From: Michal Kubecek <mkubecek@suse.cz>
To: netdev@vger.kernel.org
Cc: Kai-Heng Feng <kai.heng.feng@canonical.com>,
davem@davemloft.ne, jeffrey.t.kirsher@intel.com,
"David S. Miller" <davem@davemloft.net>,
Florian Fainelli <f.fainelli@gmail.com>,
Jiri Pirko <jiri@mellanox.com>,
Pablo Neira Ayuso <pablo@netfilter.org>,
Maxime Chevallier <maxime.chevallier@bootlin.com>,
Jakub Kicinski <jakub.kicinski@netronome.com>,
Li RongQing <lirongqing@baidu.com>,
open list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 2/2] ethtool: Call begin() and complete() in __ethtool_get_link_ksettings()
Date: Mon, 13 Jan 2020 13:51:29 +0100 [thread overview]
Message-ID: <20200113125129.GD25361@unicorn.suse.cz> (raw)
In-Reply-To: <20200110074159.18473-2-kai.heng.feng@canonical.com>
On Fri, Jan 10, 2020 at 03:41:59PM +0800, Kai-Heng Feng wrote:
> Device like igb gets runtime suspended when there's no link partner. We
> can't get correct speed under that state:
> $ cat /sys/class/net/enp3s0/speed
> 1000
>
> In addition to that, an error can also be spotted in dmesg:
> [ 385.991957] igb 0000:03:00.0 enp3s0: PCIe link lost
>
> It's because the igb device doesn't get runtime resumed before calling
> get_link_ksettings().
>
> So let's call begin() and complete() like what dev_ethtool() does, to
> runtime resume/suspend or power up/down the device properly.
>
> Once this fix is in place, igb can show the speed correctly without link
> partner:
> $ cat /sys/class/net/enp3s0/speed
> -1
I agree that we definitely should make sure ->begin() and ->complete()
are always called around ethtool_ops calls. But even if nesting should
be safe now (for in-tree drivers, that is), I'm not really happy about
calling them even in places where we positively know we are always
inside a begin / complete block as e.g. vxlan or net_failover do. (And
also linkinfo.c and linkmodes.c but it may be easier to call
->get_link_ksettings() directly there.)
How about having two helpers: one simple for "ethtool context" where we
know we already are between ->begin() and ->complete() and one with the
begin/complete calls for the rest? Another interesting question is if
any of the callers which do not have their own begin()/complete() wrap
does actually need anything more than speed and duplex (I didn't do
a full check).
Michal
> Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> ---
> net/ethtool/ioctl.c | 15 ++++++++++++++-
> 1 file changed, 14 insertions(+), 1 deletion(-)
>
> diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c
> index 182bffbffa78..c768dbf45fc4 100644
> --- a/net/ethtool/ioctl.c
> +++ b/net/ethtool/ioctl.c
> @@ -423,13 +423,26 @@ struct ethtool_link_usettings {
> int __ethtool_get_link_ksettings(struct net_device *dev,
> struct ethtool_link_ksettings *link_ksettings)
> {
> + int rc;
> +
> ASSERT_RTNL();
>
> if (!dev->ethtool_ops->get_link_ksettings)
> return -EOPNOTSUPP;
>
> + if (dev->ethtool_ops->begin) {
> + rc = dev->ethtool_ops->begin(dev);
> + if (rc < 0)
> + return rc;
> + }
> +
> memset(link_ksettings, 0, sizeof(*link_ksettings));
> - return dev->ethtool_ops->get_link_ksettings(dev, link_ksettings);
> + rc = dev->ethtool_ops->get_link_ksettings(dev, link_ksettings);
> +
> + if (dev->ethtool_ops->complete)
> + dev->ethtool_ops->complete(dev);
> +
> + return rc;
> }
> EXPORT_SYMBOL(__ethtool_get_link_ksettings);
>
> --
> 2.17.1
>
next prev parent reply other threads:[~2020-01-13 12:51 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-10 7:41 [PATCH 1/2] igb: Use device_lock() insead of rtnl_lock() Kai-Heng Feng
2020-01-10 7:41 ` [PATCH 2/2] ethtool: Call begin() and complete() in __ethtool_get_link_ksettings() Kai-Heng Feng
2020-01-13 12:51 ` Michal Kubecek [this message]
2020-01-14 10:06 ` Kai-Heng Feng
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=20200113125129.GD25361@unicorn.suse.cz \
--to=mkubecek@suse.cz \
--cc=davem@davemloft.ne \
--cc=davem@davemloft.net \
--cc=f.fainelli@gmail.com \
--cc=jakub.kicinski@netronome.com \
--cc=jeffrey.t.kirsher@intel.com \
--cc=jiri@mellanox.com \
--cc=kai.heng.feng@canonical.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lirongqing@baidu.com \
--cc=maxime.chevallier@bootlin.com \
--cc=netdev@vger.kernel.org \
--cc=pablo@netfilter.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).