Add CPUFREQ_HAVE_GOVERNOR_PER_POLICY to have individual set of tunables for each cluster of MT8173.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org --- drivers/cpufreq/mt8173-cpufreq.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 83001dc..c438109 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -469,7 +469,8 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy) }
static struct cpufreq_driver mt8173_cpufreq_driver = { - .flags = CPUFREQ_STICKY | CPUFREQ_NEED_INITIAL_FREQ_CHECK, + .flags = CPUFREQ_STICKY | CPUFREQ_NEED_INITIAL_FREQ_CHECK | + CPUFREQ_HAVE_GOVERNOR_PER_POLICY, .verify = cpufreq_generic_frequency_table_verify, .target_index = mtk_cpufreq_set_target, .get = cpufreq_generic_get,
Remove redundant regulator_get_voltage() call to get Vsram value since it will be obtained later at the beginning of voltage tracking loop.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org --- drivers/cpufreq/mt8173-cpufreq.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index c438109..750cda7 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -59,7 +59,6 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int old_vproc, old_vsram, new_vsram, vsram, vproc, ret;
old_vproc = regulator_get_voltage(proc_reg); - old_vsram = regulator_get_voltage(sram_reg); /* Vsram should not exceed the maximum allowed voltage of SoC. */ new_vsram = min(new_vproc + MIN_VOLT_SHIFT, MAX_VOLT_LIMIT);
Sometimes regulator_get_voltage() call returns negative values for reasons(e.g. underlying I2C bus timeout). Add check for the return values and fail out early.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org --- drivers/cpufreq/mt8173-cpufreq.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 750cda7..9d0fe37 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -59,6 +59,10 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int old_vproc, old_vsram, new_vsram, vsram, vproc, ret;
old_vproc = regulator_get_voltage(proc_reg); + if (old_vproc < 0) { + pr_err("%s: invalid Vproc value: %d\n", __func__, old_vproc); + return old_vproc; + } /* Vsram should not exceed the maximum allowed voltage of SoC. */ new_vsram = min(new_vproc + MIN_VOLT_SHIFT, MAX_VOLT_LIMIT);
@@ -71,7 +75,17 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, */ do { old_vsram = regulator_get_voltage(sram_reg); + if (old_vsram < 0) { + pr_err("%s: invalid Vsram value: %d\n", + __func__, old_vsram); + return old_vsram; + } old_vproc = regulator_get_voltage(proc_reg); + if (old_vproc < 0) { + pr_err("%s: invalid Vproc value: %d\n", + __func__, old_vproc); + return old_vproc; + }
vsram = min(new_vsram, old_vproc + MAX_VOLT_SHIFT);
@@ -116,7 +130,17 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, */ do { old_vproc = regulator_get_voltage(proc_reg); + if (old_vproc < 0) { + pr_err("%s: invalid Vproc value: %d\n", + __func__, old_vproc); + return old_vproc; + } old_vsram = regulator_get_voltage(sram_reg); + if (old_vsram < 0) { + pr_err("%s: invalid Vsram value: %d\n", + __func__, old_vsram); + return old_vsram; + }
vproc = max(new_vproc, old_vsram - MAX_VOLT_SHIFT); ret = regulator_set_voltage(proc_reg, vproc, @@ -184,6 +208,10 @@ static int mtk_cpufreq_set_target(struct cpufreq_policy *policy,
old_freq_hz = clk_get_rate(cpu_clk); old_vproc = regulator_get_voltage(info->proc_reg); + if (old_vproc < 0) { + pr_err("%s: invalid Vproc value: %d\n", __func__, old_vproc); + return old_vproc; + }
freq_hz = freq_table[index].frequency * 1000;
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org --- drivers/cpufreq/mt8173-cpufreq.c | 94 ++++++++++++++++++++++++++++------------ 1 file changed, 67 insertions(+), 27 deletions(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 9d0fe37..1142625 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -41,16 +41,35 @@ * the original PLL becomes stable at target frequency. */ struct mtk_cpu_dvfs_info { + struct cpumask cpus; struct device *cpu_dev; struct regulator *proc_reg; struct regulator *sram_reg; struct clk *cpu_clk; struct clk *inter_clk; struct thermal_cooling_device *cdev; + struct list_head list_head; int intermediate_voltage; bool need_voltage_tracking; };
+static LIST_HEAD(dvfs_info_list); + +struct mtk_cpu_dvfs_info *get_dvfs_info(int cpu) +{ + struct mtk_cpu_dvfs_info *info; + struct list_head *list; + + list_for_each(list, &dvfs_info_list) { + info = list_entry(list, struct mtk_cpu_dvfs_info, list_head); + + if (cpumask_test_cpu(cpu, &info->cpus)) + return info; + } + + return NULL; +} + static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int new_vproc) { @@ -402,6 +421,9 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu) */ info->need_voltage_tracking = !IS_ERR(sram_reg);
+ /* CPUs in the same cluster share a clock and power domain. */ + cpumask_copy(&info->cpus, &cpu_topology[cpu].core_sibling); + return 0;
out_free_opp_table: @@ -440,47 +462,32 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy) struct cpufreq_frequency_table *freq_table; int ret;
- info = kzalloc(sizeof(*info), GFP_KERNEL); - if (!info) - return -ENOMEM; - - ret = mtk_cpu_dvfs_info_init(info, policy->cpu); - if (ret) { - pr_err("%s failed to initialize dvfs info for cpu%d\n", - __func__, policy->cpu); - goto out_free_dvfs_info; + info = get_dvfs_info(policy->cpu); + if (!info) { + pr_err("dvfs info for cpu%d is not initialized.\n", + policy->cpu); + return -EINVAL; }
ret = dev_pm_opp_init_cpufreq_table(info->cpu_dev, &freq_table); if (ret) { pr_err("failed to init cpufreq table for cpu%d: %d\n", policy->cpu, ret); - goto out_release_dvfs_info; + return ret; }
ret = cpufreq_table_validate_and_show(policy, freq_table); if (ret) { pr_err("%s: invalid frequency table: %d\n", __func__, ret); - goto out_free_cpufreq_table; + dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table); + return ret; }
- /* CPUs in the same cluster share a clock and power domain. */ - cpumask_copy(policy->cpus, &cpu_topology[policy->cpu].core_sibling); + cpumask_copy(policy->cpus, &info->cpus); policy->driver_data = info; policy->clk = info->cpu_clk;
return 0; - -out_free_cpufreq_table: - dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table); - -out_release_dvfs_info: - mtk_cpu_dvfs_info_release(info); - -out_free_dvfs_info: - kfree(info); - - return ret; }
static int mtk_cpufreq_exit(struct cpufreq_policy *policy) @@ -489,8 +496,6 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy)
cpufreq_cooling_unregister(info->cdev); dev_pm_opp_free_cpufreq_table(info->cpu_dev, &policy->freq_table); - mtk_cpu_dvfs_info_release(info); - kfree(info);
return 0; } @@ -510,12 +515,47 @@ static struct cpufreq_driver mt8173_cpufreq_driver = {
static int mt8173_cpufreq_probe(struct platform_device *pdev) { - int ret; + struct mtk_cpu_dvfs_info *info; + struct list_head *list, *tmp; + int cpu, ret; + + for_each_possible_cpu(cpu) { + info = get_dvfs_info(cpu); + if (info) + continue; + + info = kzalloc(sizeof(*info), GFP_KERNEL); + if (!info) { + ret = -ENOMEM; + goto release_dvfs_info_list; + } + + ret = mtk_cpu_dvfs_info_init(info, cpu); + if (ret) { + pr_err("%s failed to initialize dvfs info for cpu%d\n", + __func__, cpu); + kfree(info); + goto release_dvfs_info_list; + } + + list_add(&info->list_head, &dvfs_info_list); + }
ret = cpufreq_register_driver(&mt8173_cpufreq_driver); if (ret) pr_err("failed to register mtk cpufreq driver\n");
+ return 0; + +release_dvfs_info_list: + list_for_each_safe(list, tmp, &dvfs_info_list) { + info = list_entry(list, struct mtk_cpu_dvfs_info, list_head); + + mtk_cpu_dvfs_info_release(info); + list_del(list); + kfree(info); + } + return ret; }
On 29-11-15, 16:31, Pi-Cheng Chen wrote:
Add CPUFREQ_HAVE_GOVERNOR_PER_POLICY to have individual set of tunables for each cluster of MT8173.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
drivers/cpufreq/mt8173-cpufreq.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 83001dc..c438109 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -469,7 +469,8 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy) } static struct cpufreq_driver mt8173_cpufreq_driver = {
- .flags = CPUFREQ_STICKY | CPUFREQ_NEED_INITIAL_FREQ_CHECK,
- .flags = CPUFREQ_STICKY | CPUFREQ_NEED_INITIAL_FREQ_CHECK |
CPUFREQ_HAVE_GOVERNOR_PER_POLICY,
Acked-by: Viresh Kumar viresh.kumar@linaro.org
On 29-11-15, 16:31, Pi-Cheng Chen wrote:
