On 18 November 2014 at 14:28, Scott Jiang <scott.jiang.li...@gmail.com> wrote: > 2014-11-17 3:23 GMT+08:00 Jagan Teki <jagannadh.t...@gmail.com>: >> Hi Scott, >> >> On 25 September 2014 14:55, Scott Jiang <scott.jiang.li...@gmail.com> wrote: >>> SPI3 controller is not only used on BF609 platform. So we add a common >>> controller >>> driver and leave machine specific configuration in board drivers. >>> Remove obsolete spi6xx.h and select new board driver in configuration file. >>> >>> Signed-off-by: Scott Jiang <scott.jiang.li...@gmail.com> >>> --- >>> drivers/spi/Makefile | 1 + >>> drivers/spi/adi_spi3.c | 222 ++++++++++++++++ >> >> I think adi_spi3.c is a copy of bfin_spi6xx.c with excluded-ed stuff of bf609 >> >> And also the difference is of chipselect and slave setup handling, why can't >> we use same driver my making this difference stuff in particular macro. >> >> Please try in that sense, It's not looking good to have one driver >> file with only >> having chip selct handing and slave setup. >> > > Would you mind moving spi_cs_is_valid() to arch/xx/include/asm/mach-xxx/spi.h? > Because the number of spi controllers and the cs number of each controller > might > be different for different board. It use macros to do this before, > while it's more difficult > as boards become more.
IMHO, it's not a good idea to move the driver stuff to arc/xx/include I do understand that the this logic is more designed towards specific board. Solutions on my mind: - try to use dm - then make a cs logic and get the attributes from the dts. thanks! -- Jagan. _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
