Re: [PATCH 2/3] spi: spacemit: introduce SpacemiT K1 SPI controller driver

[Alex Elder]

On 9/18/25 9:39 AM, Yixun Lan wrote:
> > > > +	virt = drv_data->ioaddr + SSP_TOP_CTRL;
> > > > +	val = readl(virt);
> > > > +	val |= TOP_TRAIL;	/* Trailing bytes handled by DMA */
> > > > +	writel(val, virt);
> > > I'd prefer to do like this, it's more easy for people to grep..
> > > 	val = readl(drv_data->ioaddr + SSP_TOP_CTRL) | TOP_TRAIL;
> > > 	writel(val, drv_data->ioaddr + SSP_TOP_CTRL);
> > This is an idiom I use to make it very clear that:
> > - The address being read is exactly the same as what's being
> >     written
> > - The value read is being updated with bits/values
> >
> > I find that putting the "| TOP_TRAIL" on the same line as the
> > readl() call obscures things a bit.  Like my eye doesn't notice
> > it as readiliy somehow...
> fair, let's put it into another line
>
> > Yours is a pure coding style comment.  There are two pieces, and
> > I'd like you to tell me how strongly you feel about them:
> > - Using virt to grab the address being written and read (versus
> >     just using drv_data->ioaddr + SSP_TOP_CTRL twice)
> > - Put the "| TOP_TRAIL" on the same line as the readl() (versus
> >     having that be assigned on a separate line).
> > To me, the second one is more important than the first.
> >
> > Let me know how strongly you feel about these and I'll update
> > my convention througout.
> I'd strongly prefer not to introduce 'virt', so be something like this:
>   	val = readl(drv_data->ioaddr + SSP_TOP_CTRL);
>   	val |= TOP_TRAIL;
>   	writel(val, drv_data->ioaddr + SSP_TOP_CTRL);

OK.  I'll do it this way throughout the driver in the
next version.

					-Alex