Re: [PATCH v4 0/7] cpufreq:boost: CPU Boost mode support

[Lukasz Majewski]

Hi Viresh,

> This patch series introduces support for CPU overclocking technique
> called Boost.
> 
> It is a follow up of a LAB governor proposal. Boost is a LAB
> component:
> http://thread.gmane.org/gmane.linux.kernel/1484746/match=cpufreq
> 
> Boost unifies hardware based solution (e.g. Intel Nehalem) with
> software oriented one (like the one done at Exynos).
> For this reason cpufreq/freq_table code has been reorganized to
> include common code.
> 
> Important design decisions:
> 
> - Boost related code is compiled-in unconditionally and disabled by
> default The cpufreq_driver is responsibile for setting
> boost_supported flag and providing enable_boost callback(if HW
> support is needed)
> 
> - struct cpufreq_driver has been extended with boost related fields:
>         -- boost_supported - when driver supports boosting
>         -- enable_boost - callback to function, which is necessary to
>            enable boost in the processor (like it is done at Intel)
> 
> - Boost sysfs attribute (/sys/devices/system/cpu/cpufreq/boost) is
> always provided.
> It will be read only, until either cpufreq_driver defines
> boost_supported flag or device tree's cpufreq "boost_mode" attribute
> is defined. It is consistent with legacy acpi API.
> 
> - No special spin_lock for Boost was created. The one from cpufreq
> core was reused.
> 
> - All available policies are now stored in a list.
> 
> - The Boost code doesn't rely on any policy. When boost state is
> changed, then the policy list is iterated and proper adjustements are
> done.
> 
> - Number of busy cpus (with load > 90%) is calculated and exported as
> a global sysfs attribute.
> 
> - The SW based boost can be only enabled when up to one core is
> regarded as a "busy" one. For safety reasons boost is disabled when
> more cores comes into play.
> 
> - To improve safety level, the thermal framework is also extended to
> disable software boosting, when thermal trip point is reached.
> 
> New patches for v4:
> cpufreq: Store cpufreq policies in a list
> cpufreq: Calculate number of busy CPUs
> cpufreq: Enable software boost only when up to one busy core is
> running thermal:boost: Disable boost when trip point is reached
> 
> Tested at: HW:
>         Exynos 4412 3.10 linux
>         Exynos 4210 3.10 linux
> Compile tested x86_64 defconfig (acpi) - help with HW (Intel Nehalem)
> test needed
> 

If this is not a problem, I'd like to kindly ask You to review those
patches.

Rafael has expressed his opinion about "boost" attribute clearly at the
other mail (as a response to v4 patches):
"Simple: Export it only when supported."

Despite the change about the "boost" attribute visibility there were
many other changes, which deserve maintainer's look.



> 
> Lukasz Majewski (7):
>   cpufreq: Store cpufreq policies in a list
>   cpufreq: Add boost frequency support in core
>   cpufreq:acpi:x86: Adjust the acpi-cpufreq.c code to work with common
>     boost solution
>   cpufreq:exynos:Extend Exynos cpufreq driver to support boost
>     framework
>   cpufreq: Calculate number of busy CPUs
>   cpufreq: Enable software boost only when up to one busy core is
>     running
>   thermal:boost: Disable boost when trip point is reached
> 
>  drivers/cpufreq/acpi-cpufreq.c     |   61 ++++++-----------
>  drivers/cpufreq/cpufreq.c          |  132
> ++++++++++++++++++++++++++++++++++++
> drivers/cpufreq/cpufreq_governor.c |    4 ++
> drivers/cpufreq/exynos-cpufreq.c   |    4 ++
> drivers/cpufreq/freq_table.c       |   43 ++++++++++--
> drivers/thermal/cpu_cooling.c      |    3 +
> include/linux/cpufreq.h            |   15 ++++ 7 files changed, 215
> insertions(+), 47 deletions(-)
> 



-- 
Best regards,

Lukasz Majewski

Samsung R&D Institute Poland (SRPOL) | Linux Platform Group
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/