[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <aa348a55d8a5574e6ceaede7d40f47e5fbcb9b6c.camel@infradead.org>
Date: Thu, 27 Mar 2025 12:01:40 +0000
From: David Woodhouse <dwmw2@...radead.org>
To: Leon Romanovsky <leon@...nel.org>, "Arinzon, David" <darinzon@...zon.com>
Cc: Andrew Lunn <andrew@...n.ch>, David Miller <davem@...emloft.net>, Jakub
Kicinski <kuba@...nel.org>, "netdev@...r.kernel.org"
<netdev@...r.kernel.org>, Eric Dumazet <edumazet@...gle.com>, Paolo Abeni
<pabeni@...hat.com>, Simon Horman <horms@...nel.org>, Richard Cochran
<richardcochran@...il.com>, "Machulsky, Zorik" <zorik@...zon.com>,
"Matushevsky, Alexander" <matua@...zon.com>, "Bshara, Saeed"
<saeedb@...zon.com>, "Wilson, Matt" <msw@...zon.com>, "Liguori, Anthony"
<aliguori@...zon.com>, "Bshara, Nafea" <nafea@...zon.com>, "Schmeilin,
Evgeny" <evgenys@...zon.com>, "Belgazal, Netanel" <netanel@...zon.com>,
"Saidi, Ali" <alisaidi@...zon.com>, "Herrenschmidt, Benjamin"
<benh@...zon.com>, "Kiyanovski, Arthur" <akiyano@...zon.com>, "Dagan,
Noam" <ndagan@...zon.com>, "Bernstein, Amit" <amitbern@...zon.com>, "Allen,
Neil" <shayagr@...zon.com>, "Ostrovsky, Evgeny" <evostrov@...zon.com>,
"Tabachnik, Ofir" <ofirt@...zon.com>, "Machnikowski, Maciek"
<maciek@...hnikowski.net>, Rahul Rameshbabu <rrameshbabu@...dia.com>, Gal
Pressman <gal@...dia.com>, Vadim Fedorenko <vadim.fedorenko@...ux.dev>
Subject: Re: [PATCH v8 net-next 5/5] net: ena: Add PHC documentation
On Thu, 2025-03-27 at 13:15 +0200, Leon Romanovsky wrote:
> On Wed, Mar 26, 2025 at 03:32:00PM +0000, Arinzon, David wrote:
> > > > > If you read the actual code in that patch, there's a hint though.
> > > > > Because the actual process in ena_phc_enable_set() does the following:
> > > > >
> > > > > + ena_destroy_device(adapter, false);
> > > > > + rc = ena_restore_device(adapter);
> > > > >
> > > > > Is that actually tearing down the whole netdev and recreating it
> > > > > when the PHC enable is flipped?
> > > >
> > > > Well Jakub said it is a pure clock, not related to the netdev.
> >
> > The PHC device is a PTP clock integrated with the networking device under the same PCI device.
> > As previously mentioned in this thread, enabling or disabling the
> > ENA PHC requires reconfiguring the ENA network device,
> > which involves tearing down and recreating the netdev.
> > This necessitates having a separate control knob.
>
> I appreciate the desire to keep everything in one place and provide
> custom, vendor specific sysfs files. However, there is standard way
> (auxbus??) to solve it without need to reinvent anything.
Using auxbus would allow the PHC driver to be loaded as a separate
module. But AIUI enabling the PHC involves reconfiguring the device and
thus bringing the network device down and back up. Which isn't ideal.
It *isn't* as separate as auxbus would make it look.
We don't want to unconditionally enable the PHC support in the device
even when it's not being used, because of the potential blast radius of
such a change if *every* guest starts enabling it even when they don't
need to use it.
Download attachment "smime.p7s" of type "application/pkcs7-signature" (5069 bytes)
Powered by blists - more mailing lists