SDIO on AT91SAMPM9263

[Bart Cerneels]

On Wed, Nov 4, 2009 at 21:03, Dan Williams <dcbw at redhat.com> wrote:
> On Wed, 2009-11-04 at 12:27 +0100, Bart Cerneels wrote:
>> Hi all
>>
>> I've hit a problem with libertas and a marvell sd8385 using FW version
>> 5.0.11p0 and 5.0.20p0.
>>
>> It's the same timeout issues like [1] but with completely difference
>> processor architecture and MMC host controller.
>> While I suspected the host controller to be the problem, this makes me
>> think its the libertas-sdio driver again.
>>
>> So what is the most likely culprit here firmware (tried 2 versions),
>> libertas-core, libertas-sdio or perhaps it's the MMC host screwing
>> everything up.
>>
>> I'm not really experience with SD but I absolutely have to fix this
>> issue this month so if you can help me you can expect a patch if it
>> turns out to be a libertas bug.
>
> Does the SDHC have any quirks or errata with respect to DMA, block size,
> multi-block transfers, or interrupts?
>
> Dan
>

It does, but I have investigated each one and they are all handled in
the host controller driver (at91_mci). Also the hardware configuration
is correct w.r.t. the errata (one slot can not handle SDIO
interrupts).

Assuming this is padding or a different quirk fix that goes wrong: how
can one tell? i.e. are there specific commands that will surely fail
and I'll see show up in the logs?
The only consistent problem I'm seeing ATM is response timeouts, which
could be FW going belly up or SD host problems. Both can possibly be
avoided in libertas.

Bart