• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 34 subscribers
  • Views 257 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DP83867IR: No ping response after LAN cable was disconnected and connected.

Naoto Uegaki
Intellectual 805 points

Part Number: DP83867IR
Other Parts Discussed in Thread: AM3352, , AM3358

We are using Etherent PHY (DP83867IR) on AM3352 and TISDK8.2.
It was reported that no ping response after LAN cable was disconnected and connected in spite of LAN LED was on.
It seemed that this problem happend with special network switch or network adaptor respectively.
It were Planet IGS-1020PTF and ICY BOX IB-AC535-C.

I'd like to know this problem is fixed if I update PHY driver (drivers/net/phy/dp83867.c)?
I'm using the attached dp83867_tisdk8.2_mine.c . It seems old.
I checked the latest file in SDK8.2, it was the attached dp83867_tisdk8.2_latest.c .
Can this problem fiexed if I update to dp83867_tisdk8.2_latest.c.

If not, I checked the latest file in SDK11.2, it was the attached dp83867_tisdk11.2_latest.c .
There are many differences. Can this problem fiexed if I just update to dp83867_tisdk11.2_latest.c. Kernel or SDK update is needed?

  • 0
    TI__Mastermind 26771 points

    Hi Uegaki-san, 

    How frequently does this happen? Does it happen all the time? Are the switches managed or unmanaged? Does this happen for multiple boards?
    How can this issue be replicated?
    If you have access to the registers on the PHY, could you read the register 1h when this happens?
    I have not heard of a case in which the driver had an issue with this type of issue?

    If you plan to upgrade the driver from 8.2 to 11.2, kernel and SDK update will be required. 

    Best,
    J

  • 0 in reply to J
    Intellectual 805 points

    Hi J -san,

    It happened all the time and there were two boards.
    These bords are set in our customer's datacenter.
    I have no information whether the switches are managed or not at the moment.
    Also, I haven't replicated this issue in my office yet.

    It's difficult to access PHY registers because it will be a hassle and our customer may not be able to understand how to access PHY register if I teach it.
    I'm trying to get more information from customer, and trying to buy Planet IGS-1020PTF and ICY BOX IB-AC535-C.

  • 0 in reply to Naoto Uegaki
    TI__Mastermind 26771 points

    Hi Uegaki-san,

    Understood. In this case, could the customer send us any log that comes up on the AM3358 side when the board is linked to these switches? 


    I also wonder if the particular switch is not able to assign ip address to the MAC that the ping is failing.

    How is the ip address assigned to the MAC? Is it fixed or dynamic?

    Best,

    J

  • 0 in reply to J
    Intellectual 805 points

    Hi J -san,

    IP address was set statically(192.168.0.102/24), not dynamically. Please refer the folowing figure.
    Before LAN cable disconnecting and connecting, PING response was OK.
    After LAN cable disconnecting and connecting, PING response was nothing. (I heard that it became OK after rebooting the borad.)
    At that time, green LED and orange LED were both ON. And "ifconfig" result showed static address "192.168.0.102/24".

    BTW, it'll difficult to get log because PHY log needs root permission and the linux on the borad profibited root access as default. But I'll try to get something log when this problem happen.

  • 0 in reply to Naoto Uegaki
    TI__Mastermind 26771 points

    Hi Uegaki-san, 

    Thank you for the information. 

    Naoto Uegaki said:
    At that time, green LED and orange LED were both ON. And "ifconfig" result showed static address "192.168.0.102/24".

    In this the case when the ping is working?

    Also, is there a chance the ip address gets wiped off when the new link is established so the ping does not end up working? I've seen that some TI processor does this (beagleplay). 

    When this issue happens, would the customer be able to get ethtool information? 
    Also, is SW restarting the PHY (0x1F = 0x4000) OR refreshing the auto-negotiation (bit 9 of 00h goes high) remove this problem?

    Lastly, in the problematic state and when they ping, do they see TX/RX counters incrementing on ifconfig? I am wondering if we can isolate which path of the connection is breaking. 

    Any information would be appreciated so I can help you narrow down where the issue could be. 

    Best,
    J

  • 0 in reply to J
    Intellectual 805 points

    Hi J -san,

    I found the cause of this problem that Gateway address was wrong as shown below.
    After setting correct gateway address 192.168.0.1, this problem was solved.

    IP: 192.168.0.102
    NETMASK: 255.255.255.0
    GATEWAY: 192.168.1.1

    In the begining, I'm sorry that I can't aware of wrong gateway address because this problem happened in datacenter and there are many network engneers.
    Thank you for your support.