Re: [PATCH V4] mtd: rawnand: qcom: update last code word register

[mdalam]

On 2021-02-12 13:49, Miquel Raynal wrote:
Hello,
mda...@codeaurora.org wrote on Fri, 12 Feb 2021 01:00:47 +0530:

On 2021-02-11 19:37, Miquel Raynal wrote:
> Hello,
>
> Manivannan Sadhasivam <manivannan.sadhasi...@linaro.org> wrote on Wed,
> 10 Feb 2021 14:31:44 +0530:
>
>> On Fri, Jan 29, 2021 at 03:09:19AM +0530, Md Sadre Alam wrote:
>> > From QPIC version 2.0 onwards new register got added to
>> > read last codeword. This change will add the READ_LOCATION_LAST_CW_n
>> > register.
>> >
>> > For first three code word READ_LOCATION_n register will be
>> > use.For last code word READ_LOCATION_LAST_CW_n register will be
>> > use.
>
> Sorry for the late notice, I think the patch is fine but if you don't
> mind I would like to propose a small change that should simplify your
> patch a lot, see below.
>
>> >
>> > Signed-off-by: Md Sadre Alam <mda...@codeaurora.org>
>> >> Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasi...@linaro.org>
>> >> Thanks,
>> Mani
>> >> > ---
>> > [V4]
>> >  * Modified condition for nandc_set_read_loc_last() in 
qcom_nandc_read_cw_raw().
>> >  * Added one additional argument "last_cw" to the function 
config_nand_cw_read()
>> >    to handle last code word condition.
>> >  * Changed total number of last code word register 
"NAND_READ_LOCATION_LAST_CW_0" to 4
>> >    while doing code word configuration.
>> >  drivers/mtd/nand/raw/qcom_nandc.c | 110 
+++++++++++++++++++++++++++++---------
>> >  1 file changed, 84 insertions(+), 26 deletions(-)
>> >
>> > diff --git a/drivers/mtd/nand/raw/qcom_nandc.c 
b/drivers/mtd/nand/raw/qcom_nandc.c
>> > index 667e4bf..9484be8 100644
>> > --- a/drivers/mtd/nand/raw/qcom_nandc.c
>> > +++ b/drivers/mtd/nand/raw/qcom_nandc.c
>> > @@ -48,6 +48,10 @@
>> >  #define  NAND_READ_LOCATION_1            0xf24
>> >  #define  NAND_READ_LOCATION_2            0xf28
>> >  #define  NAND_READ_LOCATION_3            0xf2c
>> > +#define  NAND_READ_LOCATION_LAST_CW_0    0xf40
>> > +#define  NAND_READ_LOCATION_LAST_CW_1    0xf44
>> > +#define  NAND_READ_LOCATION_LAST_CW_2    0xf48
>> > +#define  NAND_READ_LOCATION_LAST_CW_3    0xf4c
>> >
>> >  /* dummy register offsets, used by write_reg_dma */
>> >  #define  NAND_DEV_CMD1_RESTORE           0xdead
>> > @@ -187,6 +191,12 @@ nandc_set_reg(nandc, NAND_READ_LOCATION_##reg,        
               \
>> >         ((size) << READ_LOCATION_SIZE) |                    \
>> >         ((is_last) << READ_LOCATION_LAST))
>> >
>> > +#define nandc_set_read_loc_last(nandc, reg, offset, size, is_last)       \
>> > +nandc_set_reg(nandc, NAND_READ_LOCATION_LAST_CW_##reg,                   \
>> > +       ((offset) << READ_LOCATION_OFFSET) |                \
>> > +       ((size) << READ_LOCATION_SIZE) |                    \
>> > +       ((is_last) << READ_LOCATION_LAST))
>> > +
>
> You could rename the macro nandc_set_read_loc() into
> nandc_set_read_loc_first() or anything else that make sense, then have
> a helper which does:
>
> nandc_set_read_loc()
> {
>    if (condition for first)
>            return nandc_set_read_loc_first();
>    else
>            return nandc_set_read_loc_last();
> }
>

   Yes this is more precise way & simplify the patch a lot.
   But for this i have to change these two macro as a function.

   nandc_set_read_loc() & nandc_set_read_loc_last().

   Since for last code word register we are using Token Pasting 
Operator##.

   So if i am implementing like the below.

   /* helper to configure location register values */
   static void nandc_set_read_loc(struct qcom_nand_controller *nandc, 
int reg,
                   int offset, int size, int is_last, bool last_cw)
   {
           if (last_cw)
                   return nandc_set_read_loc_last(nandc, reg, offset, 
size, is_last);
           else
                   return nandc_set_read_loc_first(nandc, reg, offset, 
size, is_last);
  }

   So here for macro expansion reg should be a value not a variable 
else it will be expended like
   NAND_READ_LOCATION_LAST_CW_reg instead of 
NAND_READ_LOCATION_LAST_CW_0,1,2,3 etc.

I know it involves a little bit more computation but I wonder if using
funcs instead of macros here would not be nicer? Perhaps something 
like:

        loc = is_last ? NAND_READ_LOCATION /* 0xf20 */ :
NAND_READ_LOCATION_LAST /* 0xf40 */;
        loc += reg * 2;

  I have added a helper function to update location register value in V5 
patch.
  Please check the patch.

  the call for nandc_set_read_loc() as nandc_set_read_loc(nandc, 0, 
read_loc, data_size1, 0, true); ---> for last code word.
  nandc_set_read_loc(nandc, 0, read_loc, data_size1, 0, false); ---> 
for first three code wrod.

I think it's best to forward 'cw' as a parameter and do the
computation of is_last locally.

  So is this ok for you to convert these two macro into function ?

> And in the rest of your patch you won't have to touch anything else.
>
> Thanks,
> Miquèl

Thanks,
Miquèl