counter: Add support for Intel Quadrature Encoder Peripheral
authorJarkko Nikula <jarkko.nikula@linux.intel.com>
Wed, 2 Jun 2021 11:32:59 +0000 (14:32 +0300)
committerJonathan Cameron <Jonathan.Cameron@huawei.com>
Thu, 3 Jun 2021 17:24:14 +0000 (18:24 +0100)
Add support for Intel Quadrature Encoder Peripheral found on Intel
Elkhart Lake platform.

Initial implementation was done by Felipe Balbi while he was working at
Intel with later changes from Raymond Tan and me.

Co-developed-by: Felipe Balbi (Intel) <balbi@kernel.org>
Signed-off-by: Felipe Balbi (Intel) <balbi@kernel.org>
Co-developed-by: Raymond Tan <raymond.tan@intel.com>
Signed-off-by: Raymond Tan <raymond.tan@intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Acked-by: William Breathitt Gray <vilhelm.gray@gmail.com>
Link: https://lore.kernel.org/r/20210602113259.158674-1-jarkko.nikula@linux.intel.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Documentation/ABI/testing/sysfs-bus-counter
MAINTAINERS
drivers/counter/Kconfig
drivers/counter/Makefile
drivers/counter/intel-qep.c [new file with mode: 0644]

index 566bd99fe0a53d6240610235de3ee852e1002964..e9d9e50f03bee4e234ca93d970b4c1bdefbf3f9a 100644 (file)
@@ -193,6 +193,15 @@ Description:
                both edges:
                        Any state transition.
 
+What:          /sys/bus/counter/devices/counterX/countY/spike_filter_ns
+KernelVersion: 5.14
+Contact:       linux-iio@vger.kernel.org
+Description:
+               If the counter device supports programmable spike filter this
+               attribute indicates the value in nanoseconds where noise pulses
+               shorter or equal to configured value are ignored. Value 0 means
+               filter is disabled.
+
 What:          /sys/bus/counter/devices/counterX/name
 KernelVersion: 5.2
 Contact:       linux-iio@vger.kernel.org
index fc771d2aacef281d6fb1b598c461a0597fd94b3a..e679d422b472a9ea8cc7f23f59eef3fabd1f09c8 100644 (file)
@@ -9364,6 +9364,11 @@ L:       linux-pm@vger.kernel.org
 S:     Supported
 F:     drivers/cpufreq/intel_pstate.c
 
+INTEL QUADRATURE ENCODER PERIPHERAL DRIVER
+M:     Jarkko Nikula <jarkko.nikula@linux.intel.com>
+L:     linux-iio@vger.kernel.org
+F:     drivers/counter/intel-qep.c
+
 INTEL RDMA RNIC DRIVER
 M:     Faisal Latif <faisal.latif@intel.com>
 M:     Shiraz Saleem <shiraz.saleem@intel.com>
index 5328705aa09c685b3c9a13d97d05e83d27d95d44..d5d2540b30c21e4c9a9b63f5275626146062e116 100644 (file)
@@ -91,4 +91,14 @@ config MICROCHIP_TCB_CAPTURE
          To compile this driver as a module, choose M here: the
          module will be called microchip-tcb-capture.
 
+config INTEL_QEP
+       tristate "Intel Quadrature Encoder Peripheral driver"
+       depends on PCI
+       help
+         Select this option to enable the Intel Quadrature Encoder Peripheral
+         driver.
+
+         To compile this driver as a module, choose M here: the module
+         will be called intel-qep.
