Merge branches 'thermal-core', 'thermal-soc' and 'thermal-int340x' of .git into next
authorZhang Rui <rui.zhang@intel.com>
Wed, 24 Dec 2014 02:38:30 +0000 (10:38 +0800)
committerZhang Rui <rui.zhang@intel.com>
Wed, 24 Dec 2014 02:38:30 +0000 (10:38 +0800)
drivers/thermal/cpu_cooling.c
drivers/thermal/int340x_thermal/Makefile
drivers/thermal/int340x_thermal/acpi_thermal_rel.c
drivers/thermal/int340x_thermal/int3400_thermal.c
drivers/thermal/int340x_thermal/int3402_thermal.c
drivers/thermal/int340x_thermal/int3403_thermal.c
drivers/thermal/int340x_thermal/processor_thermal_device.c [new file with mode: 0644]

index 9b45f64..f65f0d1 100644 (file)
@@ -57,6 +57,7 @@
  * @max_level: maximum cooling level. One less than total number of valid
  *     cpufreq frequencies.
  * @allowed_cpus: all the cpus involved for this cpufreq_cooling_device.
+ * @node: list_head to link all cpufreq_cooling_device together.
  *
  * This structure is required for keeping information of each registered
  * cpufreq_cooling_device.
index ffe40bf..d441369 100644 (file)
@@ -1,4 +1,5 @@
 obj-$(CONFIG_INT340X_THERMAL)  += int3400_thermal.o
 obj-$(CONFIG_INT340X_THERMAL)  += int3402_thermal.o
 obj-$(CONFIG_INT340X_THERMAL)  += int3403_thermal.o
+obj-$(CONFIG_INT340X_THERMAL)  += processor_thermal_device.o
 obj-$(CONFIG_ACPI_THERMAL_REL) += acpi_thermal_rel.o
index e4e61b3..231cabc 100644 (file)
@@ -82,7 +82,7 @@ int acpi_parse_trt(acpi_handle handle, int *trt_count, struct trt **trtp,
        struct acpi_buffer trt_format = { sizeof("RRNNNNNN"), "RRNNNNNN" };
 
        if (!acpi_has_method(handle, "_TRT"))
-               return 0;
+               return -ENODEV;
 
        status = acpi_evaluate_object(handle, "_TRT", NULL, &buffer);
        if (ACPI_FAILURE(status))
@@ -167,7 +167,7 @@ int acpi_parse_art(acpi_handle handle, int *art_count, struct art **artp,
                sizeof("RRNNNNNNNNNNN"), "RRNNNNNNNNNNN" };
 
        if (!acpi_has_method(handle, "_ART"))
-               return 0;
+               return -ENODEV;
 
        status = acpi_evaluate_object(handle, "_ART", NULL, &buffer);
        if (ACPI_FAILURE(status))
@@ -321,8 +321,8 @@ static long acpi_thermal_rel_ioctl(struct file *f, unsigned int cmd,
        unsigned long length = 0;
        int count = 0;
        char __user *arg = (void __user *)__arg;
-       struct trt *trts;
-       struct art *arts;
+       struct trt *trts = NULL;
+       struct art *arts = NULL;
 
        switch (cmd) {
        case ACPI_THERMAL_GET_TRT_COUNT:
index dcb306e..65a98a9 100644 (file)
@@ -335,7 +335,6 @@ static struct platform_driver int3400_thermal_driver = {
        .remove = int3400_thermal_remove,
        .driver = {
                   .name = "int3400 thermal",
-                  .owner = THIS_MODULE,
                   .acpi_match_table = ACPI_PTR(int3400_thermal_match),
                   },
 };
index a5d08c1..c5cbc3a 100644 (file)
@@ -231,7 +231,6 @@ static struct platform_driver int3402_thermal_driver = {
        .remove = int3402_thermal_remove,
        .driver = {
                   .name = "int3402 thermal",
-                  .owner = THIS_MODULE,
                   .acpi_match_table = int3402_thermal_match,
                   },
 };
index 1bfa6a6..0faf500 100644 (file)
@@ -301,6 +301,8 @@ static int int3403_sensor_remove(struct int3403_priv *priv)
 {
        struct int3403_sensor *obj = priv->priv;
 
