DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Dai, Wei" <wei.dai@intel.com>
To: "Su, David W" <david.w.su@intel.com>, "dev@dpdk.org" <dev@dpdk.org>
Cc: "Su, David W" <david.w.su@intel.com>
Subject: Re: [dpdk-dev] [PATCH] net/ixgbe: keep interrupt throttling disabled
Date: Thu, 30 Mar 2017 14:28:44 +0000	[thread overview]
Message-ID: <49759EB36A64CF4892C1AFEC9231E8D650A23B96@PGSMSX101.gar.corp.intel.com> (raw)
In-Reply-To: <1484953825-3203-1-git-send-email-david.w.su@intel.com>

Hi, David

The code to be removed set the minimum interrupt interval of queue 0 to 0x79E.
According to datasheet, this value has something wrong because it wirte the least 
reserved 3 bits to 110. Anyway, it intends to set a non-zero minimum interrupt interval.

ixgbe_configure_msix( ) is only called by ixgbe_dev_start( ).
After it is called, ixgbe_dev_rx_ini( ) is called.
And ixgbe_set_rsc( ) is called in ixgbe_dev_rx_init( ).
In ixgbe_set_rsc( ), if RSC is enabled, the ITR (interrupt interval) is set to 500us again.

So I think enabling immediate interrupt response by removing these codes is not suitable.
It should be done in other code where interrupt mode Rx is configured.

> Subject: [dpdk-dev] [PATCH] net/ixgbe: keep interrupt throttling disabled
> 
> The code being removed would enable time-based interrupt throttling for the
> first MSI-X interrupt vector.  This would throttle link status change interrupts
> for NICs bound to vfio_pci driver; but for igb_uio driver, rx queue 0 interrupts
> would be throttled.  This resulted in inconsistent latencies in a DPDK interrupt
> mode packet forwarding application between the 2 drivers.
> 
> We'd like DPDK interrupt mode applications to be awaken by inbound packet
> interrupts as soon as they are received regardless of interrupt interval,
> interrupt throttling should be kept disabled.
> 
> Signed-off-by: David Su <david.w.su@intel.com>
> ---
>  drivers/net/ixgbe/ixgbe_ethdev.c |    2 --
>  1 files changed, 0 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c
> b/drivers/net/ixgbe/ixgbe_ethdev.c
> index bdf4e2b..fdfb7a6 100644
> --- a/drivers/net/ixgbe/ixgbe_ethdev.c
> +++ b/drivers/net/ixgbe/ixgbe_ethdev.c
> @@ -5815,8 +5815,6 @@ ixgbe_configure_msix(struct rte_eth_dev *dev)
>  	default:
>  		break;
>  	}
> -	IXGBE_WRITE_REG(hw, IXGBE_EITR(IXGBE_MISC_VEC_ID),
> -			IXGBE_MIN_INTER_INTERRUPT_INTERVAL_DEFAULT & 0xFFF);
> 
>  	/* set up to autoclear timer, and the vectors */
>  	mask = IXGBE_EIMS_ENABLE_MASK;
> --
> 1.7.0.4

      reply	other threads:[~2017-03-30 14:28 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-01-20 23:10 David Su
2017-03-30 14:28 ` Dai, Wei [this message]

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=49759EB36A64CF4892C1AFEC9231E8D650A23B96@PGSMSX101.gar.corp.intel.com \
    --to=wei.dai@intel.com \
    --cc=david.w.su@intel.com \
    --cc=dev@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
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).