Message ID | 20220415055916.28350-1-rex-bc.chen@mediatek.com |
---|---|
Headers | show |
Series | cpufreq: mediatek: Cleanup and support MT8183 and MT8186 | expand |
On Fri, Apr 15, 2022 at 1:59 PM Rex-BC Chen <rex-bc.chen@mediatek.com> wrote: > > Add MediaTek CCI devfreq node for MT8183. > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > --- > arch/arm64/boot/dts/mediatek/mt8183-evb.dts | 4 ++++ > arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 ++++ > arch/arm64/boot/dts/mediatek/mt8183.dtsi | 7 +++++++ > 3 files changed, 15 insertions(+) > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > index 8953dbf84f3e..7ac9864db9de 100644 > --- a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > +++ b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > @@ -412,6 +412,10 @@ > > }; > > +&cci { > + proc-supply = <&mt6358_vproc12_reg>; > +}; > + > &cpu0 { > proc-supply = <&mt6358_vproc12_reg>; > }; > diff --git a/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi b/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi > index 0f9480f91261..4786a32ee975 100644 > --- a/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi > +++ b/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi > @@ -230,6 +230,10 @@ > status = "okay"; > }; > > +&cci { > + proc-supply = <&mt6358_vproc12_reg>; > +}; > + > &cpu0 { > proc-supply = <&mt6358_vproc12_reg>; > }; > diff --git a/arch/arm64/boot/dts/mediatek/mt8183.dtsi b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > index 4ae3305d16d2..334728413582 100644 > --- a/arch/arm64/boot/dts/mediatek/mt8183.dtsi > +++ b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > @@ -280,6 +280,13 @@ > }; > }; > > + cci: cci { > + compatible = "mediatek,mt8183-cci"; > + clocks = <&apmixedsys CLK_APMIXED_CCIPLL>; > + clock-names = "cci_clock"; > + operating-points-v2 = <&cci_opp>; hi Rex, cci_opp is not defined in dts. > + }; > + > cpus { > #address-cells = <1>; > #size-cells = <0>; > -- > 2.18.0 >
On Fri, Apr 15, 2022 at 2:06 PM Hsin-Yi Wang <hsinyi@google.com> wrote: > > On Fri, Apr 15, 2022 at 1:59 PM Rex-BC Chen <rex-bc.chen@mediatek.com> wrote: > > > > Add MediaTek CCI devfreq node for MT8183. > > > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > > --- > > arch/arm64/boot/dts/mediatek/mt8183-evb.dts | 4 ++++ > > arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 ++++ > > arch/arm64/boot/dts/mediatek/mt8183.dtsi | 7 +++++++ > > 3 files changed, 15 insertions(+) > > > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > index 8953dbf84f3e..7ac9864db9de 100644 > > --- a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > +++ b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > @@ -412,6 +412,10 @@ > > > > }; > > > > +&cci { > > + proc-supply = <&mt6358_vproc12_reg>; > > +}; > > + > > &cpu0 { > > proc-supply = <&mt6358_vproc12_reg>; > > }; > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi b/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi > > index 0f9480f91261..4786a32ee975 100644 > > --- a/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi > > +++ b/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi > > @@ -230,6 +230,10 @@ > > status = "okay"; > > }; > > > > +&cci { > > + proc-supply = <&mt6358_vproc12_reg>; > > +}; > > + > > &cpu0 { > > proc-supply = <&mt6358_vproc12_reg>; > > }; > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183.dtsi b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > index 4ae3305d16d2..334728413582 100644 > > --- a/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > +++ b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > @@ -280,6 +280,13 @@ > > }; > > }; > > > > + cci: cci { > > + compatible = "mediatek,mt8183-cci"; > > + clocks = <&apmixedsys CLK_APMIXED_CCIPLL>; > > + clock-names = "cci_clock"; > > + operating-points-v2 = <&cci_opp>; > > hi Rex, > > cci_opp is not defined in dts. > It's in the previous patch. Please ignore this comment. > > + }; > > + > > cpus { > > #address-cells = <1>; > > #size-cells = <0>; > > -- > > 2.18.0 > >
