Re: [PATCH] mtd: nand: pxa3xx_nand: fix dmaengine initialization

[Robert Jarzmik]

Ezequiel Garcia <ezequiel@xxxxxxxxxxxxxxxxxxxx> writes:

> On 12 February 2016 at 19:29, Robert Jarzmik <robert.jarzmik@xxxxxxx> wrote:
>> When the driver is initialized in a pure device-tree platform, the
>> driver's probe fails allocating the dma channel :
>> [  525.624435] pxa3xx-nand 43100000.nand: no resource defined for data DMA
>> [  525.632088] pxa3xx-nand 43100000.nand: alloc nand resource failed
>>
>> The reason is that the DMA IO resource is not acquired through platform
>> resources but by OF bindings.
>>
>> Fix this by ensuring that DMA IO resources are only queried in the non
>> device-tree case.
>>
>> Fixes: 8f5ba31aa565 ("mtd: nand: pxa3xx-nand: switch to dmaengine")
>> Signed-off-by: Robert Jarzmik <robert.jarzmik@xxxxxxx>
>> ---
>>  drivers/mtd/nand/pxa3xx_nand.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/mtd/nand/pxa3xx_nand.c b/drivers/mtd/nand/pxa3xx_nand.c
>> index a168cbcc1086..afd487d4b67f 100644
>> --- a/drivers/mtd/nand/pxa3xx_nand.c
>> +++ b/drivers/mtd/nand/pxa3xx_nand.c
>> @@ -1750,7 +1750,7 @@ static int alloc_nand_resource(struct platform_device *pdev)
>>         if (ret < 0)
>>                 return ret;
>>
>> -       if (use_dma) {
>> +       if (!np && use_dma) {
>>                 r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
>>                 if (r == NULL) {
>>                         dev_err(&pdev->dev,
>
> Looking through the kernel tree, this change seems to be correct.
>
> However, I'm still wondering how DMA resources are obtained in the
> device-tree case. Can you explain it to me?

In the pxa case, look at my extract in [1].
The lines to consider are :
			dmas = <&pdma 97 3>;
			dma-names = "data";

Now have a look at the function pxad_dma_xlate() in drivers/dma/pxa_dma.c.
You'll see that the "97" ends up in the chanel drcmr, and the "3" in the
channel's prio, when the chanel is requested in the nand driver by calling
dma_request_slave_channel_compat(). In the latter case, pxad_filter_fn() is not
used, it's the pxad_dma_xlate() which is used instead.

Is this what you were looking for ?

Cheers.

-- 
Robert

[1] Extract of my zylonite310.dts
		nand0: nand@43100000 {
			compatible = "marvell,pxa3xx-nand";
			reg = <0x43100000 90>;
			interrupts = <45>;
			clocks = <&clks CLK_NAND>;
			dmas = <&pdma 97 3>;
			dma-names = "data";
			#address-cells = <1>;
			#size-cells = <1>;	
			status = "disabled";
		};