Re: [PATCH v3] mtd/nand: don't use {read,write}_buf for 8-bit transfers

[Ezequiel Garcia <ezequiel.garcia@free-electrons.com>]

Brian,

On Tue, Nov 26, 2013 at 11:35:12PM -0800, Brian Norris wrote:
> + Pekon, Ezequiel
> 
> Can one of you see how this patch works with your BeagleBones w/ x16
> NAND?
> 
> Also, do you think on of you could take a look at killing the
> nand_chip.read_word() callback, as Artem was originally requesting with
> this patch? It only has one user (for bad block checking in
> nand_block_bad(), which is currently only used when there is no BBT at
> all), and it looks like it could be replaced with a call to
> read_buf(mtd, buf, 2).
> 

Sure, sounds doable.

> Hi Uwe,
> 

Uwe,

If you're preparing a v4 to address Brian's feedback, then I'll see about testing
it with 8/16-bit flashes on my AM335x.

> On Tue, Nov 26, 2013 at 10:15:15PM +0100, Uwe Kleine-König wrote:
> > According to the Open NAND Flash Interface Specification (ONFI) Revision
> > 3.1 "Parameters are always transferred on the lower 8-bits of the data
> > bus." for the Get Features and Set Features commands.
> > 
> > So using read_buf and write_buf is wrong for 16-bit wide nand chips as
> > they use I/O[15:0]. The Get Features command is easily fixed using 4
> > times the read_byte callback. For Set Features implement a new
> > overwritable callback "write_byte". Still I expect the default to work
> > just fine for all controllers and making it overwriteable was just done
> > for symmetry.
> > 
> > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
> > ---
> >  drivers/mtd/nand/nand_base.c | 61 +++++++++++++++++++++++++++++++++++++++++---
> >  include/linux/mtd/nand.h     |  3 +++
> >  2 files changed, 60 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c
> > index bd39f7b..27d755b 100644
> > --- a/drivers/mtd/nand/nand_base.c
> > +++ b/drivers/mtd/nand/nand_base.c
> ...
> > @@ -2706,7 +2751,7 @@ static int nand_block_markbad(struct mtd_info *mtd, loff_t ofs)
> >  }
> >  
> >  /**
> > - * nand_onfi_set_features- [REPLACEABLE] set features for ONFI nand
> > + * nand_onfi_set_features - [REPLACEABLE] set features for ONFI nand
> 
> This change is unrelated. (Yes, the whitespace should be fixed, but
> probably not in this patch.)
> 
> >   * @mtd: MTD device structure
> >   * @chip: nand chip info structure
> >   * @addr: feature address.
> ...
> > @@ -2731,7 +2779,7 @@ static int nand_onfi_set_features(struct mtd_info *mtd, struct nand_chip *chip,
> >  }
> >  
> >  /**
> > - * nand_onfi_get_features- [REPLACEABLE] get features for ONFI nand
> > + * nand_onfi_get_features - [REPLACEABLE] get features for ONFI nand
> 
> Ditto.
> 
> >   * @mtd: MTD device structure
> >   * @chip: nand chip info structure
> >   * @addr: feature address.
> ...
> > @@ -2812,6 +2863,8 @@ static void nand_set_defaults(struct nand_chip *chip, int busw)
> >  		chip->block_markbad = nand_default_block_markbad;
> >  	if (!chip->write_buf || chip->write_buf == nand_write_buf)
> >  		chip->write_buf = busw ? nand_write_buf16 : nand_write_buf;
> > +	if (!chip->write_byte)
> 
> This check should be:
> 
> 	if (!chip->write_byte || chip->write_byte == nand_write_byte)
> 
> in order to match the rest of the functions, so they can be reset if the
> buswidth is detected to be x16, and we re-call nand_set_defaults()
> (e.g., when using NAND_BUSWIDTH_AUTO). See the comment:
> 
>   /* If called twice, pointers that depend on busw may need to be reset */
> 
> I know it's kind of ugly, but that's what we have for now.
> 
> > +		chip->write_byte = busw ? nand_write_byte16 : nand_write_byte;
> >  	if (!chip->read_buf || chip->read_buf == nand_read_buf)
> >  		chip->read_buf = busw ? nand_read_buf16 : nand_read_buf;
> >  	if (!chip->scan_bbt)
> 
> Brian

-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com