+       acpi_remove_notify_handler(priv->adev->handle,
+                                  ACPI_DEVICE_NOTIFY, int3403_notify);
        thermal_zone_device_unregister(obj->tzone);
        return 0;
 }
@@ -369,6 +371,7 @@ static int int3403_cdev_add(struct int3403_priv *priv)
        p = buf.pointer;
        if (!p || (p->type != ACPI_TYPE_PACKAGE)) {
                printk(KERN_WARNING "Invalid PPSS data\n");
+               kfree(buf.pointer);
                return -EFAULT;
        }
 
@@ -381,6 +384,7 @@ static int int3403_cdev_add(struct int3403_priv *priv)
 
        priv->priv = obj;
 
+       kfree(buf.pointer);
        /* TODO: add ACPI notification support */
 
        return result;
diff --git a/drivers/thermal/int340x_thermal/processor_thermal_device.c b/drivers/thermal/int340x_thermal/processor_thermal_device.c
new file mode 100644 (file)
index 0000000..31bb553
--- /dev/null
@@ -0,0 +1,309 @@
+/*
+ * processor_thermal_device.c
+ * Copyright (c) 2014, Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/acpi.h>
+
+/* Broadwell-U/HSB thermal reporting device */
+#define PCI_DEVICE_ID_PROC_BDW_THERMAL 0x1603
+#define PCI_DEVICE_ID_PROC_HSB_THERMAL 0x0A03
+
+/* Braswell thermal reporting device */
+#define PCI_DEVICE_ID_PROC_BSW_THERMAL 0x22DC
+
+struct power_config {
+       u32     index;
+       u32     min_uw;
+       u32     max_uw;
+       u32     tmin_us;
+       u32     tmax_us;
+       u32     step_uw;
+};
+
+struct proc_thermal_device {
+       struct device *dev;
+       struct acpi_device *adev;
+       struct power_config power_limits[2];
+};
+
+enum proc_thermal_emum_mode_type {
+       PROC_THERMAL_NONE,
+       PROC_THERMAL_PCI,
+       PROC_THERMAL_PLATFORM_DEV
+};
+
+/*
+ * We can have only one type of enumeration, PCI or Platform,
+ * not both. So we don't need instance specific data.
+ */
+static enum proc_thermal_emum_mode_type proc_thermal_emum_mode =
+                                                       PROC_THERMAL_NONE;
+
+#define POWER_LIMIT_SHOW(index, suffix) \
+static ssize_t power_limit_##index##_##suffix##_show(struct device *dev, \
+                                       struct device_attribute *attr, \
+                                       char *buf) \
+{ \
+       struct pci_dev *pci_dev; \
+       struct platform_device *pdev; \
+       struct proc_thermal_device *proc_dev; \
+\
+       if (proc_thermal_emum_mode == PROC_THERMAL_PLATFORM_DEV) { \
+               pdev = to_platform_device(dev); \
+               proc_dev = platform_get_drvdata(pdev); \
+       } else { \
+               pci_dev = to_pci_dev(dev); \
+               proc_dev = pci_get_drvdata(pci_dev); \
+       } \
+       return sprintf(buf, "%lu\n",\
+       (unsigned long)proc_dev->power_limits[index].suffix * 1000); \
+}
+
+POWER_LIMIT_SHOW(0, min_uw)
+POWER_LIMIT_SHOW(0, max_uw)
+POWER_LIMIT_SHOW(0, step_uw)
+POWER_LIMIT_SHOW(0, tmin_us)
+POWER_LIMIT_SHOW(0, tmax_us)
+
+POWER_LIMIT_SHOW(1, min_uw)
+POWER_LIMIT_SHOW(1, max_uw)
+POWER_LIMIT_SHOW(1, step_uw)
+POWER_LIMIT_SHOW(1, tmin_us)
+POWER_LIMIT_SHOW(1, tmax_us)
+
+static DEVICE_ATTR_RO(power_limit_0_min_uw);
+static DEVICE_ATTR_RO(power_limit_0_max_uw);
+static DEVICE_ATTR_RO(power_limit_0_step_uw);
+static DEVICE_ATTR_RO(power_limit_0_tmin_us);
+static DEVICE_ATTR_RO(power_limit_0_tmax_us);
+
