Re: [PATCH v2 7/8] [PATCH 7/8] drivers/hwmon: Add a generic PECI hwmon client driver

From: Jae Hyun Yoo
Date: Wed Feb 21 2018 - 18:08:08 EST




On 2/21/2018 1:48 PM, Guenter Roeck wrote:
On Wed, Feb 21, 2018 at 01:24:48PM -0800, Jae Hyun Yoo wrote:
Hi Guenter,

Thanks for sharing your time to review this code. Please check my answers
inline.

On 2/21/2018 10:26 AM, Guenter Roeck wrote:
On Wed, Feb 21, 2018 at 08:16:05AM -0800, Jae Hyun Yoo wrote:
This commit adds a generic PECI hwmon client driver implementation.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@xxxxxxxxxxxxxxx>
---
drivers/hwmon/Kconfig | 10 +
drivers/hwmon/Makefile | 1 +
drivers/hwmon/peci-hwmon.c | 928 +++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 939 insertions(+)
create mode 100644 drivers/hwmon/peci-hwmon.c

diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
index ef23553ff5cb..f22e0c31f597 100644
--- a/drivers/hwmon/Kconfig
+++ b/drivers/hwmon/Kconfig
@@ -1246,6 +1246,16 @@ config SENSORS_NCT7904
This driver can also be built as a module. If so, the module
will be called nct7904.
+config SENSORS_PECI_HWMON
+ tristate "PECI hwmon support"
+ depends on PECI
+ help
+ If you say yes here you get support for the generic PECI hwmon
+ driver.
+
+ This driver can also be built as a module. If so, the module
+ will be called peci-hwmon.
+
config SENSORS_NSA320
tristate "ZyXEL NSA320 and compatible fan speed and temperature sensors"
depends on GPIOLIB && OF
diff --git a/drivers/hwmon/Makefile b/drivers/hwmon/Makefile
index f814b4ace138..946f54b168e5 100644
--- a/drivers/hwmon/Makefile
+++ b/drivers/hwmon/Makefile
@@ -135,6 +135,7 @@ obj-$(CONFIG_SENSORS_NCT7802) += nct7802.o
obj-$(CONFIG_SENSORS_NCT7904) += nct7904.o
obj-$(CONFIG_SENSORS_NSA320) += nsa320-hwmon.o
obj-$(CONFIG_SENSORS_NTC_THERMISTOR) += ntc_thermistor.o
+obj-$(CONFIG_SENSORS_PECI_HWMON) += peci-hwmon.o
obj-$(CONFIG_SENSORS_PC87360) += pc87360.o
obj-$(CONFIG_SENSORS_PC87427) += pc87427.o
obj-$(CONFIG_SENSORS_PCF8591) += pcf8591.o
diff --git a/drivers/hwmon/peci-hwmon.c b/drivers/hwmon/peci-hwmon.c
new file mode 100644
index 000000000000..edd27744adcb
--- /dev/null
+++ b/drivers/hwmon/peci-hwmon.c
@@ -0,0 +1,928 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2018 Intel Corporation
+
+#include <linux/delay.h>
+#include <linux/hwmon.h>
+#include <linux/hwmon-sysfs.h>
+#include <linux/jiffies.h>
+#include <linux/module.h>
+#include <linux/of_device.h>
+#include <linux/peci.h>
+#include <linux/workqueue.h>
+
+#define DIMM_SLOT_NUMS_MAX 12 /* Max DIMM numbers (channel ranks x 2) */
+#define CORE_NUMS_MAX 28 /* Max core numbers (max on SKX Platinum) */