+
 endif # COUNTER
index cb646ed2f0395c6c4d550c1b06032d2f38d06889..19742e6f5e3ebb25fbbd61be6ac44f39a870d2aa 100644 (file)
@@ -12,3 +12,4 @@ obj-$(CONFIG_STM32_LPTIMER_CNT)       += stm32-lptimer-cnt.o
 obj-$(CONFIG_TI_EQEP)          += ti-eqep.o
 obj-$(CONFIG_FTM_QUADDEC)      += ftm-quaddec.o
 obj-$(CONFIG_MICROCHIP_TCB_CAPTURE)    += microchip-tcb-capture.o
+obj-$(CONFIG_INTEL_QEP)                += intel-qep.o
diff --git a/drivers/counter/intel-qep.c b/drivers/counter/intel-qep.c
new file mode 100644 (file)
index 0000000..ab10ba3
--- /dev/null
@@ -0,0 +1,546 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Intel Quadrature Encoder Peripheral driver
+ *
+ * Copyright (C) 2019-2021 Intel Corporation
+ *
+ * Author: Felipe Balbi (Intel)
+ * Author: Jarkko Nikula <jarkko.nikula@linux.intel.com>
+ * Author: Raymond Tan <raymond.tan@intel.com>
+ */
+#include <linux/bitops.h>
+#include <linux/counter.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/pci.h>
+#include <linux/pm_runtime.h>
+
+#define INTEL_QEPCON                   0x00
+#define INTEL_QEPFLT                   0x04
+#define INTEL_QEPCOUNT                 0x08
+#define INTEL_QEPMAX                   0x0c
+#define INTEL_QEPWDT                   0x10
+#define INTEL_QEPCAPDIV                        0x14
+#define INTEL_QEPCNTR                  0x18
+#define INTEL_QEPCAPBUF                        0x1c
+#define INTEL_QEPINT_STAT              0x20
+#define INTEL_QEPINT_MASK              0x24
+
+/* QEPCON */
+#define INTEL_QEPCON_EN                        BIT(0)
+#define INTEL_QEPCON_FLT_EN            BIT(1)
+#define INTEL_QEPCON_EDGE_A            BIT(2)
+#define INTEL_QEPCON_EDGE_B            BIT(3)
+#define INTEL_QEPCON_EDGE_INDX         BIT(4)
+#define INTEL_QEPCON_SWPAB             BIT(5)
+#define INTEL_QEPCON_OP_MODE           BIT(6)
+#define INTEL_QEPCON_PH_ERR            BIT(7)
+#define INTEL_QEPCON_COUNT_RST_MODE    BIT(8)
+#define INTEL_QEPCON_INDX_GATING_MASK  GENMASK(10, 9)
+#define INTEL_QEPCON_INDX_GATING(n)    (((n) & 3) << 9)
+#define INTEL_QEPCON_INDX_PAL_PBL      INTEL_QEPCON_INDX_GATING(0)
+#define INTEL_QEPCON_INDX_PAL_PBH      INTEL_QEPCON_INDX_GATING(1)
+#define INTEL_QEPCON_INDX_PAH_PBL      INTEL_QEPCON_INDX_GATING(2)
+#define INTEL_QEPCON_INDX_PAH_PBH      INTEL_QEPCON_INDX_GATING(3)
+#define INTEL_QEPCON_CAP_MODE          BIT(11)
+#define INTEL_QEPCON_FIFO_THRE_MASK    GENMASK(14, 12)
+#define INTEL_QEPCON_FIFO_THRE(n)      ((((n) - 1) & 7) << 12)
+#define INTEL_QEPCON_FIFO_EMPTY                BIT(15)
+
+/* QEPFLT */
+#define INTEL_QEPFLT_MAX_COUNT(n)      ((n) & 0x1fffff)
+
+/* QEPINT */
+#define INTEL_QEPINT_FIFOCRIT          BIT(5)
+#define INTEL_QEPINT_FIFOENTRY         BIT(4)
+#define INTEL_QEPINT_QEPDIR            BIT(3)
+#define INTEL_QEPINT_QEPRST_UP         BIT(2)
+#define INTEL_QEPINT_QEPRST_DOWN       BIT(1)
+#define INTEL_QEPINT_WDT               BIT(0)
+
+#define INTEL_QEPINT_MASK_ALL          GENMASK(5, 0)
+
+#define INTEL_QEP_CLK_PERIOD_NS                10
+
+#define INTEL_QEP_COUNTER_EXT_RW(_name)                                \
+{                                                              \
+       .name = #_name,                                         \
+       .read = _name##_read,                                   \
+       .write = _name##_write,                                 \
+}
+
+struct intel_qep {
+       struct counter_device counter;
+       struct mutex lock;
+       struct device *dev;
+       void __iomem *regs;
+       bool enabled;
+       /* Context save registers */
