Re: [RFT PATCH 1/9] drivers: qcom: rpmh-rsc: Clean code reading/writing regs/cmds

[Doug Anderson]

Hi,

On Wed, Mar 11, 2020 at 1:47 AM Maulik Shah <mkshah@xxxxxxxxxxxxxx> wrote:
>
> Hi,
>
> On 3/7/2020 5:29 AM, Douglas Anderson wrote:
> > This patch makes two changes, both of which should be no-ops:
> >
> > 1. Make read_tcs_reg() / read_tcs_cmd() symmetric to write_tcs_reg() /
> >    write_tcs_cmd().
>
> i agree that there are two different write function doing same thing except last addition (RSC_DRV_CMD_OFFSET * cmd_id)
>
> can you please rename write_tcs_cmd() to write_tcs_reg(), add above operation in it, and then remove existing write_tcs_reg().
> this way we have only one read and one write function.
>
> so at the end we will two function as,
>
> static u32 read_tcs_reg(struct rsc_drv *drv, int reg, int tcs_id, int cmd_id)
> {
>         return readl_relaxed(drv->tcs_base + reg + RSC_DRV_TCS_OFFSET * tcs_id +
>                              RSC_DRV_CMD_OFFSET * cmd_id);
> }
>
> static void write_tcs_reg(struct rsc_drv *drv, int reg, int tcs_id, int cmd_id,
>                           u32 data)
> {
>         writel_relaxed(data, drv->tcs_base + reg + RSC_DRV_TCS_OFFSET * tcs_id +
>                        RSC_DRV_CMD_OFFSET * cmd_id);
> }

I can if you insist and this is still better than the existing
(inconsistent) code.

...but I still feel that having two functions adds value here.


Anyone else who is CCed want to weigh in and tie break?


> > 2. Change the order of operations in the above functions to make it
> >    more obvious to me what the math is doing.  Specifically first you
> >    want to find the right TCS, then the right register, and then
> >    multiply by the command ID if necessary.
> With above change, i don't think you need to re-order this.
> specifically from tcs->base, we find right "reg" first and if it happens to be tcs then intended tcs, and then cmd inside tcs.

There was never any "need" to re-order.  That math works out to be the
same.  This is just clearer.

As an example, let's look at this:

struct point {
  int x;
  int y;
};
struct point points[10];

Let's say you have:
  void *points_base = &(points[0]);

...and now you want to find &(points[5].y).  What does your math look like?

a) points_base + (sizeof(struct point) * 5) + 4 ;

...or...

b) points_base + 4 + (sizeof(struct point) * 5);


Both calculations give the same result, but I am arguring that "a)" is
more intuitive.  Specifically you deal with the array access first and
then deal with the offset within the structure that you found.


-Doug