On Fri, Apr 15, 2022 at 1:59 PM Rex-BC Chen <rex-bc.chen@mediatek.com> wrote: > > To prevent infinite loop when tracking voltage, we calculate the maximum > value for each platform data. > We assume min voltage is 0 and tracking target voltage using > min_volt_shift for each iteration. > The retry_max is 3 times of expeted iteration count. > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > --- > drivers/cpufreq/mediatek-cpufreq.c | 16 ++++++++++++++++ > 1 file changed, 16 insertions(+) > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c b/drivers/cpufreq/mediatek-cpufreq.c > index cc44a7a9427a..d4c00237e862 100644 > --- a/drivers/cpufreq/mediatek-cpufreq.c > +++ b/drivers/cpufreq/mediatek-cpufreq.c > @@ -86,6 +86,16 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, > struct regulator *proc_reg = info->proc_reg; > struct regulator *sram_reg = info->sram_reg; > int pre_vproc, pre_vsram, new_vsram, vsram, vproc, ret; > + int retry_max; > + > + /* > + * We assume min voltage is 0 and tracking target voltage using > + * min_volt_shift for each iteration. > + * The retry_max is 3 times of expeted iteration count. > + */ > + retry_max = 3 * DIV_ROUND_UP(max(info->soc_data->sram_max_volt, > + info->soc_data->proc_max_volt), > + info->soc_data->min_volt_shift); mtk_cpufreq_voltage_tracking() will be called very frequently. retry_max is the same every time mtk_cpufreq_voltage_tracking() is called. Is it better to calculate before and store in mtk_cpu_dvfs_info? > > pre_vproc = regulator_get_voltage(proc_reg); > if (pre_vproc < 0) { > @@ -151,6 +161,12 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, > > pre_vproc = vproc; > pre_vsram = vsram; > + > + if (--retry_max < 0) { > + dev_err(info->cpu_dev, > + "over loop count, failed to set voltage\n"); > + return -EINVAL; > + } > } while (vproc != new_vproc || vsram != new_vsram); > > return 0; > -- > 2.18.0 >
On Fri, 2022-04-15 at 14:14 +0800, Hsin-Yi Wang wrote: > On Fri, Apr 15, 2022 at 1:59 PM Rex-BC Chen <rex-bc.chen@mediatek.com > > wrote: > > > > To prevent infinite loop when tracking voltage, we calculate the > > maximum > > value for each platform data. > > We assume min voltage is 0 and tracking target voltage using > > min_volt_shift for each iteration. > > The retry_max is 3 times of expeted iteration count. > > > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > > --- > > drivers/cpufreq/mediatek-cpufreq.c | 16 ++++++++++++++++ > > 1 file changed, 16 insertions(+) > > > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c > > b/drivers/cpufreq/mediatek-cpufreq.c > > index cc44a7a9427a..d4c00237e862 100644 > > --- a/drivers/cpufreq/mediatek-cpufreq.c > > +++ b/drivers/cpufreq/mediatek-cpufreq.c > > @@ -86,6 +86,16 @@ static int mtk_cpufreq_voltage_tracking(struct > > mtk_cpu_dvfs_info *info, > > struct regulator *proc_reg = info->proc_reg; > > struct regulator *sram_reg = info->sram_reg; > > int pre_vproc, pre_vsram, new_vsram, vsram, vproc, ret; > > + int retry_max; > > + > > + /* > > + * We assume min voltage is 0 and tracking target voltage > > using > > + * min_volt_shift for each iteration. > > + * The retry_max is 3 times of expeted iteration count. > > + */ > > + retry_max = 3 * DIV_ROUND_UP(max(info->soc_data- > > >sram_max_volt, > > + info->soc_data- > > >proc_max_volt), > > + info->soc_data- > > >min_volt_shift); > > mtk_cpufreq_voltage_tracking() will be called very frequently. > retry_max is the same every time mtk_cpufreq_voltage_tracking() is > called. Is it better to calculate before and store in > mtk_cpu_dvfs_info? > Hello Hsin-Yi, Thanks for your reviwew. I will do this in next version. BRs, Rex > > > > pre_vproc = regulator_get_voltage(proc_reg); > > if (pre_vproc < 0) { > > @@ -151,6 +161,12 @@ static int mtk_cpufreq_voltage_tracking(struct > > mtk_cpu_dvfs_info *info, > > > > pre_vproc = vproc; > > pre_vsram = vsram; > > + > > + if (--retry_max < 0) { > > + dev_err(info->cpu_dev, > > + "over loop count, failed to set > > voltage\n"); > > + return -EINVAL; > > + } > > } while (vproc != new_vproc || vsram != new_vsram); > > > > return 0; > > -- > > 2.18.0 > >
