Re: [PATCH v2 02/10] mtd: rawnand: denali: refactor syndrome layout handling for raw access

[Miquel Raynal]

Hi Masahiro,

Masahiro Yamada <yamada.masahiro@xxxxxxxxxxxxx> wrote on Tue, 5 Mar
2019 18:20:22 +0900:

> Hi Miquel,
> 
> On Mon, Mar 4, 2019 at 6:01 PM Miquel Raynal <miquel.raynal@xxxxxxxxxxx> wrote:
> >
> > Hi Masahiro,
> >
> > Masahiro Yamada <yamada.masahiro@xxxxxxxxxxxxx> wrote on Tue, 12 Feb
> > 2019 16:12:54 +0900:
> >  
> > > The Denali IP adopts the syndrome page layout (payload and ECC are
> > > interleaved). The *_page_raw() and *_oob() callbacks are complicated
> > > because they must hide the underlying layout used by the hardware,
> > > and always return contiguous in-band and out-of-band data.
> > >
> > > Currently, similar code is duplicated to reorganize the data layout.
> > > For example, denali_read_page_raw() and denali_write_page_raw() look
> > > almost the same.
> > >
> > > The idea for refactoring is to split the code into two parts:
> > >   [1] conversion of page layout
> > >   [2] what to do at every ECC chunk boundary
> > >
> > > For [1], I wrote denali_raw_payload_op() and denali_raw_oob_op().
> > > They manipulate data for the Denali controller's specific page layout
> > > of in-band, out-of-band, respectively.  
> >
> > Could you please comment the purpose of these two functions in the code
> > as well?  
> 
> 
> OK, I will.
> 
> 
> 
> > >
> > > The difference between write and read is just the operation at
> > > ECC chunk boundaries. For example, denali_read_oob() calls
> > > nand_change_read_column_op(), whereas denali_write_oob() calls
> > > nand_change_write_column_op(). So, I implemented [2] as a callback
> > > passed into [1].
> > >
> > > Signed-off-by: Masahiro Yamada <yamada.masahiro@xxxxxxxxxxxxx>
> > > ---
> > >
> > > Changes in v2: None
> > >
> > > drivers/mtd/nand/raw/denali.c | 354 +++++++++++++++++++-----------------------
> > > 1 file changed, 163 insertions(+), 191 deletions(-)  
> >
> > Too bad that the size of the driver did not shrink more than that :)  
> 
> Indeed, less than expected.
> 
> But, please do not miss this commit is adding
> error check to every operation.
> 
> Prior to this commit, the code just ignored the return code
> because 97d90da8a886949f simply replaced old hooks
> despite the new ones return the error code.
> 
> 
> Generally, every error check costs two lines
> in the following form:
> 
> 
>   ret = (do something)
>   if (ret)
>            return ret;
> 

Right!

> 
> 
> > > +
> > > +static int denali_memcpy_in(void *buf, unsigned int offset, unsigned int len,
> > > +                         void *priv)
> > > +{
> > > +     memcpy(buf, priv + offset, len);
> > > +     return 0;
> > >  }  
> >
> > Maybe this "callback" and the (_out cousin) could be part of you
> > controller's structure,
> > and you could use a read/write flag instead of
> > passing the functions' pointer?  
> 
> This is what the old code does.
> 
> There are 4 callbacks for the combination
> of raw/oob, and write/read.
> 
> I do not know how your suggestion looks like,
> and whether it will make the code cleaner.
> 

Please give it a try!

Thanks,
MiquÃl