+#define TEMP_TYPE_PECI 6 /* Sensor type 6: Intel PECI */
+
+#define CORE_TEMP_ATTRS 5
+#define DIMM_TEMP_ATTRS 2
+#define ATTR_NAME_LEN 24
+
+#define DEFAULT_ATTR_GRP_NUMS 5
+
+#define UPDATE_INTERVAL_MIN HZ
+#define DIMM_MASK_CHECK_DELAY msecs_to_jiffies(5000)
+
+enum sign {
+ POS,
+ NEG
+};
+
+struct temp_data {
+ bool valid;
+ s32 value;
+ unsigned long last_updated;
+};
+
+struct temp_group {
+ struct temp_data tjmax;
+ struct temp_data tcontrol;
+ struct temp_data tthrottle;
+ struct temp_data dts_margin;
+ struct temp_data die;
+ struct temp_data core[CORE_NUMS_MAX];
+ struct temp_data dimm[DIMM_SLOT_NUMS_MAX];
+};
+
+struct core_temp_group {
+ struct sensor_device_attribute sd_attrs[CORE_TEMP_ATTRS];
+ char attr_name[CORE_TEMP_ATTRS][ATTR_NAME_LEN];
+ struct attribute *attrs[CORE_TEMP_ATTRS + 1];
+ struct attribute_group attr_group;
+};
+
+struct dimm_temp_group {
+ struct sensor_device_attribute sd_attrs[DIMM_TEMP_ATTRS];
+ char attr_name[DIMM_TEMP_ATTRS][ATTR_NAME_LEN];
+ struct attribute *attrs[DIMM_TEMP_ATTRS + 1];
+ struct attribute_group attr_group;
+};
+
+struct peci_hwmon {
+ struct peci_client *client;
+ struct device *dev;
+ struct device *hwmon_dev;
+ struct workqueue_struct *work_queue;
+ struct delayed_work work_handler;
+ char name[PECI_NAME_SIZE];
+ struct temp_group temp;
+ u8 addr;
+ uint cpu_no;
+ u32 core_mask;
+ u32 dimm_mask;
+ const struct attribute_group *core_attr_groups[CORE_NUMS_MAX + 1];
+ const struct attribute_group *dimm_attr_groups[DIMM_SLOT_NUMS_MAX + 1];
+ uint global_idx;
+ uint core_idx;
+ uint dimm_idx;
+};
+
+enum label {
+ L_DIE,
+ L_DTS,
+ L_TCONTROL,
+ L_TTHROTTLE,
+ L_TJMAX,
+ L_MAX
+};
+
+static const char *peci_label[L_MAX] = {
+ "Die\n",
+ "DTS margin to Tcontrol\n",
+ "Tcontrol\n",
+ "Tthrottle\n",
+ "Tjmax\n",
+};
+
+static int send_peci_cmd(struct peci_hwmon *priv, enum peci_cmd cmd, void *msg)
+{
+ return peci_command(priv->client->adapter, cmd, msg);
+}
+
+static int need_update(struct temp_data *temp)
+{
+ if (temp->valid &&
+ time_before(jiffies, temp->last_updated + UPDATE_INTERVAL_MIN))
+ return 0;
+
+ return 1;
+}
+
+static s32 ten_dot_six_to_millidegree(s32 x)
+{
+ return ((((x) ^ 0x8000) - 0x8000) * 1000 / 64);
+}
+
+static int get_tjmax(struct peci_hwmon *priv)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ int rc;
+
+ if (!priv->temp.tjmax.valid) {
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_TEMP_TARGET;
+ msg.param = 0;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);

Is a typecast to a void * necessary ?


No. I'll remove the type cast. Thanks!

