[PATCH/RESEND #3 1/7] OMAP3+: PM: SR: add suspend/resume handlers

Menon, Nishanth nm at ti.com
Fri Nov 18 19:35:53 EST 2011


On Wed, Nov 16, 2011 at 18:02, Kevin Hilman <khilman at ti.com> wrote:
> Felipe Balbi <balbi at ti.com> writes:
>
>> From: Nishanth Menon <nm at ti.com>
>>
>> SmartReflex should be disabled while entering low power mode due to
>> the following reasons:
> [...]
>
> Nishanth, in the end, didn't you decide to drop this patch?
>

Yes, I did eventually, once we implemented DVFS for GPU, Ducati, HSI,
and other drivers, there was no real way to ensure sequence of suspend
sequencing even after moving this to suspend_noirq. some of the other
reasons:

if I disabled Smartreflex and went to Nominal voltage on MPU, and say
MPU was at "performance" mode of 1.5GHz or so, thermal scenarios got
worse due on hot corner samples - these tend to have higher leakage
and thermal characteristics tend to be more pronounced. The option of
throttling frequency down while suspend was not really a good option
in comparison to switching off smart reflex in the last possible
moment - in pmxxx.c

Regards,
Nishanth Menon



More information about the linux-arm-kernel mailing list