From: Maxime Coquelin <maxime.coquelin@redhat.com>
To: Tiwei Bie <tiwei.bie@intel.com>
Cc: zhihong.wang@intel.com, dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v2 3/5] net/vhost: improve Tx path selection
Date: Thu, 7 Jun 2018 09:34:53 +0200 [thread overview]
Message-ID: <f37e4a12-1acd-9863-f89b-56fc420b95ac@redhat.com> (raw)
In-Reply-To: <20180607051321.GC32696@debian>
On 06/07/2018 07:13 AM, Tiwei Bie wrote:
> On Wed, Jun 06, 2018 at 02:31:26PM +0200, Maxime Coquelin wrote:
> [...]
>> @@ -1886,6 +1888,15 @@ virtio_dev_configure(struct rte_eth_dev *dev)
>> (1ULL << VIRTIO_NET_F_GUEST_TSO4) |
>> (1ULL << VIRTIO_NET_F_GUEST_TSO6);
>>
>> + if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
>> + DEV_TX_OFFLOAD_UDP_CKSUM))
>> + req_features |= (1ULL << VIRTIO_NET_F_CSUM);
>
> Hmm.. I still prefer to keep the DEV_TX_OFFLOAD_TCP_CKSUM
> and DEV_TX_OFFLOAD_UDP_CKSUM aligned to make the coding
> style consistent with the existing code of Rx:
>
> 1901 ¦ if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM |
> 1902 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_CKSUM))
> 1903 ¦ ¦ req_features |= (1ULL << VIRTIO_NET_F_GUEST_CSUM);
>
> But it's up to you.
Yes, it is better to keep same alignment, even if the rx one looks a bit
weird to me.
Also, I take the opportunity to switch UDP and TXP CKSUM defines for
consistency with Rx.
>> +
>> + if (tx_offloads & DEV_TX_OFFLOAD_TCP_TSO)
>> + req_features |=
>> + (1ULL << VIRTIO_NET_F_HOST_TSO4) |
>> + (1ULL << VIRTIO_NET_F_HOST_TSO6);
>> +
>> /* if request features changed, reinit the device */
>> if (req_features != hw->req_guest_features) {
>> ret = virtio_init_device(dev, req_features);
>> @@ -1955,6 +1966,12 @@ virtio_dev_configure(struct rte_eth_dev *dev)
>> DEV_RX_OFFLOAD_TCP_CKSUM))
>> hw->use_simple_rx = 0;
>>
>> + if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
>> + DEV_TX_OFFLOAD_UDP_CKSUM |
>> + DEV_TX_OFFLOAD_TCP_TSO |
>> + DEV_TX_OFFLOAD_VLAN_INSERT))
>> + hw->use_simple_tx = 0;
>
> Ditto. Below is the code for Rx:
>
> 1987 ¦ if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM |
> 1988 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_CKSUM |
> 1989 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_LRO |
> 1990 ¦ ¦ ¦ DEV_RX_OFFLOAD_VLAN_STRIP))
> 1991 ¦ ¦ hw->use_simple_rx = 0;
>
> So, instead of:
>
> 1987 ¦ if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM |
> 1988 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_CKSUM |
> 1989 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_LRO |
> 1990 ¦ ¦ ¦ DEV_RX_OFFLOAD_VLAN_STRIP))
> 1991 ¦ ¦ hw->use_simple_rx = 0;
> 1992
> 1993 ¦ if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
> 1994 ¦ ¦ ¦ ¦ DEV_TX_OFFLOAD_UDP_CKSUM |
> 1995 ¦ ¦ ¦ ¦ DEV_TX_OFFLOAD_TCP_TSO |
> 1996 ¦ ¦ ¦ ¦ DEV_TX_OFFLOAD_VLAN_INSERT))
> 1997 ¦ ¦ hw->use_simple_tx = 0;
>
> I would prefer:
>
> 1987 ¦ if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM |
> 1988 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_CKSUM |
> 1989 ¦ ¦ ¦ DEV_RX_OFFLOAD_TCP_LRO |
> 1990 ¦ ¦ ¦ DEV_RX_OFFLOAD_VLAN_STRIP))
> 1991 ¦ ¦ hw->use_simple_rx = 0;
> 1992
> 1993 ¦ if (tx_offloads & (DEV_TX_OFFLOAD_UDP_CKSUM |
> 1994 ¦ ¦ ¦ DEV_TX_OFFLOAD_TCP_CKSUM |
> 1995 ¦ ¦ ¦ DEV_TX_OFFLOAD_TCP_TSO |
> 1996 ¦ ¦ ¦ DEV_TX_OFFLOAD_VLAN_INSERT))
> 1997 ¦ ¦ hw->use_simple_tx = 0;
>
> But anyway, it's up to you. :)
Yep, I fix it now.
> Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
Thanks,
Maxime
> Best regards,
> Tiwei Bie
>
next prev parent reply other threads:[~2018-06-07 7:34 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-06 12:31 [dpdk-dev] [PATCH v2 0/5] net/virtio: Tx path selection and offload improvements Maxime Coquelin
2018-06-06 12:31 ` [dpdk-dev] [PATCH v2 1/5] net/virtio: prevent simple Tx path selection by default Maxime Coquelin
2018-06-07 5:43 ` Tiwei Bie
2018-06-07 7:40 ` Maxime Coquelin
2018-06-07 7:53 ` Tiwei Bie
2018-06-06 12:31 ` [dpdk-dev] [PATCH v2 2/5] net/virtio: use simple path for Tx even if Rx mergeable Maxime Coquelin
2018-06-07 5:18 ` Tiwei Bie
2018-06-06 12:31 ` [dpdk-dev] [PATCH v2 3/5] net/vhost: improve Tx path selection Maxime Coquelin
2018-06-07 5:13 ` Tiwei Bie
2018-06-07 7:34 ` Maxime Coquelin [this message]
2018-06-06 12:31 ` [dpdk-dev] [PATCH v2 4/5] net/virtio: don't use simple Rx if TCP LRO or VLAN strip requested Maxime Coquelin
2018-06-07 4:58 ` Tiwei Bie
2018-06-06 12:31 ` [dpdk-dev] [PATCH v2 5/5] net/virtio: improve offload check performance Maxime Coquelin
2018-06-07 4:51 ` Tiwei Bie
2018-06-07 7:22 ` 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=f37e4a12-1acd-9863-f89b-56fc420b95ac@redhat.com \
--to=maxime.coquelin@redhat.com \
--cc=dev@dpdk.org \
--cc=tiwei.bie@intel.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).