Actions
Forum Thread Details
  • State Not Answered
  • Replies 11 replies
  • Subscribers 353 subscribers
  • Views 3177 views
  • Users 0 members are here
Related

Ethernet is not running in Zedboard

Former Member

I was trying to migrate the WinCE BSP for Xilinx ZC702 for ZedBoard and got stuck in getting the Ethernet working.
After I change the PHY address to 0 it successfully completes the autonegotiation and reports the link speed. There after it fails and looks like it is not receiving any frames.
I blamed the PHY chip differences and spend couple of days in the data sheets and no luck. At that point I thought of trying with the supplied Linux boot image(till then I used the xilinx XPS and run the sample based on the hw definition file from Zedboard and didn't check the LAN).
Surprisingly it didn't even do the self pinging, I changed the IP address but it stuck with the ping command as follows
PING 192.168.1.10 (192.168.1.10): 56 data bytes

Out of the 3 boards we have I checked two and got the same results. Any help is much appreciated.

Parents
  • 0

    Seetharam,

    Can you please elaborate on what you mean by the self-pinging?  Are you talking about pinging the address of lo or eth0 adapters? 

    I tried pinging both adapters on my ZedBoard and there was no response which seems to match what you are seeing. 

    i did some experimenting since I noticed that lo was not even listed.  First I brought lo up using the following command:

    zynq> ifconfig lo up

    Then I was able to ping 127.0.0.1 successfully.

    For grins, I brought eth0 down and then back up using the following commands:

    zynq> ifconfig eth0 down
    zynq> ifconfig eth0 up

    Then I was able to ping 192.168.1.10 successfully.

    Not sure if I have a good explanation for this behavior though.  Maybe this has something to do with the order that things are brought up in /dev/init.d/rcS?

    Please let me know if you see the same thing on your end.

    Regards,

    -Kevin

Reply
  • 0

    Seetharam,

    Can you please elaborate on what you mean by the self-pinging?  Are you talking about pinging the address of lo or eth0 adapters? 

    I tried pinging both adapters on my ZedBoard and there was no response which seems to match what you are seeing. 

    i did some experimenting since I noticed that lo was not even listed.  First I brought lo up using the following command:

    zynq> ifconfig lo up

    Then I was able to ping 127.0.0.1 successfully.

    For grins, I brought eth0 down and then back up using the following commands:

    zynq> ifconfig eth0 down
    zynq> ifconfig eth0 up

    Then I was able to ping 192.168.1.10 successfully.

    Not sure if I have a good explanation for this behavior though.  Maybe this has something to do with the order that things are brought up in /dev/init.d/rcS?

    Please let me know if you see the same thing on your end.

    Regards,

    -Kevin

Children
  • 0 in reply to zedhed

    Kevin,
    Yep. I see the same over here.
    Once I enable the loop back(ifconfig lo up) I can ping the 127.0.0.1 or the IP address of Zedboard.The other command is not required.Explanation could be ZC702 it is enabled and Zedboard it is not enabled by default.
    Looks like I am missing some setting still in driver level at BSP code.Do you have any suggestions other than the PHY address any other specific settings required for Marvell 88E1518 chip compare to 88E111R?

  • 0 in reply to Former Member

    Hi Seetharam,

    I know this thread is a little old but I figure it's worth a shot.

    I'm currently trying to accomplish something similar to you in that I'd like to port a QNX BSP from the ZC702 to the Zedboard.

    Like you, I changed the PHY address to 0 in my source code and still had no luck.

    After some more digging, I noticed that the ZC702 has a reset line connected to the PHY from MIO11 (this signal is AND'ed with the POR signal). On the Zedboard, there is only the POR signal connected to the PHY reset line meaning the SoC can't reset the PHY. So if the driver attempts to reset the PHY in software during configuration (which the QNX driver does when it's started), it will fail.

    I also noticed that the Ethernet frequencies (ENET0_FREQ and ENET1_FREQ in ps7_init.h) are different between the ZC702 and the Zedboard. I don't know if these frequencies are PHY specific as I haven't been able to find the datasheets for the 88E1518 or 88E111R (I believe you need an NDA signed with Marvell to access them and I haven't gotten that desperate yet).

    Have you happened to make any new discoveries since you posted last September?


    Best regards,
    Chris

  • 0 in reply to Former Member

    I'm posting for anyone else who happens across this thread. Hopefully, this will be useful for somebody.

    There are a few things that need to be addressed to port Ethernet connectivity from a ZC702 board support package to the Zedboard. Here are the changes I had to make to get Ethernet up and running:

    1) Hardcode the PHY address to 0
    2) The ZC702 seems to store the PHY MAC address in an I2C EEPROM. I had to eliminate any EEPROM code to make sure nothing hangs/fails in trying to access it and then hardcode the MAC address to 00:0a:35:00:01:22
    3) I eliminated any code that attempted to reset the PHY through MIO 11 as this connection is not wired on the Zedboard
    3) I didn't actually need this, but you may need to change the PHY ID to 0x1410dd1

    The good news is that the lacking reset line wiring won't prevent you from getting everything talking nicely.


    Best,
    Chris

  • 0 in reply to Former Member

    Hi Chris,

    I am also working to port QNX on zedboard, starting from zc702 BSP. Thank you for the valuable information posted above.

    However, I am stuck at your first step "Hardcode the  to 0 in source code".

    There are 3 source C files in devnp folder which possibly contains PHY address (phy_addr in mii.c and nicinfo.h, phys_addr in bdring.c, phy_address in control.c as varieties). Note nicinfo.h is in /opt/qnx650/target/qnx6/usr/include/hw/nicinfo.h

    QNX boot sequence crashes after my tries. Please advise how to make the correct modification to null the PHY address.

    Thanks in advance,


    Mike