* Rajendra Nayak [off-list ref] [120224 20:14]:

On Saturday 25 February 2012 06:03 AM, Tony Lindgren wrote:

quoted

Hi,

* Rajendra Nayak[off-list ref]  [120224 01:27]:

quoted

omap_hsmmc_late_init() adds deferred (if any) mmc devices which are
dependent on twl4030-gpio device to be available.
If twl4030-gpio is built as a module and inserted and deleted multiple
times, the mmc devices also should be added and deleted accordingly.
Split omap_hsmmc_late_init() into omap_hsmmc_deferred_add() and
omap_hsmmc_deferred_del() to handle this.
The .setup board hook for twl4030-gpio handles the runtime adding
on deferred mmc devices. Similarly a .teardown board hook for twl4030-gpio
should handle the runtime deletion by calling omap_hsmmc_deferred_del().
This is done for all existing omap3 boards in subsequent patches.

This gets complex.. But I think I've found a simpler way where we
can use the twl gpio offsets and let the drivers take care of the rest.
I'll post some patches hopefully on Monday.

I agree it gets complex. If you have a simpler way to handle this, that
would be great.

Posted now, sorry it too a bit longer, I got distracted with other
stuff. The thread is "[PATCH 0/4] Start getting rid of pdata callbacks
with gpio_find_by_chip_name()". Some further testing would be
appreciated.

Regards,

Tony