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > Add mediatek,cci property to support MediaTek CCI feature. > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > Add MediaTek CCI devfreq node for MT8183. > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > - Add cpufreq opp table. > - Add MediaTek cci opp table. > - Add property of opp table and clock fro cpufreq. > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > --- > arch/arm64/boot/dts/mediatek/mt8183-evb.dts | 32 +++ > arch/arm64/boot/dts/mediatek/mt8183.dtsi | 270 ++++++++++++++++++++ > 2 files changed, 302 insertions(+) > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > index f3fd3cca23e9..8953dbf84f3e 100644 > --- a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > +++ b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts ..snip.. > diff --git a/arch/arm64/boot/dts/mediatek/mt8183.dtsi b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > index 4b08691ed39e..4ae3305d16d2 100644 > --- a/arch/arm64/boot/dts/mediatek/mt8183.dtsi > +++ b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > @@ -42,6 +42,244 @@ > rdma1 = &rdma1; > }; > > + cluster0_opp: cluster_opp_table0 { Please don't use underscores in devicetree for anything that's not a phandle. cluster0_opp: cluster0-opp-table { > + compatible = "operating-points-v2"; > + opp-shared; > + opp0_00 { Apart from the underscore being invalid, across the kernel, you can find many instances of opp tables, containing names related to the frequencies, so, for consistency (this is not a rule!), perhaps it would be a good idea to do so. opp-793000000 { > + opp-hz = /bits/ 64 <793000000>; > + opp-microvolt = <650000>; > + required-opps = <&opp2_00>; > + }; > + opp0_01 { opp-910000000 { > + opp-hz = /bits/ 64 <910000000>; > + opp-microvolt = <687500>; > + required-opps = <&opp2_01>; > + }; > + opp0_02 { ..snip.. > + opp0_15 { > + opp-hz = /bits/ 64 <1989000000>; > + opp-microvolt = <1050000>; > + required-opps = <&opp2_15>; P.S.: Please fix the typo below! > + }; }; > + > + ...also... > + cci_opp: opp_table2 { cci_opp: cci-opp-table { > + compatible = "operating-points-v2"; > + opp-shared; > + opp2_00: opp-273000000 { This one is fine instead :)) Cheers, Angelo
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > From: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > In some MediaTek SoCs, like MT8183, CPU and CCI share the same power > supplies. Cpufreq needs to check if CCI devfreq exists and wait until > CCI devfreq ready before scaling frequency. > > Before CCI devfreq is ready, we record the voltage when booting to > kernel and use the max(cpu target voltage, booting voltage) to > prevent cpufreq adjust to the lower voltage which will cause the CCI > crash because of high frequency and low voltage. > > - Add is_ccifreq_ready() to link CCI device to CPI, and CPU will start > DVFS when CCI is ready. > - Add platform data for MT8183. > > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> I am enthusiast to see that the solution that I've proposed was welcome! I only have one nit on this patch, check below: > --- > drivers/cpufreq/mediatek-cpufreq.c | 80 +++++++++++++++++++++++++++++- > 1 file changed, 79 insertions(+), 1 deletion(-) > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c b/drivers/cpufreq/mediatek-cpufreq.c > index d4c00237e862..dd3f739fede1 100644 > --- a/drivers/cpufreq/mediatek-cpufreq.c > +++ b/drivers/cpufreq/mediatek-cpufreq.c ..snip.. > @@ -225,6 +251,14 @@ static int mtk_cpufreq_set_target(struct cpufreq_policy *policy, > vproc = dev_pm_opp_get_voltage(opp); > dev_pm_opp_put(opp); > > + /* > + * If MediaTek cci is supported but is not ready, we will use the value > + * of max(target cpu voltage, booting voltage) to prevent high freqeuncy > + * low voltage crash. > + */ > + if (info->soc_data->ccifreq_supported && !is_ccifreq_ready(info)) > + vproc = max(vproc, info->vproc_on_boot); > + > /* > * If the new voltage or the intermediate voltage is higher than the > * current voltage, scale up voltage first. ..snip.. > @@ -423,6 +484,13 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu) > if (ret) > goto out_disable_mux_clock; > > + info->vproc_on_boot = regulator_get_voltage(info->proc_reg); This result is used only if we use ccifreq, so this should be enclosed in an if condition: this will spare us some (yes, small) time on devices that don't use it. if (info->soc_data->ccifreq_supported) { info->vproc_on_boot = regulator_get_voltage(info->proc_reg); if (info->vproc_on_boot < 0) { dev_err(.... goto .. } } P.S.: While at it, since the maximum width is 100 columns, the dev_err() call fits, so don't break that line! After the requested change: Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Il 15/04/22 08:14, Hsin-Yi Wang ha scritto: > On Fri, Apr 15, 2022 at 1:59 PM Rex-BC Chen <rex-bc.chen@mediatek.com> wrote: >> >> To prevent infinite loop when tracking voltage, we calculate the maximum >> value for each platform data. >> We assume min voltage is 0 and tracking target voltage using >> min_volt_shift for each iteration. >> The retry_max is 3 times of expeted iteration count. >> >> Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> I'm sorry Rex, but this commit has to be squashed with 09/15, as the logic is that each commit has to be acceptable, and 09/15 is not, without this fix. Besides, as Hsin-Yi suggested, calculating this every time may hit performance, but at the same time I don't want to lose this explicit calculation... >> --- >> drivers/cpufreq/mediatek-cpufreq.c | 16 ++++++++++++++++ >> 1 file changed, 16 insertions(+) >> >> diff --git a/drivers/cpufreq/mediatek-cpufreq.c b/drivers/cpufreq/mediatek-cpufreq.c >> index cc44a7a9427a..d4c00237e862 100644 >> --- a/drivers/cpufreq/mediatek-cpufreq.c >> +++ b/drivers/cpufreq/mediatek-cpufreq.c >> @@ -86,6 +86,16 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, >> struct regulator *proc_reg = info->proc_reg; >> struct regulator *sram_reg = info->sram_reg; >> int pre_vproc, pre_vsram, new_vsram, vsram, vproc, ret; >> + int retry_max; >> + >> + /* >> + * We assume min voltage is 0 and tracking target voltage using >> + * min_volt_shift for each iteration. >> + * The retry_max is 3 times of expeted iteration count. >> + */ >> + retry_max = 3 * DIV_ROUND_UP(max(info->soc_data->sram_max_volt, >> + info->soc_data->proc_max_volt), >> + info->soc_data->min_volt_shift); > > mtk_cpufreq_voltage_tracking() will be called very frequently. > retry_max is the same every time mtk_cpufreq_voltage_tracking() is > called. Is it better to calculate before and store in > mtk_cpu_dvfs_info? > ...so I agree with this solution: perhaps you can add a "vtrack_max" variable to mtk_cpu_dvfs_info as suggested, and fill in that one in function mtk_cpu_dvfs_info_init(), where we effectively initialize all-the-things. Cheers, Angelo
