[PATCH v2 0/6] clk: renesas: r8a779[56]: Add Z and Z2 clock support

[Simon Horman]

On Wed, Aug 23, 2017 at 05:20:55PM +0200, Geert Uytterhoeven wrote:
> Hi Simon,
> 
> On Mon, Aug 21, 2017 at 3:01 PM, Simon Horman
> <horms+renesas at verge.net.au> wrote:
> > this patch-set adds Z and Z2 clock support.
> >
> > These are dependencies for supporting CPUFreq. The remainder of that
> > work is being posted separately and can be found at:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas.git topic/rcar-gen3-cpufreq
> >
> > A description of steps taken to lightly exercise the same feature for the
> > r88a7795 the above can be found at the link below. The results are the same
> > for the r8a7796 with the exception that it has two active CPU cores rather
> > than four.
> >
> > http://elinux.org/Tests:R-CAR-GEN3-CPUFreq
> 
> Thanks for your patches, and the wiki page!
> 
> I gave it a try on R-Car H3 (ES1.0 and ES2.0) and M3-W (ES1.0), and the
> off-by-two factor of the Z clock frequency is gone.
> I couldn't test on R-Car H3 ES1.1. Probably it's OK, too (ES1.1 fixed the
> missing PLL0/2/4 post-divider in ES1.0).
> 
> 1. After boot-up, the CPU clock frequency is 1.5 GHz, and
>    /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq and
>    /sys/kernel/debug/clk/z/clk_rate agree. Good.
> 
> 2. After switching to the conservative governor, scaling_cur_freq reports
>    either a 500 MHz or 1 GHz clock rate.
>    But /sys/kernel/debug/clk/z/clk_rate disagrees: it reports either a 200
>    or 700 MHz clock rate.
> 
>    Ah, there's also cpuinfo_cur_freq. That value matches the Z clock.
>    Interestingly, cpuinfo_cur_freq is lower than cpuinfo_min_freq?
>    /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq:200000
>    /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq:1500000
>    /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_min_freq:500000
> 
> 3. After switching back to the performance governor, scaling_cur_freq reports
>    1.5 GHz again.
>    But cpuinfo_cur_freq is still only 700 MHz, just like z/clk_rate.
> 
> Do you know what's wrong?

Unfortunately not. I will look into this.