ECC Reset problems

[Justin Ipsen]

Thanks David.

Its seems to be a timing issue, although not with ECC resets.
We had to make the following changes to affect all delays used in doc2000.c
I had forgotten that I multiplied "cycles" by 6 to make the 32M DOC work.
I increased this to 9 and the 48M DOC now works fine.

Would this just be a specific board architecture thing?

Thanks again.

Justin.

    /* Perform the required delay cycles by reading from the appropriate
register */
    static void DoC_Delay(struct DiskOnChip *doc, unsigned short cycles)
    {
        volatile char dummy;
        int i;

        for (i = 0; i < cycles*12; i++) {
            if (DoC_is_Millennium(doc))
                dummy = ReadDOC(doc->virtadr, NOP);
            else
                dummy = ReadDOC(doc->virtadr, DOCStatus);
        }
    }



David Woodhouse wrote:

> justin.ipsen at ctam.com.au said:
> >  I'm running linux 2.4.1-pre11 on a powerPC platform. I have been
> > using a 32M doc2000 with no problems. I've just started to use a 48M
> > doc2000 but the On-Chip ECC engine seems to work just once after reset
> > and not again.   It looks like the "ECC Config" Resets function has
> > stopped working?
>
> Hmmm. I can't see any documented timing constraints on the ECC Reset. But
> try adding a delay between...
>
>                 /* Prime the ECC engine */
>                 WriteDOC(DOC_ECC_RESET, docptr, ECCConf);
>                 WriteDOC(DOC_ECC_EN, docptr, ECCConf);
>
> Might also be worth looking in include/linux/mtd/doc2000.h and changing the
> definition of DOC_ECC_RESET to DOC_ECC_RESV instead of zero.
>
> --
> dwmw2