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > From: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > Voltages and shifts are defined as macros originally. > There are different requirements of these values for each MediaTek SoCs. > Therefore, we add the platform data and move these values into it. > > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > From: "Andrew-sh.Cheng" <andrew-sh.cheng@mediatek.com> > > From this opp notifier, cpufreq should listen to opp notification and do > proper actions when receiving events of disable and voltage adjustment. > > One of the user for this opp notifier is MediaTek SVS. > The MediaTek Smart Voltage Scaling (SVS) is a hardware which calculates > suitable SVS bank voltages to OPP voltage table. > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > --- > drivers/cpufreq/mediatek-cpufreq.c | 93 +++++++++++++++++++++++++++--- > 1 file changed, 85 insertions(+), 8 deletions(-) > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c b/drivers/cpufreq/mediatek-cpufreq.c > index fa8b193bf27b..221f249f8d21 100644 > --- a/drivers/cpufreq/mediatek-cpufreq.c > +++ b/drivers/cpufreq/mediatek-cpufreq.c > @@ -41,6 +41,11 @@ struct mtk_cpu_dvfs_info { > int intermediate_voltage; > bool need_voltage_tracking; > int pre_vproc; > + /* Avoid race condition for regulators between notify and policy */ > + struct mutex reg_lock; > + struct notifier_block opp_nb; > + int opp_cpu; This should be unsigned int because: 1. A negative CPU number is impossible; 2. The only usage is as a parameter of cpufreq_cpu_get(unsigned int cpu). Please change this to unsigned int, after which: Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > From: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > We found the buck voltage may not be exactly the same with what we set > because CPU may share the same buck with other module. > Therefore, we need to record the previous desired value instead of reading > it from regulators. > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > --- > drivers/cpufreq/mediatek-cpufreq.c | 17 +++++++++++++---- > 1 file changed, 13 insertions(+), 4 deletions(-) > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c b/drivers/cpufreq/mediatek-cpufreq.c > index ff27f77e8ee6..fa8b193bf27b 100644 > --- a/drivers/cpufreq/mediatek-cpufreq.c > +++ b/drivers/cpufreq/mediatek-cpufreq.c > @@ -40,6 +40,7 @@ struct mtk_cpu_dvfs_info { > struct list_head list_head; > int intermediate_voltage; > bool need_voltage_tracking; > + int pre_vproc; > }; > > static LIST_HEAD(dvfs_info_list); > @@ -191,11 +192,17 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, > > static int mtk_cpufreq_set_voltage(struct mtk_cpu_dvfs_info *info, int vproc) > { > + int ret; > + > if (info->need_voltage_tracking) > - return mtk_cpufreq_voltage_tracking(info, vproc); > + ret = mtk_cpufreq_voltage_tracking(info, vproc); > else > - return regulator_set_voltage(info->proc_reg, vproc, > - vproc + VOLT_TOL); > + ret = regulator_set_voltage(info->proc_reg, vproc, > + MAX_VOLT_LIMIT); > + if (!ret) > + info->pre_vproc = vproc; > + > + return ret; > } > > static int mtk_cpufreq_set_target(struct cpufreq_policy *policy, > @@ -213,7 +220,9 @@ static int mtk_cpufreq_set_target(struct cpufreq_policy *policy, > inter_vproc = info->intermediate_voltage; > > pre_freq_hz = clk_get_rate(cpu_clk); > - pre_vproc = regulator_get_voltage(info->proc_reg); > + pre_vproc = info->pre_vproc; > + if (pre_vproc <= 0) > + pre_vproc = regulator_get_voltage(info->proc_reg); I would do it like that, instead: if (unlikely(info->pre_vproc <= 0)) pre_vproc = regulator_get_voltage(info->proc_reg); else pre_vproc = info->pre_vproc; ....as even though it is indeed possible that info->pre_vproc is <= 0, it is very unlikely to happen ;-) This also solves a 'pre_vproc' double assignment issue, by the way. Cheers, Angelo
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > To make driver more readable, replace old_* with pre_*. > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com
Il 15/04/22 07:59, Rex-BC Chen ha scritto: > - Replace pr_* with dev_* to show logs. > - Remove usage of __func__. > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com
