[U-Boot] [RFC PATCH v2 09/11] sf_mtd: Simply mtd operations

[Miquel Raynal]

Hi Vignesh,

Vignesh R <vigneshr at ti.com> wrote on Wed, 5 Dec 2018 13:30:27 +0530:

> Hi Boris,
> 
> On 04/12/18 6:19 PM, Boris Brezillon wrote:
> > On Tue, 4 Dec 2018 17:56:57 +0530
> > Vignesh R <vigneshr at ti.com> wrote:
> >   
> >> Now that there is new SPI NOR framework, simplify mtd device
> >> registration and read/write/erase operations.
> >>
> >> Signed-off-by: Vignesh R <vigneshr at ti.com>
> >> ---
> >>  drivers/mtd/spi/sf_internal.h |  2 +-
> >>  drivers/mtd/spi/sf_mtd.c      | 39 ++++++++++++-----------------------
> >>  drivers/mtd/spi/sf_probe.c    |  2 +-
> >>  3 files changed, 15 insertions(+), 28 deletions(-)
> >>
> >> diff --git a/drivers/mtd/spi/sf_internal.h b/drivers/mtd/spi/sf_internal.h
> >> index 7e7d400cdbdf..8b445bb0b506 100644
> >> --- a/drivers/mtd/spi/sf_internal.h
> >> +++ b/drivers/mtd/spi/sf_internal.h
> >> @@ -99,6 +99,6 @@ int spi_flash_cmd_get_sw_write_prot(struct spi_flash *flash);
> >>  
> >>  #ifdef CONFIG_SPI_FLASH_MTD
> >>  int spi_flash_mtd_register(struct spi_flash *flash);
> >> -void spi_flash_mtd_unregister(void);
> >> +void spi_flash_mtd_unregister(struct spi_flash *flash);
> >>  #endif
> >>  #endif /* _SF_INTERNAL_H_ */
> >> diff --git a/drivers/mtd/spi/sf_mtd.c b/drivers/mtd/spi/sf_mtd.c
> >> index 58d7e4439903..9c07ba3cadf7 100644
> >> --- a/drivers/mtd/spi/sf_mtd.c
> >> +++ b/drivers/mtd/spi/sf_mtd.c
> >> @@ -9,17 +9,15 @@
> >>  #include <linux/mtd/mtd.h>
> >>  #include <spi_flash.h>
> >>  
> >> -static struct mtd_info sf_mtd_info;
> >>  static char sf_mtd_name[8];
> >>  
> >>  static int spi_flash_mtd_erase(struct mtd_info *mtd, struct erase_info *instr)
> >>  {
> >> -	struct spi_flash *flash = mtd->priv;
> >>  	int err;
> >>  
> >>  	instr->state = MTD_ERASING;
> >>  
> >> -	err = spi_flash_erase(flash, instr->addr, instr->len);
> >> +	err = mtd->_erase(mtd, instr);  
> > 
> > Please don't dereference mtd hooks directly. I'm pretty sure the
> > mtd_erase() overhead is negligible, and if it's not, you can
> > implement dummy wrappers for the SPL case to reduce it to zero.
> >   
> 
> I did start with mtd_* APIs here. But, ended up with bunch of patches
> just to get MTD configs in shape to satisfy compile time dependencies.
> I will send a patch to use mtd_* APIs here once Miquel's MTD config
> cleanups are in. Thanks for the review!
> 

As an FYI, it took me more than 30 Travis builds to get the
MTD Kconfigs/Makefiles (+ defconfigs) in shape :)

To find faulty configurations (for example, configurations using
CONFIG_SPI_FLASH_MTD but without enabling CONFIG_MTD) I wrote a small
script just to open these files automatically. As a reference if
someone wants to use/improve it, it is there [1].

[1] http://code.bulix.org/4ep8u5-518535

Thanks,
Miquèl