On Mon, May 25, 2009 at 01:59:20PM +0100, Russell King - ARM Linux wrote: > On Fri, May 22, 2009 at 03:45:41PM -0700, Tony Lindgren wrote: > > + /* link regulators to MMC adapters */ > > + ldp_vmmc1_supply.dev = mmc[0].dev; > This feels wrong - setup after registering devices. All it takes is to > have to move various device drivers to be earlier in the initialization > ordering and things start breaking in unexpected ways. > If MMC controls the Vmmc1 supply, then why does the Vmmc1 supply need > the struct device pointer for the MMC controller? Like the clock API the regulator API looks up clients for regulators based on a combination of the struct device for the client driver and the name of the supply. Unfortunately with both the I2C and SPI APIs the struct device is only available once the driver has probed so this can't be set up with static data prior to probe, hence this sort of bodge. Long term I think we'll want to fix I2C and SPI to let us have static data for this. -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
