MFC of em/igb drivers

[Jeremy Chadwick]

On Fri, Jun 06, 2008 at 03:49:40PM +0800, Ganbold wrote:
> Jack Vogel wrote:
>> I got the new drivers in Friday afternoon for those that don't see CVS
>> messages.
>>
>> The igb driver is for 82575 and 82576 adapters, it has multiqueue support and
>> MSIX, there will be more server type enhancements in that driver as I get the
>> time.
>>
>> The em driver now will be client oriented, the latest hardware support however
>> is for 82574, Hartwell, which is a really nice low cost PCIE dual-port adapter,
>> that actually also has MSIX. This first released support for it will
>> use 3 interrupt
>> vectors, one for TX, one for RX, and Link. The hardware actually supports 5
>> vectors, so I am planning to add support for another RX and TX queue as my
>> schedule allows.
>>
>> Hartwell is also the first adapter that has IEEE 1588 PTP support, the driver
>> provides init and an ioctl interface to use that facility, I hope we
>> see software
>> support follow on soon. This is an early announcement, I am not sure on
>> exact dates for availability but they should be soon.
>>
>> As ever, issues and bugs should be sent to me. Cheers everyone!
>
> I have Intel Gigabit card and FreeBSD-7.0-STABLE recognizes it as following:
> ...
> em0 at pci0:0:25:0:        class=0x020000 card=0x2800103c chip=0x104a8086 
> rev=0x02 hdr=0x00
>    vendor     = 'Intel Corporation'
>    device     = '82566DM Gigabit Network Connection'
>    class      = network
>    subclass   = ethernet
> ...
> # uname -an
> FreeBSD test.ub.mng.net 7.0-STABLE FreeBSD 7.0-STABLE #0: Thu Jun  5 
> 11:12:34 ULAT 2008     tsgan at test.ub.mng.net:/usr/obj/usr/src/sys/FIREWALL  
> i386
>
>
> It has some problem, 1000baseTX connection status almost never goes to 
> active in bridged mode, sometimes
> it goes to active but then immediately it goes to no carrier. (The other 
> end is Cisco4507R, Gigabit Ethernet port)
> ...
> em0: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> metric 0 
> mtu 1500
>    options=198<VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4>
>    ether 00:0f:fe:82:23:db
>    media: Ethernet 1000baseTX <full-duplex> (autoselect)
>    status: no carrier
> ...
>
> Any idea how to solve this issue?

Have you tried disabling speed and duplex negotiation and explicitly
stating speed and duplex like so?

ifconfig_em0="... media 1000baseTX mediaopt full-duplex"

Cisco switches have a notorious history of not being "friendly" with
non-Cisco hardware.  Forcing duplex on both ends of the link (that means
on both the host side, and the Cisco side!) usually fixes it.

-- 
| Jeremy Chadwick                                jdc at parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                  Mountain View, CA, USA |
| Making life hard for others since 1977.              PGP: 4BD6C0CB |