Re: [PATCH] spi/spi_s3c64xx: Move to subsys_initcall()

[Mark Brown <broonie@opensource.wolfsonmicro.com>]

On Wed, Sep 08, 2010 at 01:55:39PM +0900, Jassi Brar wrote:
> On Tue, Sep 7, 2010 at 7:29 PM, Mark Brown
> <broonie@opensource.wolfsonmicro.com> wrote:
> > Allow the use of the S3C64xx SPI controller with things like PMICs by
> > moving the init up to subsys_initcall().

> Couldn't any user ever need to load it as a module?
> If no, we might as well drop the s3c64xx_spi_exit and s3c64xx_spi_remove

This doesn't prevent building as a module - when built as a module
subsys_initcall() is identical to module_init(), the change will only
affect the order in which things are done when the code is built into
the kernel otherwise it's a noop.

> as well and save space. Rather going a step further, shouldn't then all
> spi drivers be that way? Two steps further, why not every 'bus-driver' ?

Yes, we probably do need to do the same thing for all embedded SPI
controllers (as has already happened with embedded I2C controllers),
though it's less pressing since outside of a few CPUs it's much less
common for things like PMICs to be on SPI than I2C.