Thread (17 messages) 17 messages, 4 authors, 2021-06-30

Re: [Patch v2 3/5] cpufreq: qcom-cpufreq-hw: Add dcvs interrupt support

From: Thara Gopinath <hidden>
Date: 2021-06-30 02:27:22
Also in: linux-arm-msm, linux-pm, lkml


On 6/28/21 10:50 PM, Taniya Das wrote:

On 6/24/2021 5:28 PM, Thara Gopinath wrote:
quoted
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 <redacted>
---

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
[snip]
quoted
  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);

It is better if you tag the CPU id while registering the IRQ.
"dcvsh-irq-x" (0/4/7)
Sure. Will fix it.
quoted
+    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.
+     */
+    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;
+    }
      return 0;
  error:
      kfree(data);
-- 
Warm Regards
Thara (She/Her/Hers)
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help