+ if (rc < 0)
+ return rc;
+
+ priv->temp.tjmax.value = (s32)msg.pkg_config[2] * 1000;
+ priv->temp.tjmax.valid = true;
+ }
+
+ return 0;
+}
+
+static int get_tcontrol(struct peci_hwmon *priv)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ s32 tcontrol_margin;
+ int rc;
+
+ if (!need_update(&priv->temp.tcontrol))
+ return 0;
+
+ rc = get_tjmax(priv);
+ if (rc < 0)
+ return rc;
+
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_TEMP_TARGET;
+ msg.param = 0;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ tcontrol_margin = msg.pkg_config[1];
+ tcontrol_margin = ((tcontrol_margin ^ 0x80) - 0x80) * 1000;
+
+ priv->temp.tcontrol.value = priv->temp.tjmax.value - tcontrol_margin;
+
+ if (!priv->temp.tcontrol.valid) {
+ priv->temp.tcontrol.last_updated = INITIAL_JIFFIES;
+ priv->temp.tcontrol.valid = true;
+ } else {
+ priv->temp.tcontrol.last_updated = jiffies;
+ }
+
+ return 0;
+}
+
+static int get_tthrottle(struct peci_hwmon *priv)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ s32 tthrottle_offset;
+ int rc;
+
+ if (!need_update(&priv->temp.tthrottle))
+ return 0;
+
+ rc = get_tjmax(priv);
+ if (rc < 0)
+ return rc;
+
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_TEMP_TARGET;
+ msg.param = 0;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ tthrottle_offset = (msg.pkg_config[3] & 0x2f) * 1000;
+ priv->temp.tthrottle.value = priv->temp.tjmax.value - tthrottle_offset;
+
+ if (!priv->temp.tthrottle.valid) {
+ priv->temp.tthrottle.last_updated = INITIAL_JIFFIES;
+ priv->temp.tthrottle.valid = true;
+ } else {
+ priv->temp.tthrottle.last_updated = jiffies;
+ }
+
+ return 0;
+}
+
+static int get_die_temp(struct peci_hwmon *priv)
+{
+ struct peci_get_temp_msg msg;
+ int rc;
+
+ if (!need_update(&priv->temp.die))
+ return 0;
+
+ rc = get_tjmax(priv);
+ if (rc < 0)
+ return rc;
+
+ msg.addr = priv->addr;
+
+ rc = send_peci_cmd(priv, PECI_CMD_GET_TEMP, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ priv->temp.die.value = priv->temp.tjmax.value +
+ ((s32)msg.temp_raw * 1000 / 64);
+
+ if (!priv->temp.die.valid) {
+ priv->temp.die.last_updated = INITIAL_JIFFIES;
+ priv->temp.die.valid = true;
+ } else {
+ priv->temp.die.last_updated = jiffies;
+ }
+
+ return 0;
+}
+
+static int get_dts_margin(struct peci_hwmon *priv)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ s32 dts_margin;
+ int rc;
+
+ if (!need_update(&priv->temp.dts_margin))
+ return 0;
+
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_DTS_MARGIN;
+ msg.param = 0;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ dts_margin = (msg.pkg_config[1] << 8) | msg.pkg_config[0];
+
+ /**
+ * Processors return a value of DTS reading in 10.6 format
+ * (10 bits signed decimal, 6 bits fractional).
+ * Error codes:
+ * 0x8000: General sensor error
+ * 0x8001: Reserved
+ * 0x8002: Underflow on reading value
+ * 0x8003-0x81ff: Reserved
+ */
+ if (dts_margin >= 0x8000 && dts_margin <= 0x81ff)
+ return -1;
+
+ dts_margin = ten_dot_six_to_millidegree(dts_margin);
+
+ priv->temp.dts_margin.value = dts_margin;
+
+ if (!priv->temp.dts_margin.valid) {
+ priv->temp.dts_margin.last_updated = INITIAL_JIFFIES;
+ priv->temp.dts_margin.valid = true;
+ } else {
+ priv->temp.dts_margin.last_updated = jiffies;
+ }
+
+ return 0;
+}
+
+static int get_core_temp(struct peci_hwmon *priv, int core_index)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ s32 core_dts_margin;
+ int rc;
+
+ if (!need_update(&priv->temp.core[core_index]))
+ return 0;
+
+ rc = get_tjmax(priv);
+ if (rc < 0)
+ return rc;
+
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_PER_CORE_DTS_TEMP;
+ msg.param = core_index;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ core_dts_margin = (msg.pkg_config[1] << 8) | msg.pkg_config[0];
+
+ /**
+ * Processors return a value of the core DTS reading in 10.6 format
+ * (10 bits signed decimal, 6 bits fractional).
+ * Error codes:
+ * 0x8000: General sensor error
+ * 0x8001: Reserved
+ * 0x8002: Underflow on reading value
+ * 0x8003-0x81ff: Reserved
+ */
+ if (core_dts_margin >= 0x8000 && core_dts_margin <= 0x81ff)
+ return -1;

Please use valid error codes. This value is returned to user space,
and I don't think this error should be EPERM.


Yes, you are right. I'll fix it.

+
+ core_dts_margin = ten_dot_six_to_millidegree(core_dts_margin);
+
+ priv->temp.core[core_index].value = priv->temp.tjmax.value +
+ core_dts_margin;
+
+ if (!priv->temp.core[core_index].valid) {
+ priv->temp.core[core_index].last_updated = INITIAL_JIFFIES;
+ priv->temp.core[core_index].valid = true;
+ } else {
+ priv->temp.core[core_index].last_updated = jiffies;
+ }

I don't understand the purpose of this code. Why not just set valid = true
and last_updated = jiffies ? Why set anything to INITIAL_JIFFIES some
arbitrary time after boot ? AFAICS the first read will always be followed
by another immediately afterwards if the user requests two readings in
a row. Maybe that is intentional, but not to me. If this code is on
purpose, it will require a detailed explanation.


Agreed. I'll rewrite the code as a function like this:

static void mark_updated(struct temp_data *temp)
{
if (!temp->valid)
temp->valid = true;

Are you concerned about memory write bandwidth ?

temp->valid = true;

should be sufficient. I don't see the point of the if statement.


Right. The variable can be simply overwritten without using the if statement. I'll drop the if statement.


temp->last_updated = jiffies;
}

+
+ return 0;
+}
+
+static int get_dimm_temp(struct peci_hwmon *priv, int dimm_index)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ int channel = dimm_index / 2;
+ int dimm_order = dimm_index % 2;
+ int rc;
+
+ if (!need_update(&priv->temp.dimm[dimm_index]))
+ return 0;
+
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_DDR_DIMM_TEMP;
+ msg.param = channel;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ priv->temp.dimm[dimm_index].value = msg.pkg_config[dimm_order] * 1000;
+
+ if (!priv->temp.dimm[dimm_index].valid) {
+ priv->temp.dimm[dimm_index].last_updated = INITIAL_JIFFIES;
+ priv->temp.dimm[dimm_index].valid = true;
+ } else {
+ priv->temp.dimm[dimm_index].last_updated = jiffies;
+ }
+
+ return 0;
+}
+
+static ssize_t show_tcontrol(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ int rc;
+
+ rc = get_tcontrol(priv);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.tcontrol.value);
+}
+
+static ssize_t show_tcontrol_margin(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+ int rc;
+
+ rc = get_tcontrol(priv);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", sensor_attr->index == POS ?
+ priv->temp.tjmax.value -
+ priv->temp.tcontrol.value :
+ priv->temp.tcontrol.value -
+ priv->temp.tjmax.value);
+}
+
+static ssize_t show_tthrottle(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ int rc;
+
+ rc = get_tthrottle(priv);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.tthrottle.value);
+}
+
+static ssize_t show_tjmax(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ int rc;
+
+ rc = get_tjmax(priv);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.tjmax.value);
+}
+
+static ssize_t show_die_temp(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ int rc;
+
+ rc = get_die_temp(priv);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.die.value);
+}
+
+static ssize_t show_dts_margin(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ int rc;
+
+ rc = get_dts_margin(priv);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.dts_margin.value);
+}
+
+static ssize_t show_core_temp(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+ int core_index = sensor_attr->index;
+ int rc;
+
+ rc = get_core_temp(priv, core_index);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.core[core_index].value);
+}
+
+static ssize_t show_dimm_temp(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(dev);
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+ int dimm_index = sensor_attr->index;
+ int rc;
+
+ rc = get_dimm_temp(priv, dimm_index);
+ if (rc < 0)
+ return rc;
+
+ return sprintf(buf, "%d\n", priv->temp.dimm[dimm_index].value);
+}
+
+static ssize_t show_value(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+
+ return sprintf(buf, "%d\n", sensor_attr->index);
+}
+
+static ssize_t show_label(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+
+ return sprintf(buf, peci_label[sensor_attr->index]);
+}
+
+static ssize_t show_core_label(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+
+ return sprintf(buf, "Core %d\n", sensor_attr->index);
+}
+
+static ssize_t show_dimm_label(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct sensor_device_attribute *sensor_attr = to_sensor_dev_attr(attr);
+
+ char channel = 'A' + (sensor_attr->index / 2);
+ int index = sensor_attr->index % 2;
+
+ return sprintf(buf, "DIMM %d (%c%d)\n",
+ sensor_attr->index, channel, index);
+}
+
+/* Die temperature */
+static SENSOR_DEVICE_ATTR(temp1_label, 0444, show_label, NULL, L_DIE);
+static SENSOR_DEVICE_ATTR(temp1_input, 0444, show_die_temp, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_max, 0444, show_tcontrol, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_crit, 0444, show_tjmax, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_crit_hyst, 0444, show_tcontrol_margin, NULL,
+ POS);
+
+static struct attribute *die_temp_attrs[] = {
+ &sensor_dev_attr_temp1_label.dev_attr.attr,
+ &sensor_dev_attr_temp1_input.dev_attr.attr,
+ &sensor_dev_attr_temp1_max.dev_attr.attr,
+ &sensor_dev_attr_temp1_crit.dev_attr.attr,
+ &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+ NULL
+};
+
+static struct attribute_group die_temp_attr_group = {
+ .attrs = die_temp_attrs,
+};
+
+/* DTS margin temperature */
+static SENSOR_DEVICE_ATTR(temp2_label, 0444, show_label, NULL, L_DTS);
+static SENSOR_DEVICE_ATTR(temp2_input, 0444, show_dts_margin, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp2_min, 0444, show_value, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp2_lcrit, 0444, show_tcontrol_margin, NULL, NEG);
+
+static struct attribute *dts_margin_temp_attrs[] = {
+ &sensor_dev_attr_temp2_label.dev_attr.attr,
+ &sensor_dev_attr_temp2_input.dev_attr.attr,
+ &sensor_dev_attr_temp2_min.dev_attr.attr,
+ &sensor_dev_attr_temp2_lcrit.dev_attr.attr,
+ NULL
+};
+
+static struct attribute_group dts_margin_temp_attr_group = {
+ .attrs = dts_margin_temp_attrs,
+};
+
+/* Tcontrol temperature */
+static SENSOR_DEVICE_ATTR(temp3_label, 0444, show_label, NULL, L_TCONTROL);
+static SENSOR_DEVICE_ATTR(temp3_input, 0444, show_tcontrol, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp3_crit, 0444, show_tjmax, NULL, 0);
+
+static struct attribute *tcontrol_temp_attrs[] = {
+ &sensor_dev_attr_temp3_label.dev_attr.attr,
+ &sensor_dev_attr_temp3_input.dev_attr.attr,