Remove redundant regulator_get_voltage() call to get Vsram value since it will be obtained later at the beginning of voltage tracking loop.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
drivers/cpufreq/mt8173-cpufreq.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index c438109..750cda7 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -59,7 +59,6 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int old_vproc, old_vsram, new_vsram, vsram, vproc, ret; old_vproc = regulator_get_voltage(proc_reg);
- old_vsram = regulator_get_voltage(sram_reg); /* Vsram should not exceed the maximum allowed voltage of SoC. */ new_vsram = min(new_vproc + MIN_VOLT_SHIFT, MAX_VOLT_LIMIT);
Acked-by: Viresh Kumar viresh.kumar@linaro.org
On 29-11-15, 16:31, Pi-Cheng Chen wrote:
Sometimes regulator_get_voltage() call returns negative values for reasons(e.g. underlying I2C bus timeout). Add check for the return values and fail out early.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
drivers/cpufreq/mt8173-cpufreq.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 750cda7..9d0fe37 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -59,6 +59,10 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int old_vproc, old_vsram, new_vsram, vsram, vproc, ret; old_vproc = regulator_get_voltage(proc_reg);
- if (old_vproc < 0) {
pr_err("%s: invalid Vproc value: %d\n", __func__, old_vproc);return old_vproc;- } /* Vsram should not exceed the maximum allowed voltage of SoC. */ new_vsram = min(new_vproc + MIN_VOLT_SHIFT, MAX_VOLT_LIMIT);
@@ -71,7 +75,17 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, */ do { old_vsram = regulator_get_voltage(sram_reg);
if (old_vsram < 0) {pr_err("%s: invalid Vsram value: %d\n",__func__, old_vsram);return old_vsram;} old_vproc = regulator_get_voltage(proc_reg);if (old_vproc < 0) {pr_err("%s: invalid Vproc value: %d\n",__func__, old_vproc);return old_vproc;}vsram = min(new_vsram, old_vproc + MAX_VOLT_SHIFT); @@ -116,7 +130,17 @@ static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, */ do { old_vproc = regulator_get_voltage(proc_reg);
if (old_vproc < 0) {pr_err("%s: invalid Vproc value: %d\n",__func__, old_vproc);return old_vproc;} old_vsram = regulator_get_voltage(sram_reg);if (old_vsram < 0) {pr_err("%s: invalid Vsram value: %d\n",__func__, old_vsram);return old_vsram;}vproc = max(new_vproc, old_vsram - MAX_VOLT_SHIFT); ret = regulator_set_voltage(proc_reg, vproc, @@ -184,6 +208,10 @@ static int mtk_cpufreq_set_target(struct cpufreq_policy *policy, old_freq_hz = clk_get_rate(cpu_clk); old_vproc = regulator_get_voltage(info->proc_reg);
- if (old_vproc < 0) {
pr_err("%s: invalid Vproc value: %d\n", __func__, old_vproc);return old_vproc;- }
freq_hz = freq_table[index].frequency * 1000;
Acked-by: Viresh Kumar viresh.kumar@linaro.org
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org --- v1->v2: - fix error handling path in ->probe() --- drivers/cpufreq/mt8173-cpufreq.c | 98 ++++++++++++++++++++++++++++------------ 1 file changed, 70 insertions(+), 28 deletions(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 9d0fe37..257bcb9 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -41,16 +41,35 @@ * the original PLL becomes stable at target frequency. */ struct mtk_cpu_dvfs_info { + struct cpumask cpus; struct device *cpu_dev; struct regulator *proc_reg; struct regulator *sram_reg; struct clk *cpu_clk; struct clk *inter_clk; struct thermal_cooling_device *cdev; + struct list_head list_head; int intermediate_voltage; bool need_voltage_tracking; };
+static LIST_HEAD(dvfs_info_list); + +struct mtk_cpu_dvfs_info *get_dvfs_info(int cpu) +{ + struct mtk_cpu_dvfs_info *info; + struct list_head *list; + + list_for_each(list, &dvfs_info_list) { + info = list_entry(list, struct mtk_cpu_dvfs_info, list_head); + + if (cpumask_test_cpu(cpu, &info->cpus)) + return info; + } + + return NULL; +} + static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int new_vproc) { @@ -402,6 +421,9 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu) */ info->need_voltage_tracking = !IS_ERR(sram_reg);
+ /* CPUs in the same cluster share a clock and power domain. */ + cpumask_copy(&info->cpus, &cpu_topology[cpu].core_sibling); + return 0;
out_free_opp_table: @@ -440,47 +462,32 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy) struct cpufreq_frequency_table *freq_table; int ret;
- info = kzalloc(sizeof(*info), GFP_KERNEL); - if (!info) - return -ENOMEM; - - ret = mtk_cpu_dvfs_info_init(info, policy->cpu); - if (ret) { - pr_err("%s failed to initialize dvfs info for cpu%d\n", - __func__, policy->cpu); - goto out_free_dvfs_info; + info = get_dvfs_info(policy->cpu); + if (!info) { + pr_err("dvfs info for cpu%d is not initialized.\n", + policy->cpu); + return -EINVAL; }
ret = dev_pm_opp_init_cpufreq_table(info->cpu_dev, &freq_table); if (ret) { pr_err("failed to init cpufreq table for cpu%d: %d\n", policy->cpu, ret); - goto out_release_dvfs_info; + return ret; }
ret = cpufreq_table_validate_and_show(policy, freq_table); if (ret) { pr_err("%s: invalid frequency table: %d\n", __func__, ret); - goto out_free_cpufreq_table; + dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table); + return ret; }
- /* CPUs in the same cluster share a clock and power domain. */ - cpumask_copy(policy->cpus, &cpu_topology[policy->cpu].core_sibling); + cpumask_copy(policy->cpus, &info->cpus); policy->driver_data = info; policy->clk = info->cpu_clk;
return 0; - -out_free_cpufreq_table: - dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table); - -out_release_dvfs_info: - mtk_cpu_dvfs_info_release(info); - -out_free_dvfs_info: - kfree(info); - - return ret; }
static int mtk_cpufreq_exit(struct cpufreq_policy *policy) @@ -489,8 +496,6 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy)
cpufreq_cooling_unregister(info->cdev); dev_pm_opp_free_cpufreq_table(info->cpu_dev, &policy->freq_table); - mtk_cpu_dvfs_info_release(info); - kfree(info);
return 0; } @@ -510,11 +515,48 @@ static struct cpufreq_driver mt8173_cpufreq_driver = {
static int mt8173_cpufreq_probe(struct platform_device *pdev) { - int ret; + struct mtk_cpu_dvfs_info *info; + struct list_head *list, *tmp; + int cpu, ret; + + for_each_possible_cpu(cpu) { + info = get_dvfs_info(cpu); + if (info) + continue; + + info = kzalloc(sizeof(*info), GFP_KERNEL); + if (!info) { + ret = -ENOMEM; + goto release_dvfs_info_list; + } + + ret = mtk_cpu_dvfs_info_init(info, cpu); + if (ret) { + pr_err("%s failed to initialize dvfs info for cpu%d\n", + __func__, cpu); + kfree(info); + goto release_dvfs_info_list; + } + + list_add(&info->list_head, &dvfs_info_list); + }
ret = cpufreq_register_driver(&mt8173_cpufreq_driver); - if (ret) + if (ret) { pr_err("failed to register mtk cpufreq driver\n"); + goto release_dvfs_info_list; + } + + return 0; + +release_dvfs_info_list: + list_for_each_safe(list, tmp, &dvfs_info_list) { + info = list_entry(list, struct mtk_cpu_dvfs_info, list_head); + + mtk_cpu_dvfs_info_release(info); + list_del(list); + kfree(info); + }
return ret; }
On 05-12-15, 14:01, Pi-Cheng Chen wrote:
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
v1->v2:
- fix error handling path in ->probe()
Acked-by: Viresh Kumar viresh.kumar@linaro.org
Hi Pi-Cheng,
On Sat, Dec 5, 2015 at 2:01 PM, Pi-Cheng Chen pi-cheng.chen@linaro.org wrote:
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
v1->v2:
- fix error handling path in ->probe()
drivers/cpufreq/mt8173-cpufreq.c | 98 ++++++++++++++++++++++++++++------------ 1 file changed, 70 insertions(+), 28 deletions(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 9d0fe37..257bcb9 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -41,16 +41,35 @@
- the original PLL becomes stable at target frequency.
*/ struct mtk_cpu_dvfs_info {
struct cpumask cpus; struct device *cpu_dev; struct regulator *proc_reg; struct regulator *sram_reg; struct clk *cpu_clk; struct clk *inter_clk; struct thermal_cooling_device *cdev;struct list_head list_head; int intermediate_voltage; bool need_voltage_tracking;};
+static LIST_HEAD(dvfs_info_list);
This struct is specific to this driver, so:
mtk_cpu_dvfs_info_list
+struct mtk_cpu_dvfs_info *get_dvfs_info(int cpu)
static
Also, it would be clearer to name this function something like "mtk_cpu_dfs_info_lookup" since "mtk_cpu_dfs_info" is the type this function looks up in the list. It would also match the other functions such as "mtk_cpu_dvfs_info_init()".
+{
struct mtk_cpu_dvfs_info *info;struct list_head *list;list_for_each(list, &dvfs_info_list) {info = list_entry(list, struct mtk_cpu_dvfs_info, list_head);if (cpumask_test_cpu(cpu, &info->cpus))return info;}return NULL;+}
static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int new_vproc) { @@ -402,6 +421,9 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu) */ info->need_voltage_tracking = !IS_ERR(sram_reg);
/* CPUs in the same cluster share a clock and power domain. */cpumask_copy(&info->cpus, &cpu_topology[cpu].core_sibling);return 0;out_free_opp_table: @@ -440,47 +462,32 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy) struct cpufreq_frequency_table *freq_table; int ret;
info = kzalloc(sizeof(*info), GFP_KERNEL);if (!info)return -ENOMEM;ret = mtk_cpu_dvfs_info_init(info, policy->cpu);if (ret) {pr_err("%s failed to initialize dvfs info for cpu%d\n",__func__, policy->cpu);goto out_free_dvfs_info;
info = get_dvfs_info(policy->cpu);if (!info) {pr_err("dvfs info for cpu%d is not initialized.\n",policy->cpu);return -EINVAL; } ret = dev_pm_opp_init_cpufreq_table(info->cpu_dev, &freq_table);
Hmm. Does this change at init() time? Or can we move this to probe(), too?
if (ret) { pr_err("failed to init cpufreq table for cpu%d: %d\n", policy->cpu, ret);
goto out_release_dvfs_info;
return ret; } ret = cpufreq_table_validate_and_show(policy, freq_table); if (ret) { pr_err("%s: invalid frequency table: %d\n", __func__, ret);
goto out_free_cpufreq_table;
dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table);return ret;
If dev_pm_opp_init_cpufreq_table() really is needed at init() time, then please don't change this; still use "goto out_free_cpufreq_table" here.
}
/* CPUs in the same cluster share a clock and power domain. */cpumask_copy(policy->cpus, &cpu_topology[policy->cpu].core_sibling);
cpumask_copy(policy->cpus, &info->cpus); policy->driver_data = info; policy->clk = info->cpu_clk; return 0;
-out_free_cpufreq_table:
dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table);-out_release_dvfs_info:
mtk_cpu_dvfs_info_release(info);-out_free_dvfs_info:
kfree(info);return ret;}
static int mtk_cpufreq_exit(struct cpufreq_policy *policy) @@ -489,8 +496,6 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy)
cpufreq_cooling_unregister(info->cdev); dev_pm_opp_free_cpufreq_table(info->cpu_dev, &policy->freq_table);
mtk_cpu_dvfs_info_release(info);kfree(info); return 0;} @@ -510,11 +515,48 @@ static struct cpufreq_driver mt8173_cpufreq_driver = {
static int mt8173_cpufreq_probe(struct platform_device *pdev) {
int ret;
struct mtk_cpu_dvfs_info *info;struct list_head *list, *tmp;int cpu, ret;for_each_possible_cpu(cpu) {
Is there an iterator for cpu clusters? Would be useful here.
Or, even better: * Is there a way to way to map cpu to cpu cluster ID? * Is there a way to count the number of clusters? If so you can just use an array of mtk_cpu_dvfs_info, and use a cpu's cluster ID as its index into the array.
info = get_dvfs_info(cpu);if (info)continue;
How could this ever return non-NULL? This is probe; it won't be called twice - not unless a prior probe() failed (and/or after .remove()), right?
info = kzalloc(sizeof(*info), GFP_KERNEL);
devm_kzalloc() (and you can then remove the kfree()'s below).
if (!info) {ret = -ENOMEM;goto release_dvfs_info_list;}ret = mtk_cpu_dvfs_info_init(info, cpu);if (ret) {pr_err("%s failed to initialize dvfs info for cpu%d\n",__func__, cpu);
dev_err(&pdev->dev, ) here and elsewhere.
-Dan
kfree(info);goto release_dvfs_info_list;}list_add(&info->list_head, &dvfs_info_list);} ret = cpufreq_register_driver(&mt8173_cpufreq_driver);
if (ret)
if (ret) { pr_err("failed to register mtk cpufreq driver\n");goto release_dvfs_info_list;}return 0;+release_dvfs_info_list:
list_for_each_safe(list, tmp, &dvfs_info_list) {info = list_entry(list, struct mtk_cpu_dvfs_info, list_head);mtk_cpu_dvfs_info_release(info);list_del(list);kfree(info);} return ret;}
1.9.1
Hi Daniel,
On Mon, Dec 7, 2015 at 10:44 AM, Daniel Kurtz djkurtz@chromium.org wrote:
Hi Pi-Cheng,
On Sat, Dec 5, 2015 at 2:01 PM, Pi-Cheng Chen pi-cheng.chen@linaro.org wrote:
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
v1->v2:
- fix error handling path in ->probe()
drivers/cpufreq/mt8173-cpufreq.c | 98 ++++++++++++++++++++++++++++------------ 1 file changed, 70 insertions(+), 28 deletions(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 9d0fe37..257bcb9 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -41,16 +41,35 @@
- the original PLL becomes stable at target frequency.