+       u32 qepcon;
+       u32 qepflt;
+       u32 qepmax;
+};
+
+static inline u32 intel_qep_readl(struct intel_qep *qep, u32 offset)
+{
+       return readl(qep->regs + offset);
+}
+
+static inline void intel_qep_writel(struct intel_qep *qep,
+                                   u32 offset, u32 value)
+{
+       writel(value, qep->regs + offset);
+}
+
+static void intel_qep_init(struct intel_qep *qep)
+{
+       u32 reg;
+
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+       reg &= ~INTEL_QEPCON_EN;
+       intel_qep_writel(qep, INTEL_QEPCON, reg);
+       qep->enabled = false;
+       /*
+        * Make sure peripheral is disabled by flushing the write with
+        * a dummy read
+        */
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+
+       reg &= ~(INTEL_QEPCON_OP_MODE | INTEL_QEPCON_FLT_EN);
+       reg |= INTEL_QEPCON_EDGE_A | INTEL_QEPCON_EDGE_B |
+              INTEL_QEPCON_EDGE_INDX | INTEL_QEPCON_COUNT_RST_MODE;
+       intel_qep_writel(qep, INTEL_QEPCON, reg);
+       intel_qep_writel(qep, INTEL_QEPINT_MASK, INTEL_QEPINT_MASK_ALL);
+}
+
+static int intel_qep_count_read(struct counter_device *counter,
+                               struct counter_count *count,
+                               unsigned long *val)
+{
+       struct intel_qep *const qep = counter->priv;
+
+       pm_runtime_get_sync(qep->dev);
+       *val = intel_qep_readl(qep, INTEL_QEPCOUNT);
+       pm_runtime_put(qep->dev);
+
+       return 0;
+}
+
+static const enum counter_count_function intel_qep_count_functions[] = {
+       COUNTER_COUNT_FUNCTION_QUADRATURE_X4,
+};
+
+static int intel_qep_function_get(struct counter_device *counter,
+                                 struct counter_count *count,
+                                 size_t *function)
+{
+       *function = 0;
+
+       return 0;
+}
+
+static const enum counter_synapse_action intel_qep_synapse_actions[] = {
+       COUNTER_SYNAPSE_ACTION_BOTH_EDGES,
+};
+
+static int intel_qep_action_get(struct counter_device *counter,
+                               struct counter_count *count,
+                               struct counter_synapse *synapse,
+                               size_t *action)
+{
+       *action = 0;
+       return 0;
+}
+
+static const struct counter_ops intel_qep_counter_ops = {
+       .count_read = intel_qep_count_read,
+       .function_get = intel_qep_function_get,
+       .action_get = intel_qep_action_get,
+};
+
+#define INTEL_QEP_SIGNAL(_id, _name) {                         \
+       .id = (_id),                                            \
+       .name = (_name),                                        \
+}
+
+static struct counter_signal intel_qep_signals[] = {
+       INTEL_QEP_SIGNAL(0, "Phase A"),
+       INTEL_QEP_SIGNAL(1, "Phase B"),
+       INTEL_QEP_SIGNAL(2, "Index"),
+};
+
+#define INTEL_QEP_SYNAPSE(_signal_id) {                                \
+       .actions_list = intel_qep_synapse_actions,              \
+       .num_actions = ARRAY_SIZE(intel_qep_synapse_actions),   \
+       .signal = &intel_qep_signals[(_signal_id)],             \
+}
+
+static struct counter_synapse intel_qep_count_synapses[] = {
+       INTEL_QEP_SYNAPSE(0),
+       INTEL_QEP_SYNAPSE(1),
+       INTEL_QEP_SYNAPSE(2),
+};
+
+static ssize_t ceiling_read(struct counter_device *counter,
+                           struct counter_count *count,
+                           void *priv, char *buf)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 reg;
+
+       pm_runtime_get_sync(qep->dev);
