[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAKxU2N87Nqa73M+wda3Phayu5dmkWEMhDXgxz=4bASV_-8D4yQ@mail.gmail.com>
Date: Tue, 15 Oct 2024 12:44:52 -0700
From: Rosen Penev <rosenp@...il.com>
To: Simon Horman <horms@...nel.org>
Cc: netdev@...r.kernel.org, "David S. Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
Andrew Lunn <andrew@...n.ch>, Shannon Nelson <shannon.nelson@....com>,
Uwe Kleine-König <u.kleine-koenig@...libre.com>,
Breno Leitao <leitao@...ian.org>, Jeff Johnson <quic_jjohnson@...cinc.com>,
Christian Marangi <ansuelsmth@...il.com>, open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCHv6 net-next 6/7] net: ibm: emac: generate random MAC if not found
On Sat, Oct 12, 2024 at 6:16 AM Simon Horman <horms@...nel.org> wrote:
>
> On Fri, Oct 11, 2024 at 12:56:21PM -0700, Rosen Penev wrote:
> > On this Cisco MX60W, u-boot sets the local-mac-address property.
> > Unfortunately by default, the MAC is wrong and is actually located on a
> > UBI partition. Which means nvmem needs to be used to grab it.
> >
> > In the case where that fails, EMAC fails to initialize instead of
> > generating a random MAC as many other drivers do.
> >
> > Match behavior with other drivers to have a working ethernet interface.
> >
> > Signed-off-by: Rosen Penev <rosenp@...il.com>
> > ---
> > drivers/net/ethernet/ibm/emac/core.c | 9 ++++++---
> > 1 file changed, 6 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/ibm/emac/core.c b/drivers/net/ethernet/ibm/emac/core.c
> > index b9ccaae61c48..faa483790b29 100644
> > --- a/drivers/net/ethernet/ibm/emac/core.c
> > +++ b/drivers/net/ethernet/ibm/emac/core.c
> > @@ -2937,9 +2937,12 @@ static int emac_init_config(struct emac_instance *dev)
> >
> > /* Read MAC-address */
> > err = of_get_ethdev_address(np, dev->ndev);
> > - if (err)
> > - return dev_err_probe(&dev->ofdev->dev, err,
> > - "Can't get valid [local-]mac-address from OF !\n");
> > + if (err == -EPROBE_DEFER)
> > + return err;
> > + if (err) {
> > + dev_warn(&dev->ofdev->dev, "Can't get valid mac-address. Generating random.");
> > + eth_hw_addr_random(dev->ndev);
> > + }
>
> The above seems to take the random path for all errors other than
> -EPROBE_DEFER. That seems too broad to me, and perhaps it would
> be better to be more specific. Assuming the case that needs
> to be covered is -EINVAL (a guess on my part), perhaps something like this
> would work? (Completely untested!)
>
> err = of_get_ethdev_address(np, dev->ndev);
> if (err == -EINVAL) {
> /* An explanation should go here, mentioning Cisco MX60W
> * Maybe the logic should even be specific to that hw?
> */
> dev_warn(&dev->ofdev->dev, "Can't get valid mac-address. Generating random.");
> eth_hw_addr_random(dev->ndev);
> } else if (err) {
> return dev_err_probe(&dev->ofdev->dev, err,
> "Can't get valid [local-]mac-address from OF !\n");
> }
That's just yak shaving. besides 0 and EPROBE_DEFER,
of_get_ethdev_address returns ENODEV, EINVAL, and maybe something
else. I don't see a good enough reason to diverge from convention
here. This same pattern is present in other drivers.
>
> Also, should this be a bug fix with a Fixes tag for net?
No. It's more of a feature honestly.
>
> >
> > /* IAHT and GAHT filter parameterization */
> > if (emac_has_feature(dev, EMAC_FTR_EMAC4SYNC)) {
> > --
> > 2.47.0
> >
Powered by blists - more mailing lists