[PATCH] mtd: nand: pxa3xx-nand: prevent DFI bus lockup on removal

[Ezequiel Garcia]

Robert,

On 24 August 2015 at 15:24, Robert Jarzmik <robert.jarzmik at free.fr> wrote:
> Ezequiel Garcia <ezequiel at vanguardiasur.com.ar> writes:
>
>> Should we worry about having two definitions for the same bit?
>> Would it be too ugly to mix the two meaning? Something like this:
>>
>> /* This bit has two different meanings on NFCv1 and NFCv2 */
>> #define NDCR_STOP_ON_UNCOR_ARB_CNTL (0x1 << 19)
> I don't find that very pretty, but if you want I can put that in the patch
> instead.
>

Yeah, it's far from pretty.

OK, another idea. How about this:

#define NFCV2_NDCR_STOP_ON_UNCOR (0x1 << 19)
#define NFCV1_NDCR_ARB_CNTL      (0x1 << 19)

Or maybe I'm just bike-shedding here? I figured it's important
to document the bit has different meanings in different
hardware versions.

Anyway, just submit a v2 with whatever you feel it's less ugly.

>>> @@ -1784,6 +1787,8 @@ static int pxa3xx_nand_remove(struct platform_device *pdev)
>>>              free_irq(irq, info);
>>>      pxa3xx_nand_free_buff(info);
>>>
>>
>> I think a comment here explaining how this disables DFI arbitration and
>> how clearing it grants DFI access to SMC only.
>>
>> While here, we might want to document how the whole arbiter applies to
>> PXA only, since the DFI bus is shared there.
> Ok, for v2. I take it that DFI bus is not shared on Armada, lucky you.

Right, seems not shared.

> Maybe
> something like :
>
> /*
>  * In the pxa3xx case, the DFI bus is shared between the SMC and NFC. In order
>  * to prevent a lockup of the system bus, the DFI bus arbitration is granted
>  * to SMC upon driver removal. This is done by setting the x_ARB_CNTL bit,
>  * which also prevents the NAND to have access to the bus anymore.
>  */
>

That'll work. Feel free to send a v2.
-- 
Ezequiel García, VanguardiaSur
www.vanguardiasur.com.ar