Re: [PATCH V2 1/4] cpufreq: add arm soc generic cpufreq driver

[Richard Zhao <richard.zhao@linaro.org>]

On Fri, Dec 16, 2011 at 11:59:02AM -0800, Bryan Huntsman wrote:
> On 12/16/2011 02:52 AM, Jamie Iles wrote:
> >
> >> +static DEFINE_PER_CPU(unsigned long, l_p_j_ref);
> >> +static unsigned long l_p_j_ref_freq;
> >> +
> >> +static struct clk *cpu_clk;
> > 
> > This assumes that all CPU's share the same clk and run at the same rate.  
> > Is that a fair/safe assumption?  I honestly don't know the answer to 
> > this so it's just a question!!!
> 
> On MSM, cpus independently scale both frequency and voltage.  Our clock
> driver isn't upstream yet.  David Brown has a preliminary version here:
If
 - cpu_clk are per_cpu, and get from cpu0 clk, cpu1 clk, etc.
   The same to regulators.
 - set CPUFREQ_SHARED_TYPE_NONE
 - don't set affect cpus.

Does that help you?

Thanks
Richard
> 
> https://www.codeaurora.org/gitweb/quic/kernel/?p=davidb/linux-msm.git;a=shortlog;h=refs/heads/msm-clock-rfc
> 
> Once we get our driver upstream, MSM will be an exception and not select
> ARM_GENERIC_CPUFREQ.  We'll probably have a separate msm-cpufreq.c
> driver under drivers/cpufreq/.
> 
> - Bryan
> 
> -- 
> Sent by an employee of the Qualcomm Innovation Center, Inc.
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.