Re: [PATCH v3 2/2] mtd: rawnand: Add NAND controller support on Intel LGM SoC

[Miquel Raynal <miquel.raynal@bootlin.com>]

Hi Vadivel MuruganX,

"Ramuthevar, Vadivel MuruganX"
<vadivel.muruganx.ramuthevar@linux.intel.com> wrote on Tue, 28 Apr 2020
14:50:35 +0800:

> Hi Boris,
> 
> On 28/4/2020 2:47 pm, Boris Brezillon wrote:
> > On Tue, 28 Apr 2020 14:40:58 +0800
> > "Ramuthevar, Vadivel MuruganX"
> > <vadivel.muruganx.ramuthevar@linux.intel.com> wrote:
> >   
> >> Hi Boris,
> >>
> >> On 28/4/2020 2:27 pm, Boris Brezillon wrote:  
> >>> On Tue, 28 Apr 2020 14:17:30 +0800
> >>> "Ramuthevar, Vadivel MuruganX"
> >>> <vadivel.muruganx.ramuthevar@linux.intel.com> wrote:  
> >>>    >>>> Hi Miquel,  
> >>>>
> >>>>       Thank you very much for the review comments and your time...
> >>>>
> >>>> On 27/4/2020 11:51 pm, Miquel Raynal wrote:  
> >>>>> Hi Ramuthevar,  
> >>>>>       >>>>>>> +static int ebu_nand_probe(struct platform_device *pdev)  
> >>>>>>> +{
> >>>>>>> +	struct device *dev = &pdev->dev;
> >>>>>>> +	struct ebu_nand_controller *ebu_host;
> >>>>>>> +	struct nand_chip *nand;
> >>>>>>> +	phys_addr_t nandaddr_pa;
> >>>>>>> +	struct mtd_info *mtd;
> >>>>>>> +	struct resource *res;
> >>>>>>> +	int ret;
> >>>>>>> +	u32 cs;
> >>>>>>> +
> >>>>>>> +	ebu_host = devm_kzalloc(dev, sizeof(*ebu_host), GFP_KERNEL);
> >>>>>>> +	if (!ebu_host)
> >>>>>>> +		return -ENOMEM;
> >>>>>>> +
> >>>>>>> +	ebu_host->dev = dev;
> >>>>>>> +	nand_controller_init(&ebu_host->controller);
> >>>>>>> +
> >>>>>>> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "ebunand");
> >>>>>>> +	ebu_host->ebu_addr = devm_ioremap_resource(&pdev->dev, res);
> >>>>>>> +	if (IS_ERR(ebu_host->ebu_addr))
> >>>>>>> +		return PTR_ERR(ebu_host->ebu_addr);
> >>>>>>> +
> >>>>>>> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "hsnand");
> >>>>>>> +	ebu_host->nand_addr = devm_ioremap_resource(&pdev->dev, res);
> >>>>>>> +	if (IS_ERR(ebu_host->nand_addr))
> >>>>>>> +		return PTR_ERR(ebu_host->nand_addr);
> >>>>>>> +
> >>>>>>> +	ret = device_property_read_u32(dev, "nand,cs", &cs);  
> >>>>>>
> >>>>>> CS ids should be encoded in the reg property (see [1]).  
> >>>>>
> >>>>> Is it your choice to only support a single CS or is it actually a
> >>>>> controller limitation?  
> >>>>
> >>>> Yes , its controller limitation to support only one CS  
> >>>
> >>> I'm pretty sure that's not true, otherwise you wouldn't have to select
> >>> the CS you want to use :P.  
> >>
> >> At a time it supports only one chip select.  
> > 
> > Yes, like 99% of the NAND controllers, but that doesn't mean you can't
> > support multi-CS chips. All you have to do is attach an array of
> > ebu_nand_cs to your ebu_nand_chip (as done in the atmel driver I
> > pointed to). nand_operation.cs tells you which CS (index in your
> > ebu_nand_cs array) a specific operation is targeting, and you can pick
> > the right MMIO range/reg value based on that.  
> 
> Agreed, sure I will add that and update next series of patches .

There are also many other places where you assume blindly that there
is only one CS. You can check the Atmel NAND controller driver as Boris
said and we will probably propose more little changes to be more
generic.

Thanks,
Miquèl