On Fri, 2022-04-15 at 14:24 +0200, AngeloGioacchino Del Regno wrote: > Il 15/04/22 07:59, Rex-BC Chen ha scritto: > > From: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > > > We found the buck voltage may not be exactly the same with what we > > set > > because CPU may share the same buck with other module. > > Therefore, we need to record the previous desired value instead of > > reading > > it from regulators. > > > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > > --- > > drivers/cpufreq/mediatek-cpufreq.c | 17 +++++++++++++---- > > 1 file changed, 13 insertions(+), 4 deletions(-) > > > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c > > b/drivers/cpufreq/mediatek-cpufreq.c > > index ff27f77e8ee6..fa8b193bf27b 100644 > > --- a/drivers/cpufreq/mediatek-cpufreq.c > > +++ b/drivers/cpufreq/mediatek-cpufreq.c > > @@ -40,6 +40,7 @@ struct mtk_cpu_dvfs_info { > > struct list_head list_head; > > int intermediate_voltage; > > bool need_voltage_tracking; > > + int pre_vproc; > > }; > > > > static LIST_HEAD(dvfs_info_list); > > @@ -191,11 +192,17 @@ static int > > mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, > > > > static int mtk_cpufreq_set_voltage(struct mtk_cpu_dvfs_info > > *info, int vproc) > > { > > + int ret; > > + > > if (info->need_voltage_tracking) > > - return mtk_cpufreq_voltage_tracking(info, vproc); > > + ret = mtk_cpufreq_voltage_tracking(info, vproc); > > else > > - return regulator_set_voltage(info->proc_reg, vproc, > > - vproc + VOLT_TOL); > > + ret = regulator_set_voltage(info->proc_reg, vproc, > > + MAX_VOLT_LIMIT); > > + if (!ret) > > + info->pre_vproc = vproc; > > + > > + return ret; > > } > > > > static int mtk_cpufreq_set_target(struct cpufreq_policy *policy, > > @@ -213,7 +220,9 @@ static int mtk_cpufreq_set_target(struct > > cpufreq_policy *policy, > > inter_vproc = info->intermediate_voltage; > > > > pre_freq_hz = clk_get_rate(cpu_clk); > > - pre_vproc = regulator_get_voltage(info->proc_reg); > > + pre_vproc = info->pre_vproc; > > + if (pre_vproc <= 0) > > + pre_vproc = regulator_get_voltage(info->proc_reg); > > I would do it like that, instead: > > if (unlikely(info->pre_vproc <= 0)) > pre_vproc = regulator_get_voltage(info->proc_reg); > else > pre_vproc = info->pre_vproc; > > ....as even though it is indeed possible that info->pre_vproc is <= > 0, it is > very unlikely to happen ;-) > This also solves a 'pre_vproc' double assignment issue, by the way. > > Cheers, > Angelo > > > Hello Angelo, OK, I will add this in next version. Thanks for your suggestion. BRs, Rex
On Fri, 2022-04-15 at 14:24 +0200, AngeloGioacchino Del Regno wrote: > Il 15/04/22 08:14, Hsin-Yi Wang ha scritto: > > On Fri, Apr 15, 2022 at 1:59 PM Rex-BC Chen < > > rex-bc.chen@mediatek.com> wrote: > > > > > > To prevent infinite loop when tracking voltage, we calculate the > > > maximum > > > value for each platform data. > > > We assume min voltage is 0 and tracking target voltage using > > > min_volt_shift for each iteration. > > > The retry_max is 3 times of expeted iteration count. > > > > > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > > I'm sorry Rex, but this commit has to be squashed with 09/15, as the > logic is > that each commit has to be acceptable, and 09/15 is not, without this > fix. > > Besides, as Hsin-Yi suggested, calculating this every time may hit > performance, > but at the same time I don't want to lose this explicit > calculation... > Hello Angelo, I will squash thius patch into 9/15 and will use info->vtrack_max to record the value in probe function. Thanks! BRs, Rex > > > --- > > > drivers/cpufreq/mediatek-cpufreq.c | 16 ++++++++++++++++ > > > 1 file changed, 16 insertions(+) > > > > > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c > > > b/drivers/cpufreq/mediatek-cpufreq.c > > > index cc44a7a9427a..d4c00237e862 100644 > > > --- a/drivers/cpufreq/mediatek-cpufreq.c > > > +++ b/drivers/cpufreq/mediatek-cpufreq.c > > > @@ -86,6 +86,16 @@ static int mtk_cpufreq_voltage_tracking(struct > > > mtk_cpu_dvfs_info *info, > > > struct regulator *proc_reg = info->proc_reg; > > > struct regulator *sram_reg = info->sram_reg; > > > int pre_vproc, pre_vsram, new_vsram, vsram, vproc, ret; > > > + int retry_max; > > > + > > > + /* > > > + * We assume min voltage is 0 and tracking target voltage > > > using > > > + * min_volt_shift for each iteration. > > > + * The retry_max is 3 times of expeted iteration count. > > > + */ > > > + retry_max = 3 * DIV_ROUND_UP(max(info->soc_data- > > > >sram_max_volt, > > > + info->soc_data- > > > >proc_max_volt), > > > + info->soc_data- > > > >min_volt_shift); > > > > mtk_cpufreq_voltage_tracking() will be called very frequently. > > retry_max is the same every time mtk_cpufreq_voltage_tracking() is > > called. Is it better to calculate before and store in > > mtk_cpu_dvfs_info? > > > > ...so I agree with this solution: perhaps you can add a "vtrack_max" > variable to > mtk_cpu_dvfs_info as suggested, and fill in that one in function > mtk_cpu_dvfs_info_init(), where we effectively initialize all-the- > things. > > Cheers, > Angelo
On Fri, 2022-04-15 at 14:24 +0200, AngeloGioacchino Del Regno wrote: > Il 15/04/22 07:59, Rex-BC Chen ha scritto: > > From: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > > > In some MediaTek SoCs, like MT8183, CPU and CCI share the same > > power > > supplies. Cpufreq needs to check if CCI devfreq exists and wait > > until > > CCI devfreq ready before scaling frequency. > > > > Before CCI devfreq is ready, we record the voltage when booting to > > kernel and use the max(cpu target voltage, booting voltage) to > > prevent cpufreq adjust to the lower voltage which will cause the > > CCI > > crash because of high frequency and low voltage. > > > > - Add is_ccifreq_ready() to link CCI device to CPI, and CPU will > > start > > DVFS when CCI is ready. > > - Add platform data for MT8183. > > > > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > > I am enthusiast to see that the solution that I've proposed was > welcome! > > I only have one nit on this patch, check below: > Hello Angelo, Thanks for your advice for this modification. I also reply to Kevin and describe this solution. Let's wait for Kevin's feedback and other suggestion. > > --- > > drivers/cpufreq/mediatek-cpufreq.c | 80 > > +++++++++++++++++++++++++++++- > > 1 file changed, 79 insertions(+), 1 deletion(-) > > > > diff --git a/drivers/cpufreq/mediatek-cpufreq.c > > b/drivers/cpufreq/mediatek-cpufreq.c > > index d4c00237e862..dd3f739fede1 100644 > > --- a/drivers/cpufreq/mediatek-cpufreq.c > > +++ b/drivers/cpufreq/mediatek-cpufreq.c > > ..snip.. > > > @@ -225,6 +251,14 @@ static int mtk_cpufreq_set_target(struct > > cpufreq_policy *policy, > > vproc = dev_pm_opp_get_voltage(opp); > > dev_pm_opp_put(opp); > > > > + /* > > + * If MediaTek cci is supported but is not ready, we will use > > the value > > + * of max(target cpu voltage, booting voltage) to prevent high > > freqeuncy > > + * low voltage crash. > > + */ > > + if (info->soc_data->ccifreq_supported && > > !is_ccifreq_ready(info)) > > + vproc = max(vproc, info->vproc_on_boot); > > + > > /* > > * If the new voltage or the intermediate voltage is higher > > than the > > * current voltage, scale up voltage first. > > ..snip.. > > > @@ -423,6 +484,13 @@ static int mtk_cpu_dvfs_info_init(struct > > mtk_cpu_dvfs_info *info, int cpu) > > if (ret) > > goto out_disable_mux_clock; > > > > + info->vproc_on_boot = regulator_get_voltage(info->proc_reg); > > This result is used only if we use ccifreq, so this should be > enclosed in an if > condition: this will spare us some (yes, small) time on devices that > don't use it. > > if (info->soc_data->ccifreq_supported) { > info->vproc_on_boot = regulator_get_voltage(info- > >proc_reg); > if (info->vproc_on_boot < 0) { > dev_err(.... > goto .. > } > } > > P.S.: While at it, since the maximum width is 100 columns, the > dev_err() call fits, > so don't break that line! > > After the requested change: > > Reviewed-by: AngeloGioacchino Del Regno < > angelogioacchino.delregno@collabora.com> I will add this in next version if there is no any suggestion for this patch. Thanks! BRs, Rex
