Re: [PATCH v2 0/3] net/ncsi: Add NCSI Intel OEM command to keep PHY link up

From: Paul Fertser <fercerpav@gmail.com>
To: Ivan Mikhaylov <i.mikhaylov@yadro.com>
Cc: "David S . Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	Samuel Mendoza-Jonas <sam@mendozajonas.com>,
	Joel Stanley <joel@jms.id.au>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
	openbmc@lists.ozlabs.org
Subject: Re: [PATCH v2 0/3] net/ncsi: Add NCSI Intel OEM command to keep PHY link up
Date: Tue, 20 Jul 2021 12:53:20 +0300	[thread overview]
Message-ID: <20210720095320.GB4789@home.paul.comp> (raw)
In-Reply-To: <20210708122754.555846-1-i.mikhaylov@yadro.com>

Hello,

On Thu, Jul 08, 2021 at 03:27:51PM +0300, Ivan Mikhaylov wrote:
> Add NCSI Intel OEM command to keep PHY link up and prevents any channel
> resets during the host load on i210.

There're multiple things to consider here and I have hesitations about
the way you propose to solve the issue.

While the host is booted up and fully functional it assumes it has
full proper control of network cards, and sometimes it really needs to
reset them to e.g. recover from crashed firmware. The PHY resets might
also make sense in certain cases, and so in general having this "link
up" bit set all the time might be breaking assumptions.

As far as I can tell the Intel developers assumed you would enable
this bit just prior to powering on the host and turn off after all the
POST codes are transferred and we can assume the host system is done
with the UEFI stage and the real OS took over.

OpenBMC seems to have all the necessary hooks to do it that way, and
you have a netlink command to send whatever you need for that from the
userspace, e.g. with the "C version" ncsi-netlink command to set this
bit just run:

ncsi-netlink -l 3 -c 0 -p 0 -o 0x50 0x00 0x00 0x01 0x57 0x20 0x00 0x01

https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-networkd/+/36592
would provide an OpenBMC-specific way too.

There's another related thing to consider here: by default I210 has
power-saving modes enabled and so when BMC is booting the link is
established only in 100BASE-T mode. With this configuration and this
bit always set you'd be always stuck to that, never getting Gigabit
speeds.

For server motherboards I propose to configure I210 with this:
./eeupdate64e /all /ww 0x13 0x0081 # disable Low Power Link Up
./eeupdate64e /all /ww 0x20 0x2004 # enable 1000 in non-D0a
(it's a one-time operation that's best to be performed along with the
initial I210 flashing)

Ivan, so far I have an impression that the user-space solution would
be much easier, flexible and manageable and that there's no need for
this command to be in Linux at all.

-- 
Be free, use free (http://www.gnu.org/philosophy/free-sw.html) software!
mailto:fercerpav@gmail.com