Re: [PATCH] net/hsr: Add support for the High-availability Seamless Redundancy protocol (HSRv0)

[Arvid Brodin <arvid.brodin@xdin.com>]

On 2013-06-26 00:46, Stephen Hemminger wrote:
> On Wed, 26 Jun 2013 00:30:51 +0200
> Arvid Brodin <arvid.brodin@xdin.com> wrote:
> 
>> On 2013-06-25 02:07, Stephen Hemminger wrote:
>>> On Mon, 24 Jun 2013 09:44:01 -0700
>>> Arvid Brodin <arvid.brodin@xdin.com> wrote:
>>>
>>>> +
>>>> +
>>>> +static bool is_admin_up(struct net_device *dev)
>>>> +{
>>>> +       return dev && (dev->flags & IFF_UP);
>>>> +}
>>>> +
>>>> +static bool is_operstate_up(struct net_device *dev)
>>>> +{
>>>> +       return (dev && (dev->operstate == IF_OPER_UP));
>>>> +}
>>>> +
>>>
>>> Isn't is_admin_up the same as netif_running()?
>>
>> I don't think so (but I'm unsure). This is netif_running() (from include/linux/netdevice.h):
>>
>> static inline bool netif_running(const struct net_device *dev)
>> {
>> 	return test_bit(__LINK_STATE_START, &dev->state);
>> }
>>
>> I'm not sure what this does. I don't think it's the same as checking whether the admin has
>> issued an UP command?
>>
>>
>>> And is_operstate_up the same as netif_oper_up()?
>>
>> netif_oper_up() also returns true if the device state is IF_OPER_UNKNOWN. A device can
>> have that state without even having carrier, so that's not OK for the HSR driver.
>>
> 
> Then use netif_oper_up(dev) && netif_carrier_on(dev)

That would be very weird since netif_carrier_on() signals that the carrier is on (and
returns void). :) But I get your point - i.e. check for carrier state.


This is netif_oper_up() in include/linux/netdevice.h:

static inline bool netif_oper_up(const struct net_device *dev)
{
	return (dev->operstate == IF_OPER_UP ||
		dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
}

I'm guessing that IF_OPER_UNKNOWN is just a default state and that some (old or buggy)
network drivers simply does not set this as they should? I.e. they should call
netif_carrier_off() on open, but they don't. The "backward compat" may be a way around
that? (We had problems with this on an old version of the macb driver; I think that's why
we wrote our own is_operstate_up().)

If so, I guess you are right and I should use netif_oper_up() (and any problems arising
from that would be due to bugs in the respective device driver). Does this sound reasonable?


-- 
Arvid Brodin | Consultant (Linux)
T: +46-8-56254286 | M: +46-70-9714286 | arvid.brodin@xdin.com
XDIN AB | Knarrarnäsgatan 7 | SE-164 40 Kista | Sweden | xdin.com