RE: [PATCH 2/2] arm64: dts: imx8mm: Assign highest opp as suspend opp

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi, Viresh

> On 04-07-19, 07:49, Leonard Crestez wrote:
> > On 7/4/2019 9:23 AM, Anson.Huang@xxxxxxx wrote:
> > > From: Anson Huang <Anson.Huang@xxxxxxx>
> > >
> > > Assign highest OPP as suspend OPP to reduce suspend/resume latency
> > > on i.MX8MM.
> > >
> > > Signed-off-by: Anson Huang <Anson.Huang@xxxxxxx>
> > > ---
> > >   arch/arm64/boot/dts/freescale/imx8mm.dtsi | 1 +
> > >   1 file changed, 1 insertion(+)
> > >
> > > diff --git a/arch/arm64/boot/dts/freescale/imx8mm.dtsi
> > > b/arch/arm64/boot/dts/freescale/imx8mm.dtsi
> > > index b11fc5e..3a62407 100644
> > > --- a/arch/arm64/boot/dts/freescale/imx8mm.dtsi
> > > +++ b/arch/arm64/boot/dts/freescale/imx8mm.dtsi
> > > @@ -136,6 +136,7 @@
> > >   			opp-microvolt = <1000000>;
> > >   			opp-supported-hw = <0x8>, <0x3>;
> > >   			clock-latency-ns = <150000>;
> > > +			opp-suspend;
> > >   		};
> > >   	};
> >
> > What if the highest OPP is unavailable due to speed grading?
> 
> What does this exactly mean ? How is the OPP made unavailable in your
> case ?

That is because in i.MX8M series SoCs, the speed grading and market segment
fuses settings could affect the OPP defined in DT, in a word, all possible OPPs
are defined in DT, but each parts could ONLY select some of them to be working
OPPs, so if the "opp-suspend" is added for 1 OPP in DT, if the part's speed grading or
market segment fuse settings make that OPP as unavailable,  then that "opp-suspend"
is NOT working at all.

> 
> What will dev_pm_opp_get_suspend_opp_freq() return in this case ?

If the OPP contains "opp-suspend" property is NOT supported by the HW, then there will
be no suspend OPP defined, so it will return 0. The _opp_is_supported() parses the opp-supported-hw
before opp-suspend.

> 
> > Ideally we
> > should find a way to suspend at the highest *supported* OPP.
> >
> > Maybe the opp-suspend marking could be assigned from imx-cpufreq-dt
> > driver code?

I ever tried that, go through the OPP table and check the fuse settings, then runtime add "opp-suspend"
to the opp table, but unfortunately, the " struct opp_table " is NOT opened to be used, it is a private
structure?

> 
> Sorry for jumping in late, the latest patch from Anson drew my attention to
> this topic :)

That is OK😊

Thanks,
Anson.





[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux