* [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance @ 2013-11-05 3:10 rjying 2013-11-05 4:38 ` Viresh Kumar 0 siblings, 1 reply; 9+ messages in thread From: rjying @ 2013-11-05 3:10 UTC (permalink / raw) To: Rafael J. Wysocki, Viresh Kumar, cpufreq, linux-pm Cc: linux-kernel, rongjun.ying, Rongjun Ying From: Rongjun Ying <rongjun.ying@csr.com> Sometime the regulator can't supply appropriate voltages for requestion of cpu. All voltage tolerance value can't figure out a good voltage. Signed-off-by: Rongjun Ying <rongjun.ying@csr.com> --- drivers/cpufreq/cpufreq-cpu0.c | 17 ++++++++++------- 1 files changed, 10 insertions(+), 7 deletions(-) diff --git a/drivers/cpufreq/cpufreq-cpu0.c b/drivers/cpufreq/cpufreq-cpu0.c index c522a95..0c9e6b9 100644 --- a/drivers/cpufreq/cpufreq-cpu0.c +++ b/drivers/cpufreq/cpufreq-cpu0.c @@ -23,7 +23,7 @@ #include <linux/slab.h> static unsigned int transition_latency; -static unsigned int voltage_tolerance; /* in percentage */ +static unsigned int voltage_max; static struct device *cpu_dev; static struct clk *cpu_clk; @@ -45,7 +45,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, { struct cpufreq_freqs freqs; struct opp *opp; - unsigned long volt = 0, volt_old = 0, tol = 0; + unsigned long volt = 0, volt_old = 0; long freq_Hz, freq_exact; unsigned int index; int ret; @@ -82,7 +82,6 @@ static int cpu0_set_target(struct cpufreq_policy *policy, } volt = opp_get_voltage(opp); rcu_read_unlock(); - tol = volt * voltage_tolerance / 100; volt_old = regulator_get_voltage(cpu_reg); } @@ -92,7 +91,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, /* scaling up? scale voltage before frequency */ if (!IS_ERR(cpu_reg) && freqs.new > freqs.old) { - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); + ret = regulator_set_voltage(cpu_reg, volt, voltage_max); if (ret) { pr_err("failed to scale voltage up: %d\n", ret); freqs.new = freqs.old; @@ -104,14 +103,14 @@ static int cpu0_set_target(struct cpufreq_policy *policy, if (ret) { pr_err("failed to set clock rate: %d\n", ret); if (!IS_ERR(cpu_reg)) - regulator_set_voltage_tol(cpu_reg, volt_old, tol); + regulator_set_voltage(cpu_reg, volt_old, voltage_max); freqs.new = freqs.old; goto post_notify; } /* scaling down? scale voltage after frequency */ if (!IS_ERR(cpu_reg) && freqs.new < freqs.old) { - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); + ret = regulator_set_voltage(cpu_reg, volt, voltage_max); if (ret) { pr_err("failed to scale voltage down: %d\n", ret); clk_set_rate(cpu_clk, freqs.old * 1000); @@ -224,7 +223,11 @@ static int cpu0_cpufreq_probe(struct platform_device *pdev) goto out_put_node; } - of_property_read_u32(np, "voltage-tolerance", &voltage_tolerance); + ret = of_property_read_u32(np, "voltage-max", &voltage_max); + if (ret) { + pr_err("failed to get max voltage: %d\n", ret); + goto out_put_node; + } if (of_property_read_u32(np, "clock-latency", &transition_latency)) transition_latency = CPUFREQ_ETERNAL; -- 1.7.5.4 ^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-05 3:10 [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance rjying @ 2013-11-05 4:38 ` Viresh Kumar 2013-11-05 11:53 ` Shawn Guo 0 siblings, 1 reply; 9+ messages in thread From: Viresh Kumar @ 2013-11-05 4:38 UTC (permalink / raw) To: rjying, Shawn Guo Cc: Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, rongjun.ying, Rongjun Ying Adding shawn in cc list.. On 5 November 2013 08:40, rjying <rjying@gmail.com> wrote: > From: Rongjun Ying <rongjun.ying@csr.com> > > Sometime the regulator can't supply appropriate voltages for