changed, 27 insertions(+), 23 deletions(-)
> 
> diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c
> index d481b282e..981e0994a 100644
> --- a/drivers/net/virtio/virtio_ethdev.c
> +++ b/drivers/net/virtio/virtio_ethdev.c
> @@ -1790,6 +1790,22 @@ rte_virtio_pmd_init(void)
>  	rte_pci_register(&rte_virtio_pmd);
>  }
>  
> +static inline int

Maybe it's better to return bool.

> +rx_offload_enabled(struct virtio_hw *hw)
> +{
> +	return vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM) ||
> +		vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
> +		vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6);
> +}
> +
> +static inline int

Ditto.

> +tx_offload_enabled(struct virtio_hw *hw)
> +{
> +	return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
> +		vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
> +		vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
> +}
> +
>  /*
>   * Configure virtio device
>   * It returns 0 on success.
> @@ -1869,6 +1885,9 @@ virtio_dev_configure(struct rte_eth_dev *dev)
>  		return -ENOTSUP;
>  	}
>  
> +	hw->has_tx_offload = !!tx_offload_enabled(hw);
> +	hw->has_rx_offload = !!rx_offload_enabled(hw);
> +
>  	if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
>  		/* Enable vector (0) for Link State Intrerrupt */
>  		if (VTPCI_OPS(hw)->set_config_irq(hw, 0) ==
> diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
> index a28ba8339..e0bb871f2 100644
> --- a/drivers/net/virtio/virtio_pci.h
> +++ b/drivers/net/virtio/virtio_pci.h
> @@ -233,6 +233,8 @@ struct virtio_hw {
>  	uint8_t     modern;
>  	uint8_t     use_simple_rx;
>  	uint8_t     use_simple_tx;
> +	uint8_t		has_tx_offload;
> +	uint8_t		has_rx_offload;

I think it's better to use spaces, instead of
4 spaces width tabs after uint8_t.

>  	uint16_t    port_id;
>  	uint8_t     mac_addr[ETHER_ADDR_LEN];
>  	uint32_t    notify_off_multiplier;
> diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
> index 92fab2174..3f113a118 100644
> --- a/drivers/net/virtio/virtio_rxtx.c
> +++ b/drivers/net/virtio/virtio_rxtx.c
> @@ -225,13 +225,6 @@ virtio_tso_fix_cksum(struct rte_mbuf *m)
>  	}
>  }
>  
> -static inline int
> -tx_offload_enabled(struct virtio_hw *hw)
> -{
> -	return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
> -		vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
> -		vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
> -}
>  
>  /* avoid write operation when necessary, to lessen cache issues */
>  #define ASSIGN_UNLESS_EQUAL(var, val) do {	\
> @@ -253,13 +246,15 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, struct rte_mbuf *cookie,
>  	struct virtio_net_hdr *hdr;
>  	int offload;
>  
> -	offload = tx_offload_enabled(vq->hw);
>  	head_idx = vq->vq_desc_head_idx;
>  	idx = head_idx;
>  	dxp = &vq->vq_descx[idx];
>  	dxp->cookie = (void *)cookie;
>  	dxp->ndescs = needed;
>  
> +	offload = vq->hw->has_tx_offload &&
> +		(cookie->ol_flags & PKT_TX_OFFLOAD_MASK);

If features aren't negotiated, I think there is no need to
check ol_flags and update the net header.

> +
>  	start_dp = vq->vq_ring.desc;
>  
>  	if (can_push) {
> @@ -270,7 +265,6 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, struct rte_mbuf *cookie,
>  		 * which is wrong. Below subtract restores correct pkt size.
>  		 */
>  		cookie->pkt_len -= head_size;
> -		/* if offload disabled, it is not zeroed below, do it now */
>  		if (offload == 0) {
>  			ASSIGN_UNLESS_EQUAL(hdr->csum_start, 0);
>  			ASSIGN_UNLESS_EQUAL(hdr->csum_offset, 0);
> @@ -686,14 +680,6 @@ virtio_rx_offload(struct rte_mbuf *m, struct virtio_net_hdr *hdr)
>  	return 0;
>  }
[...]

Best regards,
Tiwei Bie

  reply	other threads:[~2018-06-04 11:54 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-01 12:47 [dpdk-dev] [PATCH 0/4] net/virtio: Tx path selection and offload improvements Maxime Coquelin
2018-06-01 12:47 ` [dpdk-dev] [PATCH 1/4] net/virtio: use simple path for Tx even if Rx mergeable Maxime Coquelin
2018-06-01 12:47 ` [dpdk-dev] [PATCH 2/4] net/vhost: improve Tx path selection Maxime Coquelin
2018-06-04 12:25   ` Tiwei Bie
2018-06-01 12:47 ` [dpdk-dev] [PATCH 3/4] net/virtio: don't use simple Rx path if TCP LRO requested Maxime Coquelin
2018-06-04 11:59   ` Tiwei Bie
2018-06-04 14:32     ` Maxime Coquelin
2018-06-05 13:39       ` Maxime Coquelin
2018-06-01 12:47 ` [dpdk-dev] [PATCH 4/4] net/virtio: improve offload check performance Maxime Coquelin
2018-06-04 11:55   ` Tiwei Bie [this message]
2018-06-04 14:29     ` Maxime Coquelin
2018-06-05  3:10       ` Tiwei Bie
2018-06-05  9:43         ` Maxime Coquelin
2018-06-05 11:20           ` Tiwei Bie
2018-06-05 11:58             ` Maxime Coquelin
2018-06-05 12:21               ` Tiwei Bie
2018-06-04  7:42 ` [dpdk-dev] [PATCH 0/4] net/virtio: Tx path selection and offload improvements Maxime Coquelin

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=20180604115507.GA21406@debian \
    --to=tiwei.bie@intel.com \
    --cc=dev@dpdk.org \
    --cc=maxime.coquelin@redhat.com \
    --cc=zhihong.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).