+       reg = intel_qep_readl(qep, INTEL_QEPMAX);
+       pm_runtime_put(qep->dev);
+
+       return sysfs_emit(buf, "%u\n", reg);
+}
+
+static ssize_t ceiling_write(struct counter_device *counter,
+                            struct counter_count *count,
+                            void *priv, const char *buf, size_t len)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 max;
+       int ret;
+
+       ret = kstrtou32(buf, 0, &max);
+       if (ret < 0)
+               return ret;
+
+       mutex_lock(&qep->lock);
+       if (qep->enabled) {
+               ret = -EBUSY;
+               goto out;
+       }
+
+       pm_runtime_get_sync(qep->dev);
+       intel_qep_writel(qep, INTEL_QEPMAX, max);
+       pm_runtime_put(qep->dev);
+       ret = len;
+
+out:
+       mutex_unlock(&qep->lock);
+       return ret;
+}
+
+static ssize_t enable_read(struct counter_device *counter,
+                          struct counter_count *count,
+                          void *priv, char *buf)
+{
+       struct intel_qep *qep = counter->priv;
+
+       return sysfs_emit(buf, "%u\n", qep->enabled);
+}
+
+static ssize_t enable_write(struct counter_device *counter,
+                           struct counter_count *count,
+                           void *priv, const char *buf, size_t len)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 reg;
+       bool val, changed;
+       int ret;
+
+       ret = kstrtobool(buf, &val);
+       if (ret)
+               return ret;
+
+       mutex_lock(&qep->lock);
+       changed = val ^ qep->enabled;
+       if (!changed)
+               goto out;
+
+       pm_runtime_get_sync(qep->dev);
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+       if (val) {
+               /* Enable peripheral and keep runtime PM always on */
+               reg |= INTEL_QEPCON_EN;
+               pm_runtime_get_noresume(qep->dev);
+       } else {
+               /* Let runtime PM be idle and disable peripheral */
+               pm_runtime_put_noidle(qep->dev);
+               reg &= ~INTEL_QEPCON_EN;
+       }
+       intel_qep_writel(qep, INTEL_QEPCON, reg);
+       pm_runtime_put(qep->dev);
+       qep->enabled = val;
+
+out:
+       mutex_unlock(&qep->lock);
+       return len;
+}
+
+static ssize_t spike_filter_ns_read(struct counter_device *counter,
+                                   struct counter_count *count,
+                                   void *priv, char *buf)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 reg;
+
+       pm_runtime_get_sync(qep->dev);
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+       if (!(reg & INTEL_QEPCON_FLT_EN)) {
+               pm_runtime_put(qep->dev);
+               return sysfs_emit(buf, "0\n");
+       }
+       reg = INTEL_QEPFLT_MAX_COUNT(intel_qep_readl(qep, INTEL_QEPFLT));
+       pm_runtime_put(qep->dev);
+
+       return sysfs_emit(buf, "%u\n", (reg + 2) * INTEL_QEP_CLK_PERIOD_NS);
+}
+
+static ssize_t spike_filter_ns_write(struct counter_device *counter,
+                                    struct counter_count *count,
+                                    void *priv, const char *buf, size_t len)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 reg, length;
+       bool enable;
+       int ret;
+
+       ret = kstrtou32(buf, 0, &length);
+       if (ret < 0)
+               return ret;
+
+       /*
+        * Spike filter length is (MAX_COUNT + 2) clock periods.
+        * Disable filter when userspace writes 0, enable for valid
+        * nanoseconds values and error out otherwise.