requestion of cpu. > All voltage tolerance value can't figure out a good voltage. > > Signed-off-by: Rongjun Ying <rongjun.ying@csr.com> > --- > drivers/cpufreq/cpufreq-cpu0.c | 17 ++++++++++------- > 1 files changed, 10 insertions(+), 7 deletions(-) > > diff --git a/drivers/cpufreq/cpufreq-cpu0.c b/drivers/cpufreq/cpufreq-cpu0.c > index c522a95..0c9e6b9 100644 > --- a/drivers/cpufreq/cpufreq-cpu0.c > +++ b/drivers/cpufreq/cpufreq-cpu0.c > @@ -23,7 +23,7 @@ > #include <linux/slab.h> > > static unsigned int transition_latency; > -static unsigned int voltage_tolerance; /* in percentage */ > +static unsigned int voltage_max; > > static struct device *cpu_dev; > static struct clk *cpu_clk; > @@ -45,7 +45,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > { > struct cpufreq_freqs freqs; > struct opp *opp; > - unsigned long volt = 0, volt_old = 0, tol = 0; > + unsigned long volt = 0, volt_old = 0; > long freq_Hz, freq_exact; > unsigned int index; > int ret; > @@ -82,7 +82,6 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > } > volt = opp_get_voltage(opp); > rcu_read_unlock(); > - tol = volt * voltage_tolerance / 100; > volt_old = regulator_get_voltage(cpu_reg); > } > > @@ -92,7 +91,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > > /* scaling up? scale voltage before frequency */ > if (!IS_ERR(cpu_reg) && freqs.new > freqs.old) { > - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); > + ret = regulator_set_voltage(cpu_reg, volt, voltage_max); > if (ret) { > pr_err("failed to scale voltage up: %d\n", ret); > freqs.new = freqs.old; > @@ -104,14 +103,14 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > if (ret) { > pr_err("failed to set clock rate: %d\n", ret); > if (!IS_ERR(cpu_reg)) > - regulator_set_voltage_tol(cpu_reg, volt_old, tol); > + regulator_set_voltage(cpu_reg, volt_old, voltage_max); > freqs.new = freqs.old; > goto post_notify; > } > > /* scaling down? scale voltage after frequency */ > if (!IS_ERR(cpu_reg) && freqs.new < freqs.old) { > - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); > + ret = regulator_set_voltage(cpu_reg, volt, voltage_max); > if (ret) { > pr_err("failed to scale voltage down: %d\n", ret); > clk_set_rate(cpu_clk, freqs.old * 1000); > @@ -224,7 +223,11 @@ static int cpu0_cpufreq_probe(struct platform_device *pdev) > goto out_put_node; > } > > - of_property_read_u32(np, "voltage-tolerance", &voltage_tolerance); > + ret = of_property_read_u32(np, "voltage-max", &voltage_max); > + if (ret) { > + pr_err("failed to get max voltage: %d\n", ret); > + goto out_put_node; > + } > > if (of_property_read_u32(np, "clock-latency", &transition_latency)) > transition_latency = CPUFREQ_ETERNAL; > -- > 1.7.5.4 > ^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-05 4:38 ` Viresh Kumar @ 2013-11-05 11:53 ` Shawn Guo 2013-11-06 2:15 ` Rongjun Ying 0 siblings, 1 reply; 9+ messages in thread From: Shawn Guo @ 2013-11-05 11:53 UTC (permalink / raw) To: Viresh Kumar Cc: rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, rongjun.ying, Rongjun Ying On Tue, Nov 05, 2013 at 10:08:06AM +0530, Viresh Kumar wrote: > Adding shawn in cc list.. > > On 5 November 2013 08:40, rjying <rjying@gmail.com> wrote: > > From: Rongjun Ying <rongjun.ying@csr.com> > > > > Sometime the regulator can't supply appropriate voltages for requestion of cpu. > > All voltage tolerance value can't figure out a good voltage. The voltage-tolerance is an optional device tree property. You can omit it if you do not have a value for it. Shawn > > > > Signed-off-by: Rongjun Ying <rongjun.ying@csr.com> > > --- > > drivers/cpufreq/cpufreq-cpu0.c | 17 ++++++++++------- > > 1 files changed, 10 insertions(+), 7 deletions(-) > > > > diff --git a/drivers/cpufreq/cpufreq-cpu0.c b/drivers/cpufreq/cpufreq-cpu0.c > > index c522a95..0c9e6b9 100644 > > --- a/drivers/cpufreq/cpufreq-cpu0.c > > +++ b/drivers/cpufreq/cpufreq-cpu0.c > > @@ -23,7 +23,7 @@ > > #include <linux/slab.h> > > > > static unsigned int transition_latency; > > -static unsigned int voltage_tolerance; /* in percentage */ > > +static unsigned int voltage_max; > > > > static struct device *cpu_dev; > > static struct clk *cpu_clk; > > @@ -45,7 +45,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > > { > > struct cpufreq_freqs freqs; > > struct opp *opp; > > - unsigned long volt = 0, volt_old = 0, tol = 0; > > + unsigned long volt = 0, volt_old = 0; > > long freq_Hz, freq_exact; > > unsigned int index; > > int ret; > > @@ -82,7 +82,6 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > > } > > volt = opp_get_voltage(opp); > > rcu_read_unlock(); > > - tol = volt * voltage_tolerance / 100; > > volt_old = regulator_get_voltage(cpu_reg); > > } > > > > @@ -92,7 +91,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > > > > /* scaling up? scale voltage before frequency */ > > if (!IS_ERR(cpu_reg) && freqs.new > freqs.old) { > > - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); > > + ret = regulator_set_voltage(cpu_reg, volt, voltage_max); > > if (ret) { > > pr_err("failed to scale voltage up: %d\n", ret); > > freqs.new = freqs.old; > > @@ -104,14 +103,14 @@ static int cpu0_set_target(struct cpufreq_policy *policy, > > if (ret) { > > pr_err("failed to set clock rate: %d\n", ret); > > if (!IS_ERR(cpu_reg)) > > - regulator_set_voltage_tol(cpu_reg, volt_old, tol); > > + regulator_set_voltage(cpu_reg, volt_old, voltage_max); > > freqs.new = freqs.old; > > goto post_notify; > > } > > > > /* scaling down? scale voltage after frequency */ > > if (!IS_ERR(cpu_reg) && freqs.new < freqs.old) { > > - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); > > + ret = regulator_set_voltage(cpu_reg, volt, voltage_max); > > if (ret) { > > pr_err("failed to scale voltage down: %d\n", ret); > > clk_set_rate(cpu_clk, freqs.old * 1000); > > @@ -224,7 +223,11 @@ static int cpu0_cpufreq_probe(struct platform_device *pdev) > > goto out_put_node; > > } > > > > - of_property_read_u32(np, "voltage-tolerance", &voltage_tolerance); > > + ret = of_property_read_u32(np, "voltage-max", &voltage_max); > > + if (ret) { > > + pr_err("failed to get max voltage: %d\n", ret); > > + goto out_put_node; > > + } > > > > if (of_property_read_u32(np, "clock-latency", &transition_latency)) > > transition_latency = CPUFREQ_ETERNAL; > > -- > > 1.7.5.4 > > ^ permalink raw reply [flat|nested] 9+ messages in thread
* RE: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-05 11:53 ` Shawn Guo @ 2013-11-06 2:15 ` Rongjun Ying 2013-11-06 3:27 ` Shawn Guo 0 siblings, 1 reply; 9+ messages in thread From: Rongjun Ying @ 2013-11-06 2:15 UTC (permalink / raw) To: Shawn Guo, Viresh Kumar Cc: rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, rongjun.ying > -----Original Message----- > From: Shawn Guo [mailto:shawn.guo@linaro.org] > Sent: Tuesday, November 05, 2013 7:53 PM > To: Viresh Kumar > Cc: rjying; Rafael J. Wysocki; cpufreq@vger.kernel.org; linux- > pm@vger.kernel.org; Linux Kernel Mailing List; rongjun.ying@gcsr.com; > Rongjun Ying > Subject: Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage > instead of voltage-tolerance > > On Tue, Nov 05, 2013 at 10:08:06AM +0530, Viresh Kumar wrote: > > Adding shawn in cc list.. > > > > On 5 November 2013 08:40, rjying <rjying@gmail.com> wrote: > > > From: Rongjun Ying <rongjun.ying@csr.com> > > > > > > Sometime the regulator can't supply appropriate voltages for > requestion of cpu. > > > All voltage tolerance value can't figure out a good voltage. > > The voltage-tolerance is an optional device tree property. You can > omit it if you do not have a value for it. > > Shawn It bounced for some reason, trying to send again. If omit voltage-tolerance, sometimes also can't get appropriate voltages. For example: If the regulator IC only can supply min voltage is 1.000V and max voltage is 1.200V, and cpu work max voltage is 1.200V. But the cpu just need 1.100V when cpu run under a freq. So regulator_set_voltage_tol will return failed. Because the regulator_set_voltage will invoke with min-uV is 1100000 and max-uV is 1100000 parameters. Regulator can't supply it. As this case, the regulator just need supply 1.200V. RongJun Ying > > > > > > > Signed-off-by: Rongjun Ying <rongjun.ying@csr.com> > > > --- > > > drivers/cpufreq/cpufreq-cpu0.c | 17 ++++++++++------- > > > 1 files changed, 10 insertions(+), 7 deletions(-) > > > > > > diff --git a/drivers/cpufreq/cpufreq-cpu0.c > > > b/drivers/cpufreq/cpufreq-cpu0.c index c522a95..0c9e6b9 100644 > > > --- a/drivers/cpufreq/cpufreq-cpu0.c > > > +++ b/drivers/cpufreq/cpufreq-cpu0.c > > > @@ -23,7 +23,7 @@ > > > #include <linux/slab.h> > > > > > > static unsigned int transition_latency; -static unsigned int > > > voltage_tolerance; /* in percentage */ > > > +static unsigned int voltage_max; > > > > > > static struct device *cpu_dev; > > > static struct clk *cpu_clk; > > > @@ -45,7 +45,7 @@ static int cpu0_set_target(struct cpufreq_policy > > > *policy, { > > > struct cpufreq_freqs freqs; > > > struct opp *opp; > > > - unsigned long volt = 0, volt_old = 0, tol = 0; > > > + unsigned long volt = 0, volt_old = 0; > > > long freq_Hz, freq_exact; > > > unsigned int index; > > > int ret; > > > @@ -82,7 +82,6 @@ static int cpu0_set_target(struct cpufreq_policy > *policy, > > > } > > > volt = opp_get_voltage(opp); > > > rcu_read_unlock(); > > > - tol = volt * voltage_tolerance / 100; > > > volt_old = regulator_get_voltage(cpu_reg); > > > } > > > > > > @@ -92,7 +91,7 @@ static int cpu0_set_target(struct cpufreq_policy > > > *policy, > > > > > > /* scaling up? scale voltage before frequency */ > > > if (!IS_ERR(cpu_reg) && freqs.new > freqs.old) { > > > - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); > > > + ret = regulator_set_voltage(cpu_reg, volt, > > > + voltage_max); > > > if (ret) { > > > pr_err("failed to scale voltage up: %d\n", > ret); > > > freqs.new = freqs.old; @@ -104,14 +103,14 > @@ > > > static int cpu0_set_target(struct cpufreq_policy *policy, > > > if (ret) { > > > pr_err("failed to set clock rate: %d\n", ret); > > > if (!IS_ERR(cpu_reg)) > > > - regulator_set_voltage_tol(cpu_reg, volt_old, > tol); > > > + regulator_set_voltage(cpu_reg, volt_old, > > > + voltage_max); > > > freqs.new = freqs.old; > > > goto post_notify; > > > } > > > > > > /* scaling down? scale voltage after frequency */ > > > if (!IS_ERR(cpu_reg) && freqs.new < freqs.old) { > > > - ret = regulator_set_voltage_tol(cpu_reg, volt, tol); > > > + ret = regulator_set_voltage(cpu_reg, volt, > > > + voltage_max); > > > if (ret) { > > > pr_err("failed to scale voltage down: %d\n", > ret); > > > clk_set_rate(cpu_clk, freqs.old * 1000); @@ > > > -224,7 +223,11 @@ static int cpu0_cpufreq_probe(struct > platform_device *pdev) > > > goto