*/ struct mtk_cpu_dvfs_info {
struct cpumask cpus; struct device *cpu_dev; struct regulator *proc_reg; struct regulator *sram_reg; struct clk *cpu_clk; struct clk *inter_clk; struct thermal_cooling_device *cdev;struct list_head list_head; int intermediate_voltage; bool need_voltage_tracking;};
+static LIST_HEAD(dvfs_info_list);
This struct is specific to this driver, so:
mtk_cpu_dvfs_info_list
That's fair. Both work for me. But since it's a static variable, I prefer the shorter name since it won't be misunderstood.
+struct mtk_cpu_dvfs_info *get_dvfs_info(int cpu)
static
Will do it.
Also, it would be clearer to name this function something like "mtk_cpu_dfs_info_lookup" since "mtk_cpu_dfs_info" is the type this function looks up in the list. It would also match the other functions such as "mtk_cpu_dvfs_info_init()".
mtk_cpu_dvfs_info_lookup() looks good to me. WIll do it.
+{
struct mtk_cpu_dvfs_info *info;struct list_head *list;list_for_each(list, &dvfs_info_list) {info = list_entry(list, struct mtk_cpu_dvfs_info, list_head);if (cpumask_test_cpu(cpu, &info->cpus))return info;}return NULL;+}
static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int new_vproc) { @@ -402,6 +421,9 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu) */ info->need_voltage_tracking = !IS_ERR(sram_reg);
/* CPUs in the same cluster share a clock and power domain. */cpumask_copy(&info->cpus, &cpu_topology[cpu].core_sibling);return 0;out_free_opp_table: @@ -440,47 +462,32 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy) struct cpufreq_frequency_table *freq_table; int ret;
info = kzalloc(sizeof(*info), GFP_KERNEL);if (!info)return -ENOMEM;ret = mtk_cpu_dvfs_info_init(info, policy->cpu);if (ret) {pr_err("%s failed to initialize dvfs info for cpu%d\n",__func__, policy->cpu);goto out_free_dvfs_info;
info = get_dvfs_info(policy->cpu);if (!info) {pr_err("dvfs info for cpu%d is not initialized.\n",policy->cpu);return -EINVAL; } ret = dev_pm_opp_init_cpufreq_table(info->cpu_dev, &freq_table);Hmm. Does this change at init() time? Or can we move this to probe(), too?
It doesn't change at init() time for MT8173 since we don't add OPP dynamically and I don't see any problem to move it. I will move it to probe().
if (ret) { pr_err("failed to init cpufreq table for cpu%d: %d\n", policy->cpu, ret);
goto out_release_dvfs_info;
return ret; } ret = cpufreq_table_validate_and_show(policy, freq_table); if (ret) { pr_err("%s: invalid frequency table: %d\n", __func__, ret);
goto out_free_cpufreq_table;
dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table);return ret;If dev_pm_opp_init_cpufreq_table() really is needed at init() time, then please don't change this; still use "goto out_free_cpufreq_table" here.
}
/* CPUs in the same cluster share a clock and power domain. */cpumask_copy(policy->cpus, &cpu_topology[policy->cpu].core_sibling);
cpumask_copy(policy->cpus, &info->cpus); policy->driver_data = info; policy->clk = info->cpu_clk; return 0;
-out_free_cpufreq_table:
dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table);-out_release_dvfs_info:
mtk_cpu_dvfs_info_release(info);-out_free_dvfs_info:
kfree(info);return ret;}
static int mtk_cpufreq_exit(struct cpufreq_policy *policy) @@ -489,8 +496,6 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy)
cpufreq_cooling_unregister(info->cdev); dev_pm_opp_free_cpufreq_table(info->cpu_dev, &policy->freq_table);
mtk_cpu_dvfs_info_release(info);kfree(info); return 0;} @@ -510,11 +515,48 @@ static struct cpufreq_driver mt8173_cpufreq_driver = {
static int mt8173_cpufreq_probe(struct platform_device *pdev) {
int ret;
struct mtk_cpu_dvfs_info *info;struct list_head *list, *tmp;int cpu, ret;for_each_possible_cpu(cpu) {Is there an iterator for cpu clusters? Would be useful here.
Or, even better:
- Is there a way to way to map cpu to cpu cluster ID?
- Is there a way to count the number of clusters?