+        */
+       length /= INTEL_QEP_CLK_PERIOD_NS;
+       if (length == 0) {
+               enable = false;
+               length = 0;
+       } else if (length >= 2) {
+               enable = true;
+               length -= 2;
+       } else {
+               return -EINVAL;
+       }
+
+       if (length > INTEL_QEPFLT_MAX_COUNT(length))
+               return -EINVAL;
+
+       mutex_lock(&qep->lock);
+       if (qep->enabled) {
+               ret = -EBUSY;
+               goto out;
+       }
+
+       pm_runtime_get_sync(qep->dev);
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+       if (enable)
+               reg |= INTEL_QEPCON_FLT_EN;
+       else
+               reg &= ~INTEL_QEPCON_FLT_EN;
+       intel_qep_writel(qep, INTEL_QEPFLT, length);
+       intel_qep_writel(qep, INTEL_QEPCON, reg);
+       pm_runtime_put(qep->dev);
+       ret = len;
+
+out:
+       mutex_unlock(&qep->lock);
+       return ret;
+}
+
+static ssize_t preset_enable_read(struct counter_device *counter,
+                                 struct counter_count *count,
+                                 void *priv, char *buf)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 reg;
+
+       pm_runtime_get_sync(qep->dev);
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+       pm_runtime_put(qep->dev);
+       return sysfs_emit(buf, "%u\n", !(reg & INTEL_QEPCON_COUNT_RST_MODE));
+}
+
+static ssize_t preset_enable_write(struct counter_device *counter,
+                                  struct counter_count *count,
+                                  void *priv, const char *buf, size_t len)
+{
+       struct intel_qep *qep = counter->priv;
+       u32 reg;
+       bool val;
+       int ret;
+
+       ret = kstrtobool(buf, &val);
+       if (ret)
+               return ret;
+
+       mutex_lock(&qep->lock);
+       if (qep->enabled) {
+               ret = -EBUSY;
+               goto out;
+       }
+
+       pm_runtime_get_sync(qep->dev);
+       reg = intel_qep_readl(qep, INTEL_QEPCON);
+       if (val)
+               reg &= ~INTEL_QEPCON_COUNT_RST_MODE;
+       else
+               reg |= INTEL_QEPCON_COUNT_RST_MODE;
+
+       intel_qep_writel(qep, INTEL_QEPCON, reg);
+       pm_runtime_put(qep->dev);
+       ret = len;
+
+out:
+       mutex_unlock(&qep->lock);
+
+       return ret;
+}
+
+static const struct counter_count_ext intel_qep_count_ext[] = {
+       INTEL_QEP_COUNTER_EXT_RW(ceiling),
+       INTEL_QEP_COUNTER_EXT_RW(enable),
+       INTEL_QEP_COUNTER_EXT_RW(spike_filter_ns),
+       INTEL_QEP_COUNTER_EXT_RW(preset_enable)
+};
+
+static struct counter_count intel_qep_counter_count[] = {
+       {
+               .id = 0,
+               .name = "Channel 1 Count",
+               .functions_list = intel_qep_count_functions,
+               .num_functions = ARRAY_SIZE(intel_qep_count_functions),
+               .synapses = intel_qep_count_synapses,
+               .num_synapses = ARRAY_SIZE(intel_qep_count_synapses),
+               .ext = intel_qep_count_ext,
+               .num_ext = ARRAY_SIZE(intel_qep_count_ext),
+       },
+};
+
+static int intel_qep_probe(struct pci_dev *pci, const struct pci_device_id *id)
+{
+       struct intel_qep *qep;
+       struct device *dev = &pci->dev;
+       void __iomem *regs;
+       int ret;
+
+       qep = devm_kzalloc(dev, sizeof(*qep), GFP_KERNEL);
+       if (!qep)
+               return -ENOMEM;
+
+       ret = pcim_enable_device(pci);
+       if (ret)
+               return ret;
+
+       pci_set_master(pci);
+
+       ret = pcim_iomap_regions(pci, BIT(0), pci_name(pci));
