Re: [PATCH v3] ARM: OMAP: i2c: fix interrupt flood during resume

[Shubhrajyoti Datta <omaplinuxkernel-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

On Fri, Oct 12, 2012 at 10:13 PM, Kevin Hilman
<khilman-1D3HCaltpLuhEniVeURVKkEOCMrvLtNR@public.gmane.org> wrote:
> "Strashko, Grygorii" <grygorii.strashko-l0cyMroinI0@public.gmane.org> writes:
>
>> Hi Kevin,
>>
>> yep, [1] is the same fix - thanks.
>
>> Hi Kalle,
>>
>> i've applied these changes and PM runtime fix on top of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git (omap2plus_defconfig)
>> Could you check it with your use case, pls? (just to be sure that idea is right)
>
> I think the ideas is right, but [1] introduces more potential problems
> since it disables the IRQ at the INTC well before being disabled at the
> device.
I fail to see this point. Current driver supports master mode only.
So unless there is a msg queued by the core. May be no interrupts should fire.

what I mean

msg -> conf -> intr -> completion/error  -> autosuspend.

>  With runtime PM autosuspend timeouts, that means any IRQs
> firing during the autosuspend delay will be lost, which basically
> nullifies the use of autosuspend.

so I do not expect any interrupts between completion/error -> autosuspend.

>
> Since Shubhrajyoti didn't respond to my runtime PM comments on the
> original patch,

my apologies for the delay.

> I'll respin this patch by moving the INTC disable/enable
> into the runtime PM callbacks and make the changelog a bit more clear.
>
> Grygorii, that pm_runtime_resume() change is needed to.  Can you spin a
> patch with just that change with a nice descriptive changelog about why
> it is needed?  Thanks.
>
> Kevin