Re: [PATCH 2/2] gpio/omap: add *remove* callback in platform_driver

Hi Linus,

Linus Walleij [off-list ref] writes:

On Thu, Jul 12, 2012 at 1:25 AM, Kevin Hilman [off-list ref] wrote:

quoted

There is quite a bit of other things to do in remove to properly cleanup
what is done in probe.

OK I'm dropping this patch for now...

Thanks.

For future reference...  as one of the OMAP maintainers, I request that
you not pull/merge OMAP GPIO patches unless they've had a bit of review
and exposure.  Unfortunately, we have a history with this driver where
regressions have been introduced and the maintainers end up having to
find and fix them.  

In the case of OMAP GPIO, unless it's an obvious fix, I would recommend
you wait at least until you see some acks/tested tags from any of

- Santosh Shilimkar [off-list ref]
- Rajendra Nayak [off-list ref]
- Benoit Cousson [off-list ref]

or Tony, Paul or myself.

For major series, I have been collecting/queueing them for Grant after
ensuring they have been well reviewed and well tested (although I am
eagerly hoping to hand off this role to someone else.)

Thanks,

Kevin

On Thu, Jul 12, 2012 at 7:48 PM, Kevin Hilman [off-list ref] wrote:

In the case of OMAP GPIO, unless it's an obvious fix, I would recommend
you wait at least until you see some acks/tested tags from any of

- Santosh Shilimkar [off-list ref]
- Rajendra Nayak [off-list ref]
- Benoit Cousson [off-list ref]

or Tony, Paul or myself.

Instead of trying to store this information in my and Grants brains and
us forgetting it, what about patching MAINTAINERS to reflect the fact
instead? That's better I think, plus I use that file a lot.

For major series, I have been collecting/queueing them for Grant after
ensuring they have been well reviewed and well tested (although I am
eagerly hoping to hand off this role to someone else.)

Listing it under your GIT tree in MAINTAINERS for this driver will make
this work better I think.

One path for OMAP GPIO patches, simple. It's obviously the
ambiguity that cause the trouble. Then you can also decide
on each cycle whether to send these to GPIO or ARM SoC
etc.

Yours,
Linus Walleij