+       if (ret)
+               return ret;
+
+       regs = pcim_iomap_table(pci)[0];
+       if (!regs)
+               return -ENOMEM;
+
+       qep->dev = dev;
+       qep->regs = regs;
+       mutex_init(&qep->lock);
+
+       intel_qep_init(qep);
+       pci_set_drvdata(pci, qep);
+
+       qep->counter.name = pci_name(pci);
+       qep->counter.parent = dev;
+       qep->counter.ops = &intel_qep_counter_ops;
+       qep->counter.counts = intel_qep_counter_count;
+       qep->counter.num_counts = ARRAY_SIZE(intel_qep_counter_count);
+       qep->counter.signals = intel_qep_signals;
+       qep->counter.num_signals = ARRAY_SIZE(intel_qep_signals);
+       qep->counter.priv = qep;
+       qep->enabled = false;
+
+       pm_runtime_put(dev);
+       pm_runtime_allow(dev);
+
+       return devm_counter_register(&pci->dev, &qep->counter);
+}
+
+static void intel_qep_remove(struct pci_dev *pci)
+{
+       struct intel_qep *qep = pci_get_drvdata(pci);
+       struct device *dev = &pci->dev;
+
+       pm_runtime_forbid(dev);
+       if (!qep->enabled)
+               pm_runtime_get(dev);
+
+       intel_qep_writel(qep, INTEL_QEPCON, 0);
+}
+
+#ifdef CONFIG_PM
+static int intel_qep_suspend(struct device *dev)
+{
+       struct pci_dev *pdev = container_of(dev, struct pci_dev, dev);
+       struct intel_qep *qep = pci_get_drvdata(pdev);
+
+       qep->qepcon = intel_qep_readl(qep, INTEL_QEPCON);
+       qep->qepflt = intel_qep_readl(qep, INTEL_QEPFLT);
+       qep->qepmax = intel_qep_readl(qep, INTEL_QEPMAX);
+
+       return 0;
+}
+
+static int intel_qep_resume(struct device *dev)
+{
+       struct pci_dev *pdev = container_of(dev, struct pci_dev, dev);
+       struct intel_qep *qep = pci_get_drvdata(pdev);
+
+       /*
+        * Make sure peripheral is disabled when restoring registers and
+        * control register bits that are writable only when the peripheral
+        * is disabled
+        */
+       intel_qep_writel(qep, INTEL_QEPCON, 0);
+       intel_qep_readl(qep, INTEL_QEPCON);
+
+       intel_qep_writel(qep, INTEL_QEPFLT, qep->qepflt);
+       intel_qep_writel(qep, INTEL_QEPMAX, qep->qepmax);
+       intel_qep_writel(qep, INTEL_QEPINT_MASK, INTEL_QEPINT_MASK_ALL);
+
+       /* Restore all other control register bits except enable status */
+       intel_qep_writel(qep, INTEL_QEPCON, qep->qepcon & ~INTEL_QEPCON_EN);
+       intel_qep_readl(qep, INTEL_QEPCON);
+
+       /* Restore enable status */
+       intel_qep_writel(qep, INTEL_QEPCON, qep->qepcon);
+
+       return 0;
+}
+#endif
+
+static UNIVERSAL_DEV_PM_OPS(intel_qep_pm_ops,
+                           intel_qep_suspend, intel_qep_resume, NULL);
+
+static const struct pci_device_id intel_qep_id_table[] = {
+       /* EHL */
+       { PCI_VDEVICE(INTEL, 0x4bc3), },
+       { PCI_VDEVICE(INTEL, 0x4b81), },
+       { PCI_VDEVICE(INTEL, 0x4b82), },
+       { PCI_VDEVICE(INTEL, 0x4b83), },
+       {  } /* Terminating Entry */
+};
+MODULE_DEVICE_TABLE(pci, intel_qep_id_table);
+
+static struct pci_driver intel_qep_driver = {
+       .name = "intel-qep",
+       .id_table = intel_qep_id_table,
+       .probe = intel_qep_probe,
+       .remove = intel_qep_remove,
+       .driver = {
+               .pm = &intel_qep_pm_ops,
+       }
+};
+
+module_pci_driver(intel_qep_driver);
+
+MODULE_AUTHOR("Felipe Balbi (Intel)");
+MODULE_AUTHOR("Jarkko Nikula <jarkko.nikula@linux.intel.com>");
+MODULE_AUTHOR("Raymond Tan <raymond.tan@intel.com>");
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Intel Quadrature Encoder Peripheral driver");