From: Stephen Hemminger <shemminger@vyatta.com>
To: Thomas Jarosch <thomas.jarosch@intra2net.com>
Cc: netdev@vger.kernel.org, Shreyas Bhatewara <sbhatewara@vmware.com>
Subject: Re: [PATCH] vmxnet3: Fix inconsistent LRO state after initialization
Date: Mon, 16 May 2011 10:14:44 -0700 [thread overview]
Message-ID: <20110516101444.29eec9dc@nehalam> (raw)
In-Reply-To: <201105161828.15237.thomas.jarosch@intra2net.com>
On Mon, 16 May 2011 18:28:15 +0200
Thomas Jarosch <thomas.jarosch@intra2net.com> wrote:
> During initialization of vmxnet3, the state of LRO
> gets out of sync with netdev->features.
>
> This leads to very poor TCP performance in a IP forwarding
> setup and is hitting many VMware users.
>
> Simplified call sequence:
> 1. vmxnet3_declare_features() initializes "adapter->lro" to true.
>
> 2. The kernel automatically disables LRO if IP forwarding is enabled,
> so vmxnet3_set_flags() gets called. This also updates netdev->features.
>
> 3. Now vmxnet3_setup_driver_shared() is called. "adapter->lro" is still
> set to true and LRO gets enabled again, even though
> netdev->features shows it's disabled.
>
> Fix it by updating "adapter->lro", too.
>
>
> The private vmxnet3 adapter flags are scheduled for removal
> in net-next, see commit a0d2730c9571aeba793cb5d3009094ee1d8fda35
> "net: vmxnet3: convert to hw_features".
>
> Patch applies to 2.6.37 / 2.6.38 and 2.6.39-rc6.
>
> Please CC: comments.
>
> Signed-off-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
>
> diff -u -r -p linux-2.6.37.i686/drivers/net/vmxnet3/vmxnet3_ethtool.c linux-2.6.37.lro/drivers/net/vmxnet3/vmxnet3_ethtool.c
> --- linux-2.6.37.i686/drivers/net/vmxnet3/vmxnet3_ethtool.c 2011-05-16 17:43:26.652081801 +0200
> +++ linux-2.6.37.lro/drivers/net/vmxnet3/vmxnet3_ethtool.c 2011-05-16 17:47:52.784081802 +0200
> @@ -287,6 +287,9 @@ vmxnet3_set_flags(struct net_device *net
> /* toggle the LRO feature*/
> netdev->features ^= NETIF_F_LRO;
>
> + /* Update private LRO flag */
> + adapter->lro = lro_requested;
> +
> /* update harware LRO capability accordingly */
> if (lro_requested)
> adapter->shared->devRead.misc.uptFeatures |=
Thanks for finding this. Vyatta ended up changing the default of adapter->lro
to workaround just this problem
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
--
next prev parent reply other threads:[~2011-05-16 17:14 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-05-16 16:28 [PATCH] vmxnet3: Fix inconsistent LRO state after initialization Thomas Jarosch
2011-05-16 17:14 ` Stephen Hemminger [this message]
2011-05-16 19:05 ` David Miller
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=20110516101444.29eec9dc@nehalam \
--to=shemminger@vyatta.com \
--cc=netdev@vger.kernel.org \
--cc=sbhatewara@vmware.com \
--cc=thomas.jarosch@intra2net.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).