RTT increasing issue in vxworks END drivers - VxWorks

This is a discussion on RTT increasing issue in vxworks END drivers - VxWorks ; Hi, This is Bobby. I have problem while doing Ping test from a lnPci0 to lnPci0 interfaces. When this ping process is running, if i remove my gei0 cable & Plug it back to the same interface, this is effecting ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: RTT increasing issue in vxworks END drivers

  1. RTT increasing issue in vxworks END drivers

    Hi,

    This is Bobby. I have problem while doing Ping test from a lnPci0 to
    lnPci0 interfaces. When this ping process is running, if i remove my
    gei0 cable & Plug it back to the same interface, this is effecting on
    the increased RTT for the immediate reply in the above said Ping
    process & after this the rest of the replies are normal. I don't know
    why this is happening. Pls go through below for details....

    The test setup details are, i have 2No's of industrial PC's which
    consists of gei0 (82541PI Gigabit Ethernet controller) and lnPci0 (AMD
    PC net AM79C73BVC Fast Ethernet) interfaces. In both of which, gei0
    was configured as Master and slave is lnPci0. (standby for gei0, just
    in case if gei0 fails in real time application).

    In this situation, I observed that while pinging between 2 lnPci0
    interfaces from one PC to another, & if i remove the cable connected
    to gei0 (which is in idle state, & not related to lnPci0 pinging
    process) and plug it back, the immediate reply message coming from 2nd
    PC, in the above said pinging process is getting delayed upto 1740
    Milli seconds randomly (not fixed value) and after that the subsequent
    replies are normal or without delay...

    So here the problem is that the real time application can bear RTT
    delay only upto 300 milli seconds not more than that....

    My Application is running on Vxworks5.5 with gei0 & lnPci0 END driver
    support. And the BSP is Pcpentium3 generic BSP (default BSP in
    tornado2.2) for Vxworks5.5.
    Vxworks loads from gei0 interface, & also i tried to loading Vxworks
    from lnPci0. But issue is same...

    Any idea on this issue???? Pls help me out!

    Regards
    Bobby

  2. Re: RTT increasing issue in vxworks END drivers

    On 10월8일, 오후1시50분, honey2...@gmail.com wrote:
    > Hi,
    >
    > This is Bobby. I have problem while doing Ping test from a lnPci0 to
    > lnPci0 interfaces. When this ping process is running, if i remove my
    > gei0 cable & Plug it back to the same interface, this is effecting on
    > the increased RTT for the immediate reply in the above said Ping
    > process & after this the rest of the replies are normal. I don't know
    > why this is happening. Pls go through below for details....
    >
    > The test setup details are, i have 2No's of industrial PC's which
    > consists of gei0 (82541PI Gigabit Ethernet controller) and lnPci0 (AMD
    > PC net AM79C73BVC Fast Ethernet) interfaces. In both of which, gei0
    > was configured as Master and slave is lnPci0. (standby for gei0, just
    > in case if gei0 fails in real time application).
    >
    > In this situation, I observed that while pinging between 2 lnPci0
    > interfaces from one PC to another, & if i remove the cable connected
    > to gei0 (which is in idle state, & not related to lnPci0 pinging
    > process) and plug it back, the immediate reply message coming from 2nd
    > PC, in the above said pinging process is getting delayed upto 1740
    > Milli seconds randomly (not fixed value) and after that the subsequent
    > replies are normal or without delay...
    >
    > So here the problem is that the real time application can bear RTT
    > delay only upto 300 milli seconds not more than that....
    >
    > My Application is running on Vxworks5.5 with gei0 & lnPci0 END driver
    > support. And the BSP is Pcpentium3 generic BSP (default BSP in
    > tornado2.2) for Vxworks5.5.
    > Vxworks loads from gei0 interface, & also i tried to loading Vxworks
    > from lnPci0. But issue is same...
    >
    > Any idea on this issue???? Pls help me out!
    >
    > Regards
    > Bobby


    I had a similar problem with ethernet.

    The problem was that network is stopped for a few seconds when pull
    out ethernet cable from geisc0 which is primary ethernet port for
    MVME6100 board under vxworks 5.5.

    tNetTask is used to handle link status change.

    tNetTask may check link status every 5(?).. secs.

    In my opinion, the delay of ping's respones time is affected by msec
    timeout for checking link status.

    I have changed ethernet device driver by windriver's aid but ping's
    response time is slower then noraml. I'm going to try it for vxworks
    6.x.

    Regards
    Jin



+ Reply to Thread