+static DEVICE_ATTR_RO(power_limit_1_min_uw);
+static DEVICE_ATTR_RO(power_limit_1_max_uw);
+static DEVICE_ATTR_RO(power_limit_1_step_uw);
+static DEVICE_ATTR_RO(power_limit_1_tmin_us);
+static DEVICE_ATTR_RO(power_limit_1_tmax_us);
+
+static struct attribute *power_limit_attrs[] = {
+       &dev_attr_power_limit_0_min_uw.attr,
+       &dev_attr_power_limit_1_min_uw.attr,
+       &dev_attr_power_limit_0_max_uw.attr,
+       &dev_attr_power_limit_1_max_uw.attr,
+       &dev_attr_power_limit_0_step_uw.attr,
+       &dev_attr_power_limit_1_step_uw.attr,
+       &dev_attr_power_limit_0_tmin_us.attr,
+       &dev_attr_power_limit_1_tmin_us.attr,
+       &dev_attr_power_limit_0_tmax_us.attr,
+       &dev_attr_power_limit_1_tmax_us.attr,
+       NULL
+};
+
+static struct attribute_group power_limit_attribute_group = {
+       .attrs = power_limit_attrs,
+       .name = "power_limits"
+};
+
+static int proc_thermal_add(struct device *dev,
+                           struct proc_thermal_device **priv)
+{
+       struct proc_thermal_device *proc_priv;
+       struct acpi_device *adev;
+       acpi_status status;
+       struct acpi_buffer buf = { ACPI_ALLOCATE_BUFFER, NULL };
+       union acpi_object *elements, *ppcc;
+       union acpi_object *p;
+       int i;
+       int ret;
+
+       adev = ACPI_COMPANION(dev);
+
+       status = acpi_evaluate_object(adev->handle, "PPCC", NULL, &buf);
+       if (ACPI_FAILURE(status))
+               return -ENODEV;
+
+       p = buf.pointer;
+       if (!p || (p->type != ACPI_TYPE_PACKAGE)) {
+               dev_err(dev, "Invalid PPCC data\n");
+               ret = -EFAULT;
+               goto free_buffer;
+       }
+       if (!p->package.count) {
+               dev_err(dev, "Invalid PPCC package size\n");
+               ret = -EFAULT;
+               goto free_buffer;
+       }
+
+       proc_priv = devm_kzalloc(dev, sizeof(*proc_priv), GFP_KERNEL);
+       if (!proc_priv) {
+               ret = -ENOMEM;
+               goto free_buffer;
+       }
+
+       proc_priv->dev = dev;
+       proc_priv->adev = adev;
+
+       for (i = 0; i < min((int)p->package.count - 1, 2); ++i) {
+               elements = &(p->package.elements[i+1]);
+               if (elements->type != ACPI_TYPE_PACKAGE ||
+                   elements->package.count != 6) {
+                       ret = -EFAULT;
+                       goto free_buffer;
+               }
+               ppcc = elements->package.elements;
+               proc_priv->power_limits[i].index = ppcc[0].integer.value;
+               proc_priv->power_limits[i].min_uw = ppcc[1].integer.value;
+               proc_priv->power_limits[i].max_uw = ppcc[2].integer.value;
+               proc_priv->power_limits[i].tmin_us = ppcc[3].integer.value;
+               proc_priv->power_limits[i].tmax_us = ppcc[4].integer.value;
+               proc_priv->power_limits[i].step_uw = ppcc[5].integer.value;
+       }
+
+       *priv = proc_priv;
+
+       ret = sysfs_create_group(&dev->kobj,
+                                &power_limit_attribute_group);
+
+free_buffer:
+       kfree(buf.pointer);
+
+       return ret;
+}
+
+void proc_thermal_remove(struct proc_thermal_device *proc_priv)
+{
+       sysfs_remove_group(&proc_priv->dev->kobj,
+                          &power_limit_attribute_group);
+}
+
