[ofa-general] Does ib0 always map to port1?

[Sumeet Lahorani]

Thanks all. When I look at the ifconfig ib0 output, it looks as though 
only the first 6 bytes of the HWaddr are printed correctly and the rest 
are 0's.

ib0       Link encap:InfiniBand  HWaddr 
80:00:00:48:FE:80:00:00:00:00:00:00:00:00:00:00:00:00:00:00 
          inet6 addr: fe80::21e:bff:ff4c:8a2d/64 Scope:Link
          UP BROADCAST RUNNING SLAVE MULTICAST  MTU:1500  Metric:1
          RX packets:7359 errors:0 dropped:0 overruns:0 frame:0
          TX packets:7054 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:256
          RX bytes:974584 (951.7 KiB)  TX bytes:562068 (548.8 KiB)

The IPv6 address seems to more closely match the GUID of the port

# ibstatus
Infiniband device 'mlx4_0' port 1 status:
        default gid:     fe80:0000:0000:0000:001e:0bff:ff4c:8a2d
        base lid:        0x1f
        sm lid:          0x4
        state:           4: ACTIVE
        phys state:      5: LinkUp
        rate:            20 Gb/sec (4X DDR)

But this is not an exact match either (0x21e vs 0x1e). Is there another 
command (apart from ifconfig) I can use to get a better match?

- Sumeet

David Dillow wrote:
> On Thu, 2009-01-22 at 10:18 +0200, Dotan Barak wrote:
>   
>> On Thu, Jan 22, 2009 at 9:45 AM, Or Gerlitz <ogerlitz at voltaire.com> wrote:
>>     
>>> Sumeet Lahorani wrote:
>>>       
>>>> I see that ib0 always maps to port1 and ib1 always maps to port2 on the
>>>> HCA. I'm trying to find out if this will always be the case and if so which
>>>> script ensures this mapping?
>>>>         
>>> Yes, on a dual ported HCA, ib0 maps to port1 and ib1 to port2, this is a
>>> property of the ipoib driver regardless of which HW is used, see
>>> drivers/infiniband/ulps/ipoib/ipoib_main :: ipoib_add_one()
>>>
>>>       
>> Part of the "MAC" address of the I/F is the port GUID, so you can use
>> this value to determine which HCA.port is mapped to that I/Fs...
>> (if there are several HCAs in the same host, this can be very useful ...)
>>     
>
> This is the better way to go for general use, as it is entirely possible
> for a user to change the device names such that ib0 swaps with ib1:
>
> ip link set ib0 name ib0_rename
> ip link set ib1 name ib0
> ip link set ib0_rename ib1
>
> Now, it is probably not very likely that a user will do this, but
> Dotan's right that it can get fun when there are multiple HCA's, so it
> is a good idea in any event.
>