DP83630-EVK: Software for the DP83630-EVK/NOPB development kit

Hi Bill,

When you say snooping the traffic on switch, you meant you mirrored all the traffic from DP83630 mirrored to port having L3/L3 and wireshark connected ? Ideally the packets should show up.

Alternatively, given you can snoop traffic on Perle system , you can perform PTP sync between Perle and and DP83630 and check the packet exchange.

For PTP verification, you can use the status register provided in Phy as well.

Regards,
Geet

Hello Geet,

Sorry for the late reply, was off the past two days. So I went ahead and setup port mirroring on my setup and I'm still not seeing ANY PTP packets. I have both ports sending BOTH TX and RX packets to the mirrored port where I am running Wireshark but haven't seen a single PTP packet.

I've seen some multicast packets using the destination address of 224.0.0.251, but these don't appear to be related to PTP messages.

One question I have, how is this supposed to work when it appears that the default configuration for the MAC_SRC_ADD (set in Page 5 - 0x18 - PTP Packet-based Status Configuration Register 0) is to Use Mac Address [08 00 17 0B 6B 0F] (by having bits 11 and 12 both 0).

Since, by default, this is the SAME on all boards connected together, doesn't this mess up the ARP table in the switch if that SAME source address is seen on multiple ports?

Any help would be greatly appreciated!

Much thanks,
Bill

An update on this.

So it turns out that the DP83630-EVK development boards being used are definitely NOT generating and sending out PTP packets.  

The reason that we're seeing the clocks in sync on the scope (scoping the clock output on each board) is that BIT 13 in Register Page 0 - 0x1C - PHY Control Register 2 (PHYCR2) is set to a 1 on both boards.

According to the documentation, this bit "enables fully synchronous communication relative to the recovered receive clock".

So it is using the CLOCK to synchronize, NOT PTP packets.  When that bit is turned off on either of the Development Boards, the boards are no longer in sync.

So my question is, do you guys have some Linux test code I could use to get this going?  I just downloaded the LinuxPtp source code and sure enough, then only supported Hardware timestamping at the PHY level is on your device! :-)   (From the README.org file: )

** Hardware Timestamping - PHY

|---------+-------------------------------+---------|
| Driver | Hardware | Version |
|---------+-------------------------------+---------|
| dp83640 | National Semiconductor PHYTER | 3.0 |
|---------+-------------------------------+---------|

If you have ANY test code you could point me to, it would save me a lot of time.

Much thanks,

Bill

So in the version of Linux I'm using, there's no support for the National Semi dp83640 device.  Running 'make menuconfig' the only National PHY drivers listed are:

x x [*] National Semi-conductor devices (NEW) x x
x x <M> National Semiconductor DP8381x series PCI Ethernet support x x
x x <M> National Semiconductor DP83820 support x x
x x [*] National Semi-conductor 8390 devices (NEW) x x
x x <M> Asix AX88190 PCMCIA support x x
x x <M> NE2000/NE1000 support x x
x x <M> PCI NE2000 and clones support (see help) x x
x x <M> NE2000 compatible PCMCIA support x x
x x <M> SMC Ultra support x x
x x <M> WD80*3 support

Also, under TI, the only supported devices are:

x [*] Texas Instruments (TI) devices (NEW) x x
x x < > TI CPSW ALE Support (NEW) x x
x x <M> TI ThunderLAN support

Any help on which Linux branch has support for the National dp83640 device?

Thanks,

Bill

Hi,

We don't have latest linux based implementation available. However we had created a TI Design using DP838630 (www.ti.com/.../TIDA-00496)

Please go thru the PTP section in below user guide to follow step by step procedure to enable PTP and how to verify.

www.ti.com/.../tiduat6a.pdf


Regards,
Geet

So I no longer am using the Pearle switch. We don't want all of our customers to have to purchase a separate Pearle switch. The idea is to make one of the Access Points that have these IEEE 1588 capable PHYs become the grandmaster using NTP and send PTP packets to the other Access Points. The master will be running ntpd and get the correct time from an NTP server, then use ptpd to send out PTP messages to all other Access Points in the system.

I'm hoping that you guys have some version of this already worked out and have some software I could start with.. hopefully I won't have to write this myself. We do have a proprietary solution but we want to move away it in favor of an IEEE standard.

Any help would be greatly appreciated.

Much thanks,
Bill

Hi Geet,

Apologies, it's 1am and I'm reading through your last reply on my cell phone nd i actually missed your last response. I saw the link you sent. The first link didnt work for me but the 2nd did. I'll definitely go through that document tomorrow when I get in and see how easy it is to get PTP packets generated. Thanks a lot for sending it. I'll check out that other PHY as well.

Thanks again!
Bill

Hi Geet, is thefe any chance toy could resend a link to the design (i.e. get the 1sr link you sent to work). We'd love to take a look at it.

Thanks again, Bill

Hi Geet, was able to get the the first link to work, no need to resend it.

I'm looking at all of the documentation now.  Thanks again for sending it out!

Regards, Bill