Message ID | 20230809173905.1844132-3-a-nandan@ti.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | Add support for thermal mitigation for K3 J7200 SoC | expand |
On 23:09-20230809, Apurva Nandan wrote: > From: Keerthy <j-keerthy@ti.com> > > Add 750M, 1G, 1.5G & 2G as the supported frequencies for A72. > This enables support for Dynamic Frequency Scaling(DFS) > > Signed-off-by: Keerthy <j-keerthy@ti.com> > Signed-off-by: Apurva Nandan <a-nandan@ti.com> > --- > arch/arm64/boot/dts/ti/k3-j7200.dtsi | 28 ++++++++++++++++++++++++++++ > 1 file changed, 28 insertions(+) > > diff --git a/arch/arm64/boot/dts/ti/k3-j7200.dtsi b/arch/arm64/boot/dts/ti/k3-j7200.dtsi > index ef73e6d7e858..7222c453096f 100644 > --- a/arch/arm64/boot/dts/ti/k3-j7200.dtsi > +++ b/arch/arm64/boot/dts/ti/k3-j7200.dtsi > @@ -48,6 +48,10 @@ cpu0: cpu@0 { > d-cache-line-size = <64>; > d-cache-sets = <256>; > next-level-cache = <&L2_0>; > + clocks = <&k3_clks 202 2>; > + clock-names = "cpu"; > + operating-points-v2 = <&cpu0_opp_table>; > + #cooling-cells = <2>; /* min followed by max */ > }; > > cpu1: cpu@1 { > @@ -62,6 +66,30 @@ cpu1: cpu@1 { > d-cache-line-size = <64>; > d-cache-sets = <256>; > next-level-cache = <&L2_0>; > + clocks = <&k3_clks 203 0>; > + clock-names = "cpu"; > + operating-points-v2 = <&cpu0_opp_table>; > + #cooling-cells = <2>; /* min followed by max */ > + }; > + }; > + > + cpu0_opp_table: opp-table { > + compatible = "operating-points-v2"; > + > + opp4-2000000000 { > + opp-hz = /bits/ 64 <2000000000>; > + }; > + > + opp3-1500000000 { > + opp-hz = /bits/ 64 <1500000000>; > + }; > + > + opp2-1000000000 { > + opp-hz = /bits/ 64 <1000000000>; > + }; > + > + opp1-750000000 { > + opp-hz = /bits/ 64 <750000000>; > }; > }; > > -- > 2.34.1 > Are you sure this is correct to enable all OPPs without efuse bit checks? https://www.ti.com/lit/ds/symlink/dra821u-q1.pdf 7.5 Operating Performance Points DRA821xC operates only upto 750MHz DRA821xE at 1GHz DRA821xL upto 1.5GHz and DRA821xT upto 2GHz
On 8/10/2023 12:39 AM, Nishanth Menon wrote: > On 23:09-20230809, Apurva Nandan wrote: >> From: Keerthy <j-keerthy@ti.com> >> >> Add 750M, 1G, 1.5G & 2G as the supported frequencies for A72. >> This enables support for Dynamic Frequency Scaling(DFS) >> >> Signed-off-by: Keerthy <j-keerthy@ti.com> >> Signed-off-by: Apurva Nandan <a-nandan@ti.com> >> --- >> arch/arm64/boot/dts/ti/k3-j7200.dtsi | 28 ++++++++++++++++++++++++++++ >> 1 file changed, 28 insertions(+) >> >> diff --git a/arch/arm64/boot/dts/ti/k3-j7200.dtsi b/arch/arm64/boot/dts/ti/k3-j7200.dtsi >> index ef73e6d7e858..7222c453096f 100644 >> --- a/arch/arm64/boot/dts/ti/k3-j7200.dtsi >> +++ b/arch/arm64/boot/dts/ti/k3-j7200.dtsi >> @@ -48,6 +48,10 @@ cpu0: cpu@0 { >> d-cache-line-size = <64>; >> d-cache-sets = <256>; >> next-level-cache = <&L2_0>; >> + clocks = <&k3_clks 202 2>; >> + clock-names = "cpu"; >> + operating-points-v2 = <&cpu0_opp_table>; >> + #cooling-cells = <2>; /* min followed by max */ >> }; >> >> cpu1: cpu@1 { >> @@ -62,6 +66,30 @@ cpu1: cpu@1 { >> d-cache-line-size = <64>; >> d-cache-sets = <256>; >> next-level-cache = <&L2_0>; >> + clocks = <&k3_clks 203 0>; >> + clock-names = "cpu"; >> + operating-points-v2 = <&cpu0_opp_table>; >> + #cooling-cells = <2>; /* min followed by max */ >> + }; >> + }; >> + >> + cpu0_opp_table: opp-table { >> + compatible = "operating-points-v2"; >> + >> + opp4-2000000000 { >> + opp-hz = /bits/ 64 <2000000000>; >> + }; >> + >> + opp3-1500000000 { >> + opp-hz = /bits/ 64 <1500000000>; >> + }; >> + >> + opp2-1000000000 { >> + opp-hz = /bits/ 64 <1000000000>; >> + }; >> + >> + opp1-750000000 { >> + opp-hz = /bits/ 64 <750000000>; >> }; >> }; >> >> -- >> 2.34.1 >> > Are you sure this is correct to enable all OPPs without efuse bit checks? > > https://www.ti.com/lit/ds/symlink/dra821u-q1.pdf > 7.5 Operating Performance Points > DRA821xC operates only upto 750MHz > DRA821xE at 1GHz > DRA821xL upto 1.5GHz and > DRA821xT upto 2GHz Looks, top SKUs is considered here . After detecting which SKU we are running (I hope TRM should have this information- through efuse or some other register) I think, we can follow two approaches. 1) have OPP table for each SKU and select based SKUs type or 2) Do run time fixup by u-boot based upon SKU type
On 17:23-20230810, Kumar, Udit wrote: [..] > > > + opp1-750000000 { > > > + opp-hz = /bits/ 64 <750000000>; > > > }; > > > }; > > > -- > > > 2.34.1 > > > > > Are you sure this is correct to enable all OPPs without efuse bit checks? > > > > https://www.ti.com/lit/ds/symlink/dra821u-q1.pdf > > 7.5 Operating Performance Points > > DRA821xC operates only upto 750MHz > > DRA821xE at 1GHz > > DRA821xL upto 1.5GHz and > > DRA821xT upto 2GHz > > Looks, top SKUs is considered here . > > After detecting which SKU we are running (I hope TRM should have this > information- through efuse or some other register) > > I think, we can follow two approaches. Both of these are wrong approaches. > > 1) have OPP table for each SKU and select based SKUs type or This proliferates cpu dtsi to make it hard to manage > > 2) Do run time fixup by u-boot based upon SKU type This wont work: a) in u-boot's falcon boot mode and puts unrelated responsibility to bootloader (u-boot is not the only bootloader in the party here). b) Further, the reason for doing the opp detection in the kernel is due to the severity of consequence of attempting to run a lower rated chip at higher frequency - PoH (Power on Hours) or physical damage can result. c) Finally, in a virtualized environment: TISCI will get DM (Device Manager) to arbitrate between the each of the VM's request, but if the VM's are'nt self sufficient, we will have DM making wrong choices resulting in (b) condition again. This is the reason why drivers/cpufreq/ti-cpufreq.c exists and all SoCs that have OPPs from TI is handled in the kernel itself.
On 8/10/2023 6:23 PM, Nishanth Menon wrote: > On 17:23-20230810, Kumar, Udit wrote: > [..] >>>> + opp1-750000000 { >>>> + opp-hz = /bits/ 64 <750000000>; >>>> }; >>>> }; >>>> -- >>>> 2.34.1 >>>> >>> [..] >>> This wont work: >>> >>> a) in u-boot's falcon boot mode and puts unrelated responsibility to >>> bootloader (u-boot is not the only bootloader in the party here). >>> b) Further, the reason for doing the opp detection in the kernel is >>> due to the severity of consequence of attempting to run a lower rated >>> chip at higher frequency - PoH (Power on Hours) or physical damage can >>> result. >>> c) Finally, in a virtualized environment: TISCI will get DM (Device >>> Manager) to arbitrate between the each of the VM's request, but if >>> the VM's are'nt self sufficient, we will have DM making wrong choices >>> resulting in (b) condition again. >>> >>> This is the reason why drivers/cpufreq/ti-cpufreq.c exists and all SoCs >>> that have OPPs from TI is handled in the kernel itself. Thanks to pointing to this driver.
diff --git a/arch/arm64/boot/dts/ti/k3-j7200.dtsi b/arch/arm64/boot/dts/ti/k3-j7200.dtsi index ef73e6d7e858..7222c453096f 100644 --- a/arch/arm64/boot/dts/ti/k3-j7200.dtsi +++ b/arch/arm64/boot/dts/ti/k3-j7200.dtsi @@ -48,6 +48,10 @@ cpu0: cpu@0 { d-cache-line-size = <64>; d-cache-sets = <256>; next-level-cache = <&L2_0>; + clocks = <&k3_clks 202 2>; + clock-names = "cpu"; + operating-points-v2 = <&cpu0_opp_table>; + #cooling-cells = <2>; /* min followed by max */ }; cpu1: cpu@1 { @@ -62,6 +66,30 @@ cpu1: cpu@1 { d-cache-line-size = <64>; d-cache-sets = <256>; next-level-cache = <&L2_0>; + clocks = <&k3_clks 203 0>; + clock-names = "cpu"; + operating-points-v2 = <&cpu0_opp_table>; + #cooling-cells = <2>; /* min followed by max */ + }; + }; + + cpu0_opp_table: opp-table { + compatible = "operating-points-v2"; + + opp4-2000000000 { + opp-hz = /bits/ 64 <2000000000>; + }; + + opp3-1500000000 { + opp-hz = /bits/ 64 <1500000000>; + }; + + opp2-1000000000 { + opp-hz = /bits/ 64 <1000000000>; + }; + + opp1-750000000 { + opp-hz = /bits/ 64 <750000000>; }; };