If so you can just use an array of mtk_cpu_dvfs_info, and use a cpu's cluster ID as its index into the array.
As I originally planned, this driver should be easily extended support for other Mediatek platforms. For this reason, we should not assume there's a fixed mapping for CPU ID and cluster ID and the CPU DVFS domain and cluster neither. Furthermore, the CPU DVFS domain sharing information will be provided by operating-points-v2 bindings on which I am working to migrate mt8173-cpufreq driver. We should not put the same information in two different places.
info = get_dvfs_info(cpu);if (info)continue;How could this ever return non-NULL? This is probe; it won't be called twice - not unless a prior probe() failed (and/or after .remove()), right?
Yes. it does return non-NULL for CPU1 and CPU3. When it goes into the loop with cpu=0, mtk_cpu_dvfs_info_init() will copy the cluster core_sibling mask to info->cpus:
cpumask_copy(&info->cpus, &cpu_topology[cpu].core_sibling);
which means cpu0 and cpu1 share this struct mtk_cpu_dvfs_info variable. So when it goes into the loop with cpu=1, get_dvfs_info() will return the shared data structure. Similar for CPU2 and CPU3.
BTW, the shared information from core_sibling cpumask will be replaced by operating-points-v2 bindings later.
info = kzalloc(sizeof(*info), GFP_KERNEL);devm_kzalloc() (and you can then remove the kfree()'s below).
Yes. Will do it.
if (!info) {ret = -ENOMEM;goto release_dvfs_info_list;}ret = mtk_cpu_dvfs_info_init(info, cpu);if (ret) {pr_err("%s failed to initialize dvfs info for cpu%d\n",__func__, cpu);dev_err(&pdev->dev, ) here and elsewhere.
Will do it. Thanks for reviewing.
Pi-Cheng
-Dan
kfree(info);goto release_dvfs_info_list;}list_add(&info->list_head, &dvfs_info_list);} ret = cpufreq_register_driver(&mt8173_cpufreq_driver);
if (ret)
if (ret) { pr_err("failed to register mtk cpufreq driver\n");goto release_dvfs_info_list;}return 0;+release_dvfs_info_list:
list_for_each_safe(list, tmp, &dvfs_info_list) {info = list_entry(list, struct mtk_cpu_dvfs_info, list_head);mtk_cpu_dvfs_info_release(info);list_del(list);kfree(info);} return ret;}
1.9.1
On 07-12-15, 20:42, Pi-Cheng Chen wrote:
ret = dev_pm_opp_init_cpufreq_table(info->cpu_dev, &freq_table);Hmm. Does this change at init() time? Or can we move this to probe(), too?
It doesn't change at init() time for MT8173 since we don't add OPP dynamically and I don't see any problem to move it. I will move it to probe().
No. None of this belongs to probe(), we do it there just to handle the EPROBE_DEFER thing. Please don't move anything that isn't required to be.
On Saturday, December 05, 2015 02:01:46 PM Pi-Cheng Chen wrote:
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org
I'm expecting a v3 here. Please let me know if I should apply this instead.
Thanks, Rafael
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org Acked-by: Viresh Kumar viresh.kumar@linaro.org --- v2->v3: - rename get_dvfs_info() as mtk_cpu_dvfs_info_lookup() - fix some minor issues of v2 v1->v2: - fix error handling path in ->probe() --- drivers/cpufreq/mt8173-cpufreq.c | 92 +++++++++++++++++++++++++++++----------- 1 file changed, 68 insertions(+), 24 deletions(-)
diff --git a/drivers/cpufreq/mt8173-cpufreq.c b/drivers/cpufreq/mt8173-cpufreq.c index 9d0fe37..fd601b9 100644 --- a/drivers/cpufreq/mt8173-cpufreq.c +++ b/drivers/cpufreq/mt8173-cpufreq.c @@ -41,16 +41,35 @@ * the original PLL becomes stable at target frequency. */ struct mtk_cpu_dvfs_info { + struct cpumask cpus; struct device *cpu_dev; struct regulator *proc_reg; struct regulator *sram_reg; struct clk *cpu_clk; struct clk *inter_clk; struct thermal_cooling_device *cdev; + struct list_head list_head; int intermediate_voltage; bool need_voltage_tracking; };
