Vasu Dev wrote:
> On Mon, 2009-03-23 at 18:04 -0700, Joe Eykholt wrote:
>> Vasu Dev wrote:
>>>
>>> + if (netdev->netdev_ops && netdev->netdev_ops->ndo_get_storage_address) {
>>> + memcpy(fc->ctlr.ctl_src_addr,
>>> + netdev->netdev_ops->ndo_get_storage_address(netdev),
>>> + ETH_ALEN);
>>> + if (is_valid_ether_addr(fc->ctlr.ctl_src_addr))
>>> + fc->ctlr.spma = 1;
>>> + }
>> If SPMA was a requirement, we could just set the SPMA flag even
>> without a special MAC address.
>
> The spma flag is set only if NIC provides this another valid MAC
> address, so not sure where else we could have set this spma flag and
> what would be use of only setting up spma flag.
There's no reason we couldn't use the native MAC address for SPMA.
BTW, I still don't think we need SPMA support at all, but I don't mind
that it's getting added if nobody else does.
>> The above code seems to have little
>> to do with SPMA and just allows the LLD to provide another MAC
>> address to be used for FIP and FCoE.
>>
>
> The spma mode allows initiator to propose a MAC address to switch and
> above code is just using above mentioned another MAC address for that,
> therefore above code is related to spma.
True, it's related in a way. But it also sets a control address that'll be used
even if FPMA is used, or even if FIP isn't used, for the FIP
frame source address (FCoE controller address) and the non-FIP FLOGI
source address.
>> That part of the patch could be separated from the rest.
>
> This is the only place we are making use of patch 1/3 to read spma MAC
> address from NIC and we need above code in this patch adding spma mode
> with NIC proposed MAC address in spma mode, therefor above code cannot
> be separated from adding spma mode support in this patch.
OK.
Joe
_______________________________________________
devel mailing list
[email protected]
http://www.open-fcoe.org/mailman/listinfo/devel
