On 6/28/21 10:35 PM, Viresh Kumar wrote:
On 24-06-21, 07:58, Thara Gopinath wrote:
Add interrupt support to notify the kernel of h/w initiated frequency
throttling by LMh. Convey this to scheduler via thermal presssure
interface.
Signed-off-by: Thara Gopinath <thara.gopinath@xxxxxxxxxx>
---
v1->v2:
- Introduced qcom_cpufreq_hw_lmh_init to consolidate LMh related initializations
as per Viresh's review comment.
- Moved the piece of code restarting polling/re-enabling LMh interrupt to
qcom_lmh_dcvs_notify therby simplifying isr and timer callback as per Viresh's
suggestion.
- Droped cpus from qcom_cpufreq_data and instead using cpus from cpufreq_policy in
qcom_lmh_dcvs_notify as per Viresh's review comment.
- Dropped dt property qcom,support-lmh as per Bjorn's suggestion.
- Other minor/cosmetic fixes
drivers/cpufreq/qcom-cpufreq-hw.c | 103 ++++++++++++++++++++++++++++++
1 file changed, 103 insertions(+)
diff --git a/drivers/cpufreq/qcom-cpufreq-hw.c b/drivers/cpufreq/qcom-cpufreq-hw.c
index f86859bf76f1..241f6f2b441f 100644
--- a/drivers/cpufreq/qcom-cpufreq-hw.c
+++ b/drivers/cpufreq/qcom-cpufreq-hw.c
@@ -13,6 +13,7 @@
#include <linux/of_platform.h>
#include <linux/pm_opp.h>
#include <linux/slab.h>
+#include <linux/interrupt.h>
Please don't break the alphabetical order here.
#define LUT_MAX_ENTRIES 40U
#define LUT_SRC GENMASK(31, 30)
@@ -22,10 +23,13 @@
#define CLK_HW_DIV 2
#define LUT_TURBO_IND 1
+#define HZ_PER_KHZ 1000
struct qcom_cpufreq_soc_data {
u32 reg_enable;
u32 reg_freq_lut;
u32 reg_volt_lut;
+ u32 reg_current_vote;
u32 reg_perf_state;
u8 lut_row_size;
};
@@ -33,7 +37,10 @@ struct qcom_cpufreq_soc_data {
struct qcom_cpufreq_data {
void __iomem *base;
struct resource *res;
+ struct delayed_work lmh_dcvs_poll_work;
const struct qcom_cpufreq_soc_data *soc_data;
+ struct cpufreq_policy *policy;
+ int lmh_dcvs_irq;
};
static unsigned long cpu_hw_rate, xo_rate;
@@ -251,10 +258,79 @@ static void qcom_get_related_cpus(int index, struct cpumask *m)
}
}
+static inline unsigned long qcom_lmh_vote_to_freq(u32 val)
+{
+ return (val & 0x3FF) * 19200;
+}
+
+static void qcom_lmh_dcvs_notify(struct qcom_cpufreq_data *data)
+{
+ struct cpufreq_policy *policy = data->policy;
+ struct dev_pm_opp *opp;
+ struct device *dev;
+ unsigned long max_capacity, capacity, freq_hz, throttled_freq;
+ unsigned int val, freq;
+
+ /*
+ * Get the h/w throttled frequency, normalize it using the
+ * registered opp table and use it to calculate thermal pressure.
+ */
+ val = readl_relaxed(data->base + data->soc_data->reg_current_vote);
+ freq = qcom_lmh_vote_to_freq(val);
+ freq_hz = freq * HZ_PER_KHZ;
+
+ dev = get_cpu_device(cpumask_first(policy->cpus));
+ opp = dev_pm_opp_find_freq_floor(dev, &freq_hz);
+ if (IS_ERR(opp) && PTR_ERR(opp) == -ERANGE)
+ opp = dev_pm_opp_find_freq_ceil(dev, &freq_hz);
+
+ throttled_freq = freq_hz / HZ_PER_KHZ;
+
+ /* Update thermal pressure */
+ max_capacity = arch_scale_cpu_capacity(cpumask_first(policy->cpus));
+ capacity = throttled_freq * max_capacity;
+ capacity /= policy->cpuinfo.max_freq;
+ /* Don't pass boost capacity to scheduler */
+ if (capacity > max_capacity)
+ capacity = max_capacity;
I wonder why this check isn't present for cpufreq_cooling.c .
