Enabling Interface 2 (Port 2) of DM648 in NDK Ver.1.94.1 helloworld.pjt

Hi SP,

A couple of comments:

SP said:

cpsw_MDIO_FindingState: Timed Out looking for a Phy!
cpsw_MDIO_FindingState: Timed Out looking for a Phy!
cpsw_MDIO_FindingState: Timed Out looking for a Phy!

These error messages are coming from the DM648 driver.  From what I can tell, any problem you are having at this point is due to an issue in the DM648 driver.

I want to warn you, this is a very old driver at this point and support will be limited for it.  You may find that you are mostly on your own in debugging this driver, especially since very few in the community seem to be using this device any more.

Having said that, I spent some time digging through the driver code and I think the question you need to answer is this:

What is the result of the following code?

    #define myMDIO_ALIVE           MDIO_ALIVE(PhyDev->miibase)
    ...
    PhyAcks=myMDIO_ALIVE;
    PhyAcks&=PhyMask;   /* Only interested in 'our' Phys */

The above code is coming from the DM648 driver file cpsw_miimdio.c, in the function _cpsw_MDIO_FindingState() (around lines 545-546).

I think for some reason the "MDIO is not alive" and that is causing the failure (if you look down in the code just below the above two lines, you will see the error message that's showing up in your output).

I also checked into the PhyMask, and I think this is being set properly.  It is being set in ethdriver.c, in Setup_EmacInitConfig():

    /* Program default MAC SL config */
    for (i=0; i < CPSW3G_NUM_MAC_PORTS; i++)
    {
	    if(0 == i)
	    {
		    /* program phy_mask for port 0 */
		    iCfg->macInitCfg[i].PhyMask = 0x1;/*corresponding to Addr=0*/
		    iCfg->macInitCfg[i].MLinkMask = 0x0;
	    }
	    else
	    {
		    /* program phy_mask for port 1 */
		    iCfg->macInitCfg[i].PhyMask = 0x2;/*corresponding to address 1 */
		    iCfg->macInitCfg[i].MLinkMask = 0x0;
	    }/* end of else block covering settings for port 1 */

Again, I think the mask is being set properly, but the code is failing because the same corresponding bit is not set in:

 PhyAcks=myMDIO_ALIVE;

Resulting in the AND operation against the PhyMask to be zero:

    PhyAcks&=PhyMask;   /* Only interested in 'our' Phys */

For your reference, here are some other notes I have while tracing through the code, as this may be helpful for you:

1. cpgmacMacOpen() is called for each phy (?) but cpsw_MDIO_Init() is called with hard coded MDIO_BASEADDR.  Is this ok?

static void cpgmacMacOpen(Cpsw3gPort *hPort)
{
...
    cpsw_MDIO_Init(&gmacObject->PhyDev, MDIO_BASEADDR, instId,macInitCfg->PhyMask,
              macInitCfg->MLinkMask, 0,0, 0,
              cpswCfg->mdioBusFrequency,cpswCfg->mdioClockFrequency,0);
...
}

2. cpgmacMacOpen() is called (indirectly through fxn cpsw3g_port_open()) for each phy in EMAC_open()

    /* Open all ports */
    for(i=0;i<CPSW3G_NUM_PORTS; i++)
    {
        localDev.cpsw3gPort[i].portNum = i;
        localDev.cpsw3gPort[i].chNum = i;
        cpsw3g_port_open(&localDev.cpsw3gPort[i]);
    }/* end of for loop over PORTS */


Steve