On Fri, 2022-04-15 at 14:24 +0200, AngeloGioacchino Del Regno wrote: > Il 15/04/22 07:59, Rex-BC Chen ha scritto: > > - Add cpufreq opp table. > > - Add MediaTek cci opp table. > > - Add property of opp table and clock fro cpufreq. > > > > Signed-off-by: Andrew-sh.Cheng <andrew-sh.cheng@mediatek.com> > > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> > > --- > > arch/arm64/boot/dts/mediatek/mt8183-evb.dts | 32 +++ > > arch/arm64/boot/dts/mediatek/mt8183.dtsi | 270 > > ++++++++++++++++++++ > > 2 files changed, 302 insertions(+) > > > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > index f3fd3cca23e9..8953dbf84f3e 100644 > > --- a/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > +++ b/arch/arm64/boot/dts/mediatek/mt8183-evb.dts > > ..snip.. > > > diff --git a/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > index 4b08691ed39e..4ae3305d16d2 100644 > > --- a/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > +++ b/arch/arm64/boot/dts/mediatek/mt8183.dtsi > > @@ -42,6 +42,244 @@ > > rdma1 = &rdma1; > > }; > > > > + cluster0_opp: cluster_opp_table0 { > > Please don't use underscores in devicetree for anything that's not a > phandle. > > cluster0_opp: cluster0-opp-table { > > > + compatible = "operating-points-v2"; > > + opp-shared; > > + opp0_00 { > > Apart from the underscore being invalid, across the kernel, you can > find many > instances of opp tables, containing names related to the frequencies, > so, for > consistency (this is not a rule!), perhaps it would be a good idea to > do so. > > opp-793000000 { > > > + opp-hz = /bits/ 64 <793000000>; > > + opp-microvolt = <650000>; > > + required-opps = <&opp2_00>; > > + }; > > + opp0_01 { > > opp-910000000 { > > > + opp-hz = /bits/ 64 <910000000>; > > + opp-microvolt = <687500>; > > + required-opps = <&opp2_01>; > > + }; > > + opp0_02 { > > ..snip.. > > > + opp0_15 { > > + opp-hz = /bits/ 64 <1989000000>; > > + opp-microvolt = <1050000>; > > + required-opps = <&opp2_15>; > > P.S.: Please fix the typo below! > > > + }; }; > > + > > + > > ...also... > > > + cci_opp: opp_table2 { > > cci_opp: cci-opp-table { > > > + compatible = "operating-points-v2"; > > + opp-shared; > > + opp2_00: opp-273000000 { > > This one is fine instead :)) > > Cheers, > Angelo Hello Angelo, OK, I will modify this issue in next version. Thanks! BRs, Rex
Hi Rex, Rex-BC Chen <rex-bc.chen@mediatek.com> writes: > From: Jia-Wei Chang <jia-wei.chang@mediatek.com> > > In some MediaTek SoCs, like MT8183, CPU and CCI share the same power > supplies. Cpufreq needs to check if CCI devfreq exists and wait until > CCI devfreq ready before scaling frequency. > > Before CCI devfreq is ready, we record the voltage when booting to > kernel and use the max(cpu target voltage, booting voltage) to > prevent cpufreq adjust to the lower voltage which will cause the CCI > crash because of high frequency and low voltage. > > - Add is_ccifreq_ready() to link CCI device to CPI, and CPU will start > DVFS when CCI is ready. > - Add platform data for MT8183. > > Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com> > Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> The solution of keeping the max of the CPU voltage from OPP and boot-up voltage makes sense until CCI is ready. Thank you for the rework and the detailed technical explanations. Reviewed-by: Kevin Hilman <khilman@baylibre.com> Kevin