[PATCH v3 1/9] Input: pmic8xxx-pwrkey - Migrate to regmap APIs

On Wed, Feb 26, 2014 at 04:43:03PM -0800, Stephen Boyd wrote:

On 02/26/14 16:30, Josh Cartwright wrote:

quoted

On Wed, Feb 26, 2014 at 04:20:03PM -0800, Stephen Boyd wrote:

quoted

On 02/26/14 16:13, Josh Cartwright wrote:

quoted

On Wed, Feb 26, 2014 at 04:05:40PM -0800, Stephen Boyd wrote:

quoted

On 02/26/14 15:59, Josh Cartwright wrote:

However, I think it needs to be made clear that your if "Modernize pm8921
with irqdomains, regmap, DT" lands before this patchset, this will start
breaking randconfig builds.  It isn't clear to me how this dependency
should be handled.

Hmm? If pm8921 lands before this what build breakage is there? The
pm8xxx_read/write APIs are still exposed.

Well, with the pm8921 patchset in place, the keypad driver becomes
selectable, which, when built leads to the error below.  But AFAICT,
this isn't even addressed in this patchset?!

drivers/input/keyboard/pmic8xxx-keypad.c:24:35: fatal error: linux/mfd/pm8xxx/gpio.h: No such file or directory
 #include <linux/mfd/pm8xxx/gpio.h>
                                   ^

So, I'm confused...maybe the pm8921 cleanup patchset should be marking
the keypad driver BROKEN?

Ah, yes probably. I have a patch locally that rips out the gpio stuff
because this driver has been uncompilable since it was introduced. I may
as well send that along with this series so that people can actually
compile this driver. Or like you say, send it along with the removal of
the BROKEN in pm8921 Kconfig.

Yeah, I think it should probably land before MFD_PM8XXX's BROKEN
removal, just to eliminate randconfig breakage noise.  I double-checked,
and it looks like at least the other drivers that depend on MFD_PM8XXX
still build (with a couple sparse warnings).

Also.....it's driving me nuts that some of the pm8xxx drivers use
"pm8xxx-foo", and others use "pmic8xxx-foo"...

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation