Re: [PATCH v3 13/13] spi: atmel-quadspi: add support for sam9x60 qspi controller

[Boris Brezillon <bbrezillon@xxxxxxxxxx>]

On Sat, 2 Feb 2019 08:58:25 +0000
<Tudor.Ambarus@xxxxxxxxxxxxx> wrote:

> >> @@ -117,6 +120,7 @@
> >>  #define QSPI_IFR_CRM                    BIT(14)
> >>  #define QSPI_IFR_NBDUM_MASK             GENMASK(20, 16)
> >>  #define QSPI_IFR_NBDUM(n)               (((n) << 16) & QSPI_IFR_NBDUM_MASK)
> >> +#define QSPI_IFR_APBTFRTYP_READ		BIT(24)  
> > 
> > Maybe add a comment saying it's only available on SAM9X60 or prefix it
> > with SAM9X60.  
> 
> I'll add a comment. The macro name is more generic how it is now and can be used
> by future versions of the IP. Hypothetically speaking, if we rename it to
> QSPI_IFR_SAM9x60_TFSFR_READ and other sam9x will come out, then I'll have to
> rename this macro again, to make it more generic.	

Okay.


> >> +static int atmel_qspi_set_cfg(struct atmel_qspi *aq,
> >> +			      const struct spi_mem_op *op,
> >> +			      struct atmel_qspi_cfg *cfg)
> >> +{
> >> +	void __iomem *base = aq->regs;
> >> +	int ret;
> >> +
> >> +	/* Set the QSPI controller in Serial Memory Mode */
> >> +	if (aq->smm != QSPI_MR_SMM) {
> >> +		writel_relaxed(QSPI_MR_SMM, base + QSPI_MR);  
> > 
> > 					    aq->reqs +
> > 
> > and you can get rid of base.  
> 
> I will wait your reasons on this, see 3/13

ad->regs is only dereferenced once in this function, so there's even
less reasons to add a local var. 

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/