out_put_node; > > > } > > > > > > - of_property_read_u32(np, "voltage-tolerance", > &voltage_tolerance); > > > + ret = of_property_read_u32(np, "voltage-max", &voltage_max); > > > + if (ret) { > > > + pr_err("failed to get max voltage: %d\n", ret); > > > + goto out_put_node; > > > + } > > > > > > if (of_property_read_u32(np, "clock-latency", > &transition_latency)) > > > transition_latency = CPUFREQ_ETERNAL; > > > -- > > > 1.7.5.4 > > > > > > > To report this email as spam click > https://www.mailcontrol.com/sr/MZbqvYs5QwJvpeaetUwhCQ== > X88IR7qi8t8XvyHQDjR!61qPoTyr8GNFXkv3KeH4!zzvzA== . Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog ^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-06 2:15 ` Rongjun Ying @ 2013-11-06 3:27 ` Shawn Guo 2013-11-07 1:55 ` Rongjun Ying 0 siblings, 1 reply; 9+ messages in thread From: Shawn Guo @ 2013-11-06 3:27 UTC (permalink / raw) To: Rongjun Ying Cc: Viresh Kumar, rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, rongjun.ying On Wed, Nov 06, 2013 at 02:15:07AM +0000, Rongjun Ying wrote: > If omit voltage-tolerance, sometimes also can't get appropriate voltages. > For example: > If the regulator IC only can supply min voltage is 1.000V and max voltage is 1.200V, and cpu work max voltage is 1.200V. > But the cpu just need 1.100V when cpu run under a freq. > So regulator_set_voltage_tol will return failed. > Because the regulator_set_voltage will invoke with min-uV is 1100000 and max-uV is 1100000 parameters. > Regulator can't supply it. > As this case, the regulator just need supply 1.200V. For given board, what voltages could be provided is known. So you can just define OPP table in <board>.dts and specify the voltage as the value that the regulator IC can supply, e.g. 1.200V in above example. This is not nice, as OPP table is CPU/SoC specific and should be ideally defined in <soc>.dtsi. But still it's a way out for you to use cpufreq-cpu0 driver as it is. In any case, you can not just change voltage-tolerance to voltage-max with no care about the existing users. Shawn ^ permalink raw reply [flat|nested] 9+ messages in thread
* RE: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-06 3:27 ` Shawn Guo @ 2013-11-07 1:55 ` Rongjun Ying 2013-11-07 2:26 ` Shawn Guo 0 siblings, 1 reply; 9+ messages in thread From: Rongjun Ying @ 2013-11-07 1:55 UTC (permalink / raw) To: Shawn Guo Cc: Viresh Kumar, rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, Barry Song > -----Original Message----- > From: Shawn Guo [mailto:shawn.guo@linaro.org] > Sent: Wednesday, November 06, 2013 11:28 AM > To: Rongjun Ying > Cc: Viresh Kumar; rjying; Rafael J. Wysocki; cpufreq@vger.kernel.org; > linux-pm@vger.kernel.org; Linux Kernel Mailing List; > rongjun.ying@gcsr.com > Subject: Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage > instead of voltage-tolerance > > On Wed, Nov 06, 2013 at 02:15:07AM +0000, Rongjun Ying wrote: > > If omit voltage-tolerance, sometimes also can't get appropriate > voltages. > > For example: > > If the regulator IC only can supply min voltage is 1.000V and max > voltage is 1.200V, and cpu work max voltage is 1.200V. > > But the cpu just need 1.100V when cpu run under a freq. > > So regulator_set_voltage_tol will return failed. > > Because the regulator_set_voltage will invoke with min-uV is 1100000 > and max-uV is 1100000 parameters. > > Regulator can't supply it. > > As this case, the regulator just need supply 1.200V. > > For given board, what voltages could be provided is known. So you can > just define OPP table in <board>.dts and specify the voltage as the > value that the regulator IC can supply, e.g. 1.200V in above example. > > This is not nice, as OPP table is CPU/SoC specific and should be > ideally defined in <soc>.dtsi. But still it's a way out for you to use > cpufreq-cpu0 driver as it is. > > In any case, you can not just change voltage-tolerance to voltage-max > with no care about the existing users. > > Shawn > I don't think so. The voltage/freq pairs are attribute of the CPU. Any boards can choose regulator IC base the cost and other reasons. If the opp table defined in <board>.dts, we can set exact voltage/freq pairs, and not need use the voltage-tolerance to set tolerance. Thanks -RongJun Ying > > > To report this email as spam click > https://www.mailcontrol.com/sr/MZbqvYs5QwJvpeaetUwhCQ== > ch0RRg48Kxe8iwle4qrOOz8w5o+DwGNVtWag1M3YVHc55g== . Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog ^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-07 1:55 ` Rongjun Ying @ 2013-11-07 2:26 ` Shawn Guo 2013-11-07 3:32 ` Rongjun Ying 0 siblings, 1 reply; 9+ messages in thread From: Shawn Guo @ 2013-11-07 2:26 UTC (permalink / raw) To: Rongjun Ying Cc: Viresh Kumar, rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, Barry Song On Thu, Nov 07, 2013 at 01:55:21AM +0000, Rongjun Ying wrote: > > For given board, what voltages could be provided is known. So you can > > just define OPP table in <board>.dts and specify the voltage as the > > value that the regulator IC can supply, e.g. 1.200V in above example. > > > > This is not nice, as OPP table is CPU/SoC specific and should be > > ideally defined in <soc>.dtsi. But still it's a way out for you to use > > cpufreq-cpu0 driver as it is. > > > > In any case, you can not just change voltage-tolerance to voltage-max > > with no care about the existing users. > > > > Shawn > > > I don't think so. The voltage/freq pairs are attribute of the CPU. > Any boards can choose regulator IC base the cost and other reasons. > If the opp table defined in <board>.dts, we can set exact voltage/freq pairs, > and not need use the voltage-tolerance to set tolerance. If you read my comment above, you should see that I agree OPP is CPU/SoC specific and should be defined in <soc>.dtsi. But property operating-points can reasonably be overwritten by particular <board>.dts for some reason like some voltages cannot be supplied on that board. Again, this is just a way for you to use generic cpufreq-cpu0 driver as it is, but not necessarily the best one. Shawn ^ permalink raw reply [flat|nested] 9+ messages in thread
* RE: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-07 2:26 ` Shawn Guo @ 2013-11-07 3:32 ` Rongjun Ying 2013-11-07 4:55 ` Shawn Guo 0 siblings, 1 reply; 9+ messages in thread From: Rongjun Ying @ 2013-11-07 3:32 UTC (permalink / raw) To: Shawn Guo Cc: Viresh Kumar, rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, Barry Song > -----Original Message----- > From: Shawn Guo [mailto:shawn.guo@linaro.org] > Sent: Thursday, November 07, 2013 10:27 AM > To: Rongjun Ying > Cc: Viresh Kumar; rjying; Rafael J. Wysocki; cpufreq@vger.kernel.org; > linux-pm@vger.kernel.org; Linux Kernel Mailing List; Barry Song > Subject: Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage > instead of voltage-tolerance > > On Thu, Nov 07, 2013 at 01:55:21AM +0000, Rongjun Ying wrote: > > > For given board, what voltages could be provided is known. So you > > > can just define OPP table in <board>.dts and specify the voltage as > > > the value that the regulator IC can supply, e.g. 1.200V in above > example. > > > > > > This is not