* [PATCH 1/4] cpufreq: mt8173: add CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag
@ 2015-11-29 8:31 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: Viresh Kumar, Matthias Brugger
Cc: Daniel Kurtz, linux-mediatek, linaro-kernel, linux-arm-kernel, linux-pm
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,
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 1/4] cpufreq: mt8173: add CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag
@ 2015-11-29 8:31 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: linux-arm-kernel
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,
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 2/4] cpufreq: mt8173: remove redundant regulator_get_voltage() call
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-11-29 8:31 ` Pi-Cheng Chen
-1 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: Viresh Kumar, Matthias Brugger
Cc: Daniel Kurtz, linux-mediatek, linaro-kernel, linux-arm-kernel, linux-pm
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);
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 2/4] cpufreq: mt8173: remove redundant regulator_get_voltage() call
@ 2015-11-29 8:31 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: linux-arm-kernel
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);
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 3/4] cpufreq: mt8173: check return value of regulator_get_voltage() call
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-11-29 8:31 ` Pi-Cheng Chen
-1 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: Viresh Kumar, Matthias Brugger
Cc: Daniel Kurtz, linux-mediatek, linaro-kernel, linux-arm-kernel, linux-pm
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;
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 3/4] cpufreq: mt8173: check return value of regulator_get_voltage() call
@ 2015-11-29 8:31 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: linux-arm-kernel
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;
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 4/4] cpufreq: mt8173: move resources allocation into ->probe()
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-11-29 8:31 ` Pi-Cheng Chen
-1 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: Viresh Kumar, Matthias Brugger
Cc: Daniel Kurtz, linux-mediatek, linaro-kernel, linux-arm-kernel, linux-pm
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;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH 4/4] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-11-29 8:31 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-11-29 8:31 UTC (permalink / raw)
To: linux-arm-kernel
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;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH 1/4] cpufreq: mt8173: add CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-11-30 5:20 ` Viresh Kumar
-1 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-11-30 5:20 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Matthias Brugger, Daniel Kurtz, linux-mediatek, linaro-kernel,
linux-arm-kernel, linux-pm
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,
> .verify = cpufreq_generic_frequency_table_verify,
> .target_index = mtk_cpufreq_set_target,
> .get = cpufreq_generic_get,
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 1/4] cpufreq: mt8173: add CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag
@ 2015-11-30 5:20 ` Viresh Kumar
0 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-11-30 5:20 UTC (permalink / raw)
To: linux-arm-kernel
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,
> .verify = cpufreq_generic_frequency_table_verify,
> .target_index = mtk_cpufreq_set_target,
> .get = cpufreq_generic_get,
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 2/4] cpufreq: mt8173: remove redundant regulator_get_voltage() call
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-11-30 5:23 ` Viresh Kumar
-1 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-11-30 5:23 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Matthias Brugger, Daniel Kurtz, linux-mediatek, linaro-kernel,
linux-arm-kernel, linux-pm
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>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 2/4] cpufreq: mt8173: remove redundant regulator_get_voltage() call
@ 2015-11-30 5:23 ` Viresh Kumar
0 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-11-30 5:23 UTC (permalink / raw)
To: linux-arm-kernel
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>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH 3/4] cpufreq: mt8173: check return value of regulator_get_voltage() call
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-11-30 5:27 ` Viresh Kumar
-1 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-11-30 5:27 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Matthias Brugger, Daniel Kurtz, linux-mediatek, linaro-kernel,
linux-arm-kernel, linux-pm
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>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH 3/4] cpufreq: mt8173: check return value of regulator_get_voltage() call
@ 2015-11-30 5:27 ` Viresh Kumar
0 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-11-30 5:27 UTC (permalink / raw)
To: linux-arm-kernel
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>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
2015-11-29 8:31 ` Pi-Cheng Chen
@ 2015-12-05 6:01 ` Pi-Cheng Chen
-1 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-12-05 6:01 UTC (permalink / raw)
To: Viresh Kumar, Matthias Brugger
Cc: Daniel Kurtz, Ricky Liang, linux-mediatek, linaro-kernel,
linux-arm-kernel, linux-pm
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;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-05 6:01 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-12-05 6:01 UTC (permalink / raw)
To: linux-arm-kernel
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;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-05 6:01 ` Pi-Cheng Chen
@ 2015-12-06 2:24 ` Viresh Kumar
-1 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-12-06 2:24 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Matthias Brugger, Daniel Kurtz, Ricky Liang, linux-mediatek,
linaro-kernel, linux-arm-kernel, linux-pm
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>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-06 2:24 ` Viresh Kumar
0 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-12-06 2:24 UTC (permalink / raw)
To: linux-arm-kernel
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>
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-05 6:01 ` Pi-Cheng Chen
@ 2015-12-07 2:44 ` Daniel Kurtz
-1 siblings, 0 replies; 30+ messages in thread
From: Daniel Kurtz @ 2015-12-07 2:44 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Viresh Kumar, Matthias Brugger, Ricky Liang, linux-mediatek,
Linaro Kernel Mailman List, linux-arm-kernel, linux-pm
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
>
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-07 2:44 ` Daniel Kurtz
0 siblings, 0 replies; 30+ messages in thread
From: Daniel Kurtz @ 2015-12-07 2:44 UTC (permalink / raw)
To: linux-arm-kernel
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
>
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-07 2:44 ` Daniel Kurtz
@ 2015-12-07 12:42 ` Pi-Cheng Chen
-1 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-12-07 12:42 UTC (permalink / raw)
To: Daniel Kurtz
Cc: Viresh Kumar, Matthias Brugger, Ricky Liang,
moderated list:ARM/Mediatek SoC...,
Linaro Kernel Mailman List, linux-arm-kernel, linux-pm
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
>>
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-07 12:42 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-12-07 12:42 UTC (permalink / raw)
To: linux-arm-kernel
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
>>
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-07 12:42 ` Pi-Cheng Chen
@ 2015-12-07 16:03 ` Viresh Kumar
-1 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-12-07 16:03 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Daniel Kurtz, Matthias Brugger, Ricky Liang,
moderated list:ARM/Mediatek SoC...,
Linaro Kernel Mailman List, linux-arm-kernel, linux-pm
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.
