RFE submitted under drivers / other.
Garrett D'Amore wrote:
> Your example looks reasonable. I'm not sure about the problem of
> multiple MIIs. Sort of like the problem of XGMII, I just don't have
> knowledge in the area.
>
> -- Garrett
>
> Steven Stallion wrote:
>> Garrett D'Amore wrote:
>>
>>> You need a way to register your mac callbacks with Nemo. :-) Probably
>>> you want to have a way to associate the MII with a state structure at
>>> registration time as well... unless you believe that the same soft state
>>> used by the rest of nemo will be sufficient (it might be ... I don't
>>> have experience with NICs that support more than a single MII part on
>>> them to know whether an intermediate state structure would be useful or
>>> not. You'd still need a way to identify *which* MII part is active in
>>> the case where multiple MII devices are present on a single NIC part,
>>> though.)
>>>
>>
>> I figured a concrete example may help:
>>
>> 2838 boolean_t
>>
>> 2839 afe_m_getcapab(void *arg, mac_capab_t cap, void *cap_data)
>>
>> 2840 {
>>
>> 2841 switch (cap) {
>>
>> 2842 case MAC_CAPAB_MII:
>>
>> 2843 mac_capab_mii_t *miip = cap_data;
>>
>> 2844
>>
>> 2845 miip->mm_read = afe_mii_read;
>>
>> 2846 miip->mm_write = afe_mii_write;
>>
>> 2847 break;
>>
>> 2848
>>
>> 2849 default:
>>
>> 2850 return (B_FALSE);
>>
>> 2851 }
>>
>> 2852 return (B_TRUE);
>>
>> 2853 }
>>
>>
>> WRT having multiple MII parts I am not sure how this should be handled.
>> Is it possible that devices which use multiple MII's also make use of
>> m_instance?
>>
>> Thoughts?
>>
>>
>
--
--
Yet magic and hierarchy
arise from the same source,
and this source has a null pointer.
Reference the NULL within NULL,
it is the gateway to all wizardry.
_______________________________________________
networking-discuss mailing list
[email protected]