nice, as OPP table is CPU/SoC specific and should be > > > ideally defined in <soc>.dtsi. But still it's a way out for you to > > > use > > > cpufreq-cpu0 driver as it is. > > > > > > In any case, you can not just change voltage-tolerance to > > > voltage-max with no care about the existing users. > > > > > > Shawn > > > > > I don't think so. The voltage/freq pairs are attribute of the CPU. > > Any boards can choose regulator IC base the cost and other reasons. > > If the opp table defined in <board>.dts, we can set exact > voltage/freq > > pairs, and not need use the voltage-tolerance to set tolerance. > > If you read my comment above, you should see that I agree OPP is > CPU/SoC specific and should be defined in <soc>.dtsi. But property > operating-points can reasonably be overwritten by particular > <board>.dts for some reason like some voltages cannot be supplied on > that board. > > Again, this is just a way for you to use generic cpufreq-cpu0 driver as > it is, but not necessarily the best one. > > Shawn > I think my patch is better. All soc have a max work voltage. If regulator IC can't supply min voltage, It can supply max voltage to soc. As my patch, not need overwrite OPP in <board>.dts. As your comment above, almost all board need overwrite OPP, unless the regulator IC can supply all voltage ranges of SoC need. Thanks -RongJun Ying > > > To report this email as spam click > https://www.mailcontrol.com/sr/MZbqvYs5QwJvpeaetUwhCQ== > aM23S6+s5uWr7qEkawiSK+0AUqjCk0gm6KMeRLY+cWrx9g== . Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom More information can be found at www.csr.com. Follow CSR on Twitter at http://twitter.com/CSR_PLC and read our blog at www.csr.com/blog ^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance 2013-11-07 3:32 ` Rongjun Ying @ 2013-11-07 4:55 ` Shawn Guo 0 siblings, 0 replies; 9+ messages in thread From: Shawn Guo @ 2013-11-07 4:55 UTC (permalink / raw) To: Rongjun Ying Cc: Viresh Kumar, rjying, Rafael J. Wysocki, cpufreq, linux-pm, Linux Kernel Mailing List, Barry Song On Thu, Nov 07, 2013 at 03:32:09AM +0000, Rongjun Ying wrote: > I think my patch is better. Well, unfortunately your patch is plainly wrong, because there are in-tree users of voltage-tolerance right now. How can you simply remove it from driver with ignoring these users? $ git grep voltage-tolerance arch/arm/boot/dts arch/arm/boot/dts/am33xx.dtsi: voltage-tolerance = <2>; /* 2 percentage */ arch/arm/boot/dts/imx27.dtsi: voltage-tolerance = <5>; arch/arm/boot/dts/r8a73a4-ape6evm-reference.dts: voltage-tolerance = <1>; /* 1% */ arch/arm/boot/dts/r8a73a4-ape6evm.dts: voltage-tolerance = <1>; /* 1% */ arch/arm/boot/dts/sh73a0-kzm9g-reference.dts: voltage-tolerance = <1>; /* 1% */ Shawn > All soc have a max work voltage. If regulator IC can't supply min voltage, > It can supply max voltage to soc. > As my patch, not need overwrite OPP in <board>.dts. > As your comment above, almost all board need overwrite OPP, unless the regulator > IC can supply all voltage ranges of SoC need. ^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2013-11-07 4:55 UTC | newest] Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2013-11-05 3:10 [PATCH 1/1] cpufreq: cpufreq-cpu0: use the max voltage instead of voltage-tolerance rjying 2013-11-05 4:38 ` Viresh Kumar 2013-11-05 11:53 ` Shawn Guo 2013-11-06 2:15 ` Rongjun Ying 2013-11-06 3:27 ` Shawn Guo 2013-11-07 1:55 ` Rongjun Ying 2013-11-07 2:26 ` Shawn Guo 2013-11-07 3:32 ` Rongjun Ying 2013-11-07 4:55 ` Shawn Guo
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).