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-07 16:03 ` Viresh Kumar
0 siblings, 0 replies; 30+ messages in thread
From: Viresh Kumar @ 2015-12-07 16:03 UTC (permalink / raw)
To: linux-arm-kernel
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.
--
viresh
^ permalink raw reply [flat|nested] 30+ messages in thread
* Re: [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-05 6:01 ` Pi-Cheng Chen
@ 2015-12-09 23:53 ` Rafael J. Wysocki
-1 siblings, 0 replies; 30+ messages in thread
From: Rafael J. Wysocki @ 2015-12-09 23:53 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Viresh Kumar, Matthias Brugger, Daniel Kurtz, Ricky Liang,
linux-mediatek, linaro-kernel, linux-arm-kernel, linux-pm
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
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v2] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-09 23:53 ` Rafael J. Wysocki
0 siblings, 0 replies; 30+ messages in thread
From: Rafael J. Wysocki @ 2015-12-09 23:53 UTC (permalink / raw)
To: linux-arm-kernel
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
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v3] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-09 23:53 ` Rafael J. Wysocki
@ 2015-12-10 3:48 ` Pi-Cheng Chen
-1 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-12-10 3:48 UTC (permalink / raw)
To: Rafael J. Wysocki, Viresh Kumar, Matthias Brugger
Cc: Daniel Kurtz, Ricky Liang, linux-mediatek, linaro-kernel,
linux-arm-kernel, linux-pm
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;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* [PATCH v3] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-10 3:48 ` Pi-Cheng Chen
0 siblings, 0 replies; 30+ messages in thread
From: Pi-Cheng Chen @ 2015-12-10 3:48 UTC (permalink / raw)
To: linux-arm-kernel
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;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 30+ messages in thread
* Re: [PATCH v3] cpufreq: mt8173: move resources allocation into ->probe()
2015-12-10 3:48 ` Pi-Cheng Chen
@ 2015-12-14 23:59 ` Rafael J. Wysocki
-1 siblings, 0 replies; 30+ messages in thread
From: Rafael J. Wysocki @ 2015-12-14 23:59 UTC (permalink / raw)
To: Pi-Cheng Chen
Cc: Viresh Kumar, Matthias Brugger, Daniel Kurtz, Ricky Liang,
linux-mediatek, linaro-kernel, linux-arm-kernel, linux-pm
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
^ permalink raw reply [flat|nested] 30+ messages in thread
* [PATCH v3] cpufreq: mt8173: move resources allocation into ->probe()
@ 2015-12-14 23:59 ` Rafael J. Wysocki
0 siblings, 0 replies; 30+ messages in thread
From: Rafael J. Wysocki @ 2015-12-14 23:59 UTC (permalink / raw)
To: linux-arm-kernel
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
^ permalink raw reply [flat|nested] 30+ messages in thread
end of thread, other threads:[~2015-12-14 23:59 UTC | newest]
Thread overview: 30+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-11-29 8:31 [PATCH 1/4] cpufreq: mt8173: add CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag Pi-Cheng Chen
2015-11-29 8:31 ` Pi-Cheng Chen
2015-11-29 8:31 ` [PATCH 2/4] cpufreq: mt8173: remove redundant regulator_get_voltage() call Pi-Cheng Chen
2015-11-29 8:31 ` Pi-Cheng Chen
2015-11-30 5:23 ` Viresh Kumar
2015-11-30 5:23 ` Viresh Kumar
2015-11-29 8:31 ` [PATCH 3/4] cpufreq: mt8173: check return value of " Pi-Cheng Chen
2015-11-29 8:31 ` Pi-Cheng Chen
2015-11-30 5:27 ` Viresh Kumar
2015-11-30 5:27 ` Viresh Kumar
2015-11-29 8:31 ` [PATCH 4/4] cpufreq: mt8173: move resources allocation into ->probe() Pi-Cheng Chen
2015-11-29 8:31 ` Pi-Cheng Chen
2015-12-05 6:01 ` [PATCH v2] " Pi-Cheng Chen
2015-12-05 6:01 ` Pi-Cheng Chen
2015-12-06 2:24 ` Viresh Kumar
2015-12-06 2:24 ` Viresh Kumar
2015-12-07 2:44 ` Daniel Kurtz
2015-12-07 2:44 ` Daniel Kurtz
2015-12-07 12:42 ` Pi-Cheng Chen
2015-12-07 12:42 ` Pi-Cheng Chen
2015-12-07 16:03 ` Viresh Kumar
2015-12-07 16:03 ` Viresh Kumar
2015-12-09 23:53 ` Rafael J. Wysocki
2015-12-09 23:53 ` Rafael J. Wysocki
2015-12-10 3:48 ` [PATCH v3] " Pi-Cheng Chen
2015-12-10 3:48 ` Pi-Cheng Chen
2015-12-14 23:59 ` Rafael J. Wysocki
2015-12-14 23:59 ` Rafael J. Wysocki
2015-11-30 5:20 ` [PATCH 1/4] cpufreq: mt8173: add CPUFREQ_HAVE_GOVERNOR_PER_POLICY flag Viresh Kumar
2015-11-30 5:20 ` Viresh Kumar
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.