+static LIST_HEAD(dvfs_info_list); + +static struct mtk_cpu_dvfs_info *mtk_cpu_dvfs_info_lookup(int cpu) +{ + struct mtk_cpu_dvfs_info *info; + struct list_head *list; + + list_for_each(list, &dvfs_info_list) { + info = list_entry(list, struct mtk_cpu_dvfs_info, list_head); + + if (cpumask_test_cpu(cpu, &info->cpus)) + return info; + } + + return NULL; +} + static int mtk_cpufreq_voltage_tracking(struct mtk_cpu_dvfs_info *info, int new_vproc) { @@ -402,6 +421,9 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu) */ info->need_voltage_tracking = !IS_ERR(sram_reg);
+ /* CPUs in the same cluster share a clock and power domain. */ + cpumask_copy(&info->cpus, &cpu_topology[cpu].core_sibling); + return 0;
out_free_opp_table: @@ -440,22 +462,18 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy) struct cpufreq_frequency_table *freq_table; int ret;
- info = kzalloc(sizeof(*info), GFP_KERNEL); - if (!info) - return -ENOMEM; - - ret = mtk_cpu_dvfs_info_init(info, policy->cpu); - if (ret) { - pr_err("%s failed to initialize dvfs info for cpu%d\n", - __func__, policy->cpu); - goto out_free_dvfs_info; + info = mtk_cpu_dvfs_info_lookup(policy->cpu); + if (!info) { + pr_err("dvfs info for cpu%d is not initialized.\n", + policy->cpu); + return -EINVAL; }
ret = dev_pm_opp_init_cpufreq_table(info->cpu_dev, &freq_table); if (ret) { pr_err("failed to init cpufreq table for cpu%d: %d\n", policy->cpu, ret); - goto out_release_dvfs_info; + return ret; }
ret = cpufreq_table_validate_and_show(policy, freq_table); @@ -464,8 +482,7 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy) goto out_free_cpufreq_table; }
- /* CPUs in the same cluster share a clock and power domain. */ - cpumask_copy(policy->cpus, &cpu_topology[policy->cpu].core_sibling); + cpumask_copy(policy->cpus, &info->cpus); policy->driver_data = info; policy->clk = info->cpu_clk;
@@ -473,13 +490,6 @@ static int mtk_cpufreq_init(struct cpufreq_policy *policy)
out_free_cpufreq_table: dev_pm_opp_free_cpufreq_table(info->cpu_dev, &freq_table); - -out_release_dvfs_info: - mtk_cpu_dvfs_info_release(info); - -out_free_dvfs_info: - kfree(info); - return ret; }
@@ -489,8 +499,6 @@ static int mtk_cpufreq_exit(struct cpufreq_policy *policy)
cpufreq_cooling_unregister(info->cdev); dev_pm_opp_free_cpufreq_table(info->cpu_dev, &policy->freq_table); - mtk_cpu_dvfs_info_release(info); - kfree(info);
return 0; } @@ -510,11 +518,47 @@ static struct cpufreq_driver mt8173_cpufreq_driver = {
static int mt8173_cpufreq_probe(struct platform_device *pdev) { - int ret; + struct mtk_cpu_dvfs_info *info; + struct list_head *list, *tmp; + int cpu, ret; + + for_each_possible_cpu(cpu) { + info = mtk_cpu_dvfs_info_lookup(cpu); + if (info) + continue; + + info = devm_kzalloc(&pdev->dev, sizeof(*info), GFP_KERNEL); + if (!info) { + ret = -ENOMEM; + goto release_dvfs_info_list; + } + + ret = mtk_cpu_dvfs_info_init(info, cpu); + if (ret) { + dev_err(&pdev->dev, + "failed to initialize dvfs info for cpu%d\n", + cpu); + goto release_dvfs_info_list; + } + + list_add(&info->list_head, &dvfs_info_list); + }
ret = cpufreq_register_driver(&mt8173_cpufreq_driver); - if (ret) - pr_err("failed to register mtk cpufreq driver\n"); + if (ret) { + dev_err(&pdev->dev, "failed to register mtk cpufreq driver\n"); + goto release_dvfs_info_list; + } + + return 0; + +release_dvfs_info_list: + list_for_each_safe(list, tmp, &dvfs_info_list) { + info = list_entry(list, struct mtk_cpu_dvfs_info, list_head); + + mtk_cpu_dvfs_info_release(info); + list_del(list); + }
return ret; }
On Thursday, December 10, 2015 11:48:13 AM Pi-Cheng Chen wrote:
Since the return value of ->init() of cpufreq driver is not propagated to the device driver model now, move resources allocation into ->probe() to handle -EPROBE_DEFER properly.
Signed-off-by: Pi-Cheng Chen pi-cheng.chen@linaro.org Acked-by: Viresh Kumar viresh.kumar@linaro.org
Applied along with the [1-3] in this series.
Thanks, Rafael
linaro-kernel@lists.linaro.org
-
Daniel Kurtz -
Pi-Cheng Chen -
Rafael J. Wysocki -
Viresh Kumar