• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 97 subscribers
  • Views 494 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.

AM3517 and LAN9514 Driver Problems

Greg Sulfid
Intellectual 610 points
Other Parts Discussed in Thread: AM3517

Hello,

I'm using the AM3517 and the latest CE BSP (ARM_A8_01_01_00_patch_01) with an extra USB LAN chip from SMSC (LAN9514) on the OTG port.

My problem is, that the initialization failed in the omap_usbh driver module (chw.cpp:145 / ReadFifo) - it pop-up with a datatype misalignment:

BOOL ReadFIFO(DWORD* pFifoBase, void *pData, DWORD size)
{
    DWORD total  = size / 4;
    DWORD remain = size % 4;
    DWORD i         = 0;

    DWORD* pDword = (DWORD*)pData;

    volatile ULONG *pReg = (volatile ULONG*)pFifoBase;

    // this is 32-bit align
    for (i = 0; i < total; i++)
    {
        *pDword++ = INREG32(pReg);
    }

The LAN9514 chip has also an USB hub inside, and an USB stick on one of its port did its job without problems.

I would appreciate every hint you can give me.

Many thanks ahead!

  • 0
    Intellectual 610 points

    What confused me with this function is, that the comment says "this is 32-bit align", so I thought that the address from pData make absolutely no sense when its not aligned.

    But I changed the code to a memcpy version and then I get the LAN part of the 9514 up and running and it did its work:

        for (i = 0; i < total; i++)
        {
             dwTemp = INREG32(pReg);
             memcpy((DWORD*)pData + i, &dwTemp, sizeof(dwTemp));
        }
        if (remain)
        {
            dwTemp = INREG32(pReg);
            memcpy((DWORD*)pData + total, &dwTemp, remain);
        }

    But could a problem show up because the address is not 32-bit aligned? It would be great if someone had an answer so I must not look at all the USB stuff in detail :-)

  • 0 in reply to Greg Sulfid
    Intellectual 610 points

    It seems that memcpy is also not unalignment "safe" in the release version: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.faqs/ka3934.html

    Defining the pointer as a possible unaligned address did work for all circumstances: http://msdn.microsoft.com/en-us/library/ms177389.aspx

    DWORD __unaligned * pDword = (DWORD*)pData;