+static int int3401_add(struct platform_device *pdev)
+{
+       struct proc_thermal_device *proc_priv;
+       int ret;
+
+       if (proc_thermal_emum_mode == PROC_THERMAL_PCI) {
+               dev_err(&pdev->dev, "error: enumerated as PCI dev\n");
+               return -ENODEV;
+       }
+
+       ret = proc_thermal_add(&pdev->dev, &proc_priv);
+       if (ret)
+               return ret;
+
+       platform_set_drvdata(pdev, proc_priv);
+       proc_thermal_emum_mode = PROC_THERMAL_PLATFORM_DEV;
+
+       return 0;
+}
+
+static int int3401_remove(struct platform_device *pdev)
+{
+       proc_thermal_remove(platform_get_drvdata(pdev));
+
+       return 0;
+}
+
+static int  proc_thermal_pci_probe(struct pci_dev *pdev,
+                                  const struct pci_device_id *unused)
+{
+       struct proc_thermal_device *proc_priv;
+       int ret;
+
+       if (proc_thermal_emum_mode == PROC_THERMAL_PLATFORM_DEV) {
+               dev_err(&pdev->dev, "error: enumerated as platform dev\n");
+               return -ENODEV;
+       }
+
+       ret = pci_enable_device(pdev);
+       if (ret < 0) {
+               dev_err(&pdev->dev, "error: could not enable device\n");
+               return ret;
+       }
+
+       ret = proc_thermal_add(&pdev->dev, &proc_priv);
+       if (ret) {
+               pci_disable_device(pdev);
+               return ret;
+       }
+
+       pci_set_drvdata(pdev, proc_priv);
+       proc_thermal_emum_mode = PROC_THERMAL_PCI;
+
+       return 0;
+}
+
+static void  proc_thermal_pci_remove(struct pci_dev *pdev)
+{
+       proc_thermal_remove(pci_get_drvdata(pdev));
+       pci_disable_device(pdev);
+}
+
+static const struct pci_device_id proc_thermal_pci_ids[] = {
+       { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_PROC_BDW_THERMAL)},
+       { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_PROC_HSB_THERMAL)},
+       { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_PROC_BSW_THERMAL)},
+       { 0, },
+};
+
+MODULE_DEVICE_TABLE(pci, proc_thermal_pci_ids);
+
+static struct pci_driver proc_thermal_pci_driver = {
+       .name           = "proc_thermal",
+       .probe          = proc_thermal_pci_probe,
+       .remove         = proc_thermal_pci_remove,
+       .id_table       = proc_thermal_pci_ids,
+};
+
+static const struct acpi_device_id int3401_device_ids[] = {
+       {"INT3401", 0},
+       {"", 0},
+};
+MODULE_DEVICE_TABLE(acpi, int3401_device_ids);
+
+static struct platform_driver int3401_driver = {
+       .probe = int3401_add,
+       .remove = int3401_remove,
+       .driver = {
+               .name = "int3401 thermal",
+               .acpi_match_table = int3401_device_ids,
+       },
+};
+
+static int __init proc_thermal_init(void)
+{
+       int ret;
+
+       ret = platform_driver_register(&int3401_driver);
+       if (ret)
+               return ret;
+
+       ret = pci_register_driver(&proc_thermal_pci_driver);
+
+       return ret;
+}
+
+static void __exit proc_thermal_exit(void)
+{
+       platform_driver_unregister(&int3401_driver);
+       pci_unregister_driver(&proc_thermal_pci_driver);
+}
+
+module_init(proc_thermal_init);
+module_exit(proc_thermal_exit);
+
+MODULE_AUTHOR("Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>");
+MODULE_DESCRIPTION("Processor Thermal Reporting Device Driver");
+MODULE_LICENSE("GPL v2");