Hi Viresh,
I don't think cpufreq_cooling recognizes boost frequencies. The max
state there is the max of nominal frequencies , right? If not, it might
be a good idea to add this check there as well.
I will fix rest of your comments in v3.
--
Warm Regards
Thara (She/Her/Hers)
+ arch_set_thermal_pressure(policy->cpus, max_capacity - capacity);
+ /*
Whenever you mix code and comments, please separate them with a blank
line, else it becomes a bit messy and harder to read.
+ * If h/w throttled frequency is higher than what cpufreq has requested for, stop
+ * polling and switch back to interrupt mechanism
+ */
+ if (throttled_freq >= qcom_cpufreq_hw_get(cpumask_first(policy->cpus)))
+ /* Clear the existing interrupts and enable it back */
+ enable_irq(data->lmh_dcvs_irq);
+ else
+ mod_delayed_work(system_highpri_wq, &data->lmh_dcvs_poll_work,
+ msecs_to_jiffies(10));
+}
+
+static void qcom_lmh_dcvs_poll(struct work_struct *work)
+{
+ struct qcom_cpufreq_data *data;
+
+ data = container_of(work, struct qcom_cpufreq_data, lmh_dcvs_poll_work.work);
+
+ qcom_lmh_dcvs_notify(data);
+}
+
+static irqreturn_t qcom_lmh_dcvs_handle_irq(int irq, void *data)
+{
+ struct qcom_cpufreq_data *c_data = data;
+
+ /* Disable interrupt and enable polling */
+ disable_irq_nosync(c_data->lmh_dcvs_irq);
+ qcom_lmh_dcvs_notify(c_data);
+
+ return 0;
+}
+
static const struct qcom_cpufreq_soc_data qcom_soc_data = {
.reg_enable = 0x0,
.reg_freq_lut = 0x110,
.reg_volt_lut = 0x114,
+ .reg_current_vote = 0x704,
.reg_perf_state = 0x920,
.lut_row_size = 32,
};
@@ -274,6 +350,23 @@ static const struct of_device_id qcom_cpufreq_hw_match[] = {
};
MODULE_DEVICE_TABLE(of, qcom_cpufreq_hw_match);
+static void qcom_cpufreq_hw_lmh_init(struct cpufreq_policy *policy)
+{
+ struct qcom_cpufreq_data *data = policy->driver_data;
+ struct platform_device *pdev = cpufreq_get_driver_data();
+ struct device *dev = &pdev->dev;
+ int ret;
+
+ ret = devm_request_irq(dev, data->lmh_dcvs_irq, qcom_lmh_dcvs_handle_irq,
+ 0, "dcvsh-irq", data);
+ if (ret) {
+ dev_err(dev, "Error %d registering irq %x\n", ret, data->lmh_dcvs_irq);
+ return;
+ }
+ data->policy = policy;
+ INIT_DEFERRABLE_WORK(&data->lmh_dcvs_poll_work, qcom_lmh_dcvs_poll);
+}
+
static int qcom_cpufreq_hw_cpu_init(struct cpufreq_policy *policy)
{
struct platform_device *pdev = cpufreq_get_driver_data();
@@ -370,6 +463,16 @@ static int qcom_cpufreq_hw_cpu_init(struct cpufreq_policy *policy)
dev_warn(cpu_dev, "failed to enable boost: %d\n", ret);
}
+ /* Look for LMh interrupt. If no interrupt line is specified /
+ * if there is an error, allow cpufreq to be enabled as usual.
+ */
Proper comment style please..
+ data->lmh_dcvs_irq = platform_get_irq(pdev, index);
+ if (data->lmh_dcvs_irq > 0) {
+ qcom_cpufreq_hw_lmh_init(policy);
+ } else if (data->lmh_dcvs_irq != -ENXIO) {
+ ret = data->lmh_dcvs_irq;
+ goto error;
+ }
Move all of this to qcom_cpufreq_hw_lmh_init().
And I don't see any cleanup for this stuff. There is no guarantee that
the irq won't fire and queue up a work right after cpufreq driver is
unregistered and before the devm_ stuff gets released.
return 0;
error:
kfree(data);
--
2.25.1
