From: Arvid Brodin <arvid.brodin@xdin.com>
To: <netdev@vger.kernel.org>
Cc: David Miller <davem@davemloft.net>, <shemminger@vyatta.com>,
<joe@perches.com>, <jboticario@gmail.com>,
<balferreira@googlemail.com>
Subject: Re: [PATCH v2] net/hsr: Add support for the High-availability Seamless Redundancy protocol (HSRv0)
Date: Tue, 20 Aug 2013 19:36:22 +0200 [thread overview]
Message-ID: <5213A916.2020505@xdin.com> (raw)
In-Reply-To: <51D2156A.3070403@xdin.com>
On 2013-07-02 01:48, Arvid Brodin wrote:
> On 2013-06-29 06:16, David Miller wrote:
>> From: Arvid Brodin <arvid.brodin@xdin.com>
>> Date: Wed, 26 Jun 2013 17:16:11 +0200
>>
>>> +static int hsr_dev_open(struct net_device *dev)
>>> +{
>>> + struct hsr_priv *hsr_priv;
>>> +
>>> + hsr_priv = netdev_priv(dev);
>>> +
>>> + if (hsr_priv->slave[0])
>>> + dev_open(hsr_priv->slave[0]);
>>> + if (hsr_priv->slave[1])
>>> + dev_open(hsr_priv->slave[1]);
>>> +
>>> + return 0;
>>> +}
>>
>> dev_open() can and does fail, you must thus check for error returns, undo any
>> necessary state, and propagate that error to callers of hsr_dev_open.
>>
>
> I'm not sure it's an error (from the HSR interface's point of view) if the slave(s)
> won't come up here. The calls to dev_open() can be seen more like a convenience
> than a necessity (I actually left them out to begin with). If none of the slaves
> could go up the HSR interface would end up as admin UP but with operstate
> IF_OPER_LOWERLAYERDOWN until any of the slaves went up.
[snip]
Stephen Hemminger explained to me (in private email) that many of the errors
from dev_open() aren't soft errors - i.e. they should be propagated back up
even if the HSR driver doesn't care (out of memory, and hardware dead being
examples).
But HSR is often used in safety critical systems, and it may be important that
a hsr_dev->dev_open() does not fail just because one of the slave's cable
happens to be unplugged (or cut) - redundancy being the whole point of HSR.
I think the only reasonable thing to do is not to call dev_open() at all here,
and instead just inform the user if a slave is not already up. This way user
space must explicitly set both slaves UP and thus must also decide how to treat
different kinds of "slave errors".
I will send a new patch with this change.
Thanks to Stephen for helping me with this!
--
Arvid Brodin | Consultant (Linux)
XDIN AB | Knarrarnäsgatan 7 | SE-164 40 Kista | Sweden | xdin.com
prev parent reply other threads:[~2013-08-20 17:45 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-06-26 15:16 [PATCH v2] net/hsr: Add support for the High-availability Seamless Redundancy protocol (HSRv0) Arvid Brodin
2013-06-29 4:16 ` David Miller
2013-07-01 23:48 ` Arvid Brodin
2013-07-08 15:54 ` Arvid Brodin
2013-08-20 17:36 ` Arvid Brodin [this message]
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=5213A916.2020505@xdin.com \
--to=arvid.brodin@xdin.com \
--cc=balferreira@googlemail.com \
--cc=davem@davemloft.net \
--cc=jboticario@gmail.com \
--cc=joe@perches.com \
--cc=netdev@vger.kernel.org \
--cc=shemminger@vyatta.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.