+ &sensor_dev_attr_temp3_crit.dev_attr.attr,
+ NULL
+};
+
+static struct attribute_group tcontrol_temp_attr_group = {
+ .attrs = tcontrol_temp_attrs,
+};
+
+/* Tthrottle temperature */
+static SENSOR_DEVICE_ATTR(temp4_label, 0444, show_label, NULL, L_TTHROTTLE);
+static SENSOR_DEVICE_ATTR(temp4_input, 0444, show_tthrottle, NULL, 0);
+
+static struct attribute *tthrottle_temp_attrs[] = {
+ &sensor_dev_attr_temp4_label.dev_attr.attr,
+ &sensor_dev_attr_temp4_input.dev_attr.attr,
+ NULL
+};
+
+static struct attribute_group tthrottle_temp_attr_group = {
+ .attrs = tthrottle_temp_attrs,
+};
+
+/* Tjmax temperature */
+static SENSOR_DEVICE_ATTR(temp5_label, 0444, show_label, NULL, L_TJMAX);
+static SENSOR_DEVICE_ATTR(temp5_input, 0444, show_tjmax, NULL, 0);
+
+static struct attribute *tjmax_temp_attrs[] = {
+ &sensor_dev_attr_temp5_label.dev_attr.attr,
+ &sensor_dev_attr_temp5_input.dev_attr.attr,
+ NULL
+};
+
+static struct attribute_group tjmax_temp_attr_group = {
+ .attrs = tjmax_temp_attrs,
+};
+
+static const struct attribute_group *
+default_attr_groups[DEFAULT_ATTR_GRP_NUMS + 1] = {
+ &die_temp_attr_group,
+ &dts_margin_temp_attr_group,
+ &tcontrol_temp_attr_group,
+ &tthrottle_temp_attr_group,
+ &tjmax_temp_attr_group,
+ NULL
+};
+
+/* Core temperature */
+static ssize_t (*const core_show_fn[CORE_TEMP_ATTRS]) (struct device *dev,
+ struct device_attribute *devattr, char *buf) = {
+ show_core_label,
+ show_core_temp,
+ show_tcontrol,
+ show_tjmax,
+ show_tcontrol_margin,
+};
+
+static const char *const core_suffix[CORE_TEMP_ATTRS] = {
+ "label",
+ "input",
+ "max",
+ "crit",
+ "crit_hyst",
+};
+
+static int check_resolved_cores(struct peci_hwmon *priv)
+{
+ struct peci_rd_pci_cfg_local_msg msg;
+ int rc;
+
+ if (!(priv->client->adapter->cmd_mask & BIT(PECI_CMD_RD_PCI_CFG_LOCAL)))
+ return -EINVAL;
+
+ /* Get the RESOLVED_CORES register value */
+ msg.addr = priv->addr;
+ msg.bus = 1;
+ msg.device = 30;
+ msg.function = 3;
+ msg.reg = 0xB4;
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PCI_CFG_LOCAL, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ priv->core_mask = msg.pci_config[3] << 24 |
+ msg.pci_config[2] << 16 |
+ msg.pci_config[1] << 8 |
+ msg.pci_config[0];
+
+ if (!priv->core_mask)
+ return -EAGAIN;
+
+ dev_dbg(priv->dev, "Scanned resolved cores: 0x%x\n", priv->core_mask);
+ return 0;
+}
+
+static int create_core_temp_group(struct peci_hwmon *priv, int core_no)
+{
+ struct core_temp_group *data;
+ int i;
+
+ data = devm_kzalloc(priv->dev, sizeof(struct core_temp_group),
+ GFP_KERNEL);
+ if (!data)
+ return -ENOMEM;
+
+ for (i = 0; i < CORE_TEMP_ATTRS; i++) {
+ snprintf(data->attr_name[i], ATTR_NAME_LEN,
+ "temp%d_%s", priv->global_idx, core_suffix[i]);
+ sysfs_attr_init(&data->sd_attrs[i].dev_attr.attr);
+ data->sd_attrs[i].dev_attr.attr.name = data->attr_name[i];
+ data->sd_attrs[i].dev_attr.attr.mode = 0444;
+ data->sd_attrs[i].dev_attr.show = core_show_fn[i];
+ if (i == 0 || i == 1) /* label or temp */
+ data->sd_attrs[i].index = core_no;
+ data->attrs[i] = &data->sd_attrs[i].dev_attr.attr;
+ }
+
+ data->attr_group.attrs = data->attrs;
+ priv->core_attr_groups[priv->core_idx++] = &data->attr_group;
+ priv->global_idx++;
+
+ return 0;
+}
+
+static int create_core_temp_groups(struct peci_hwmon *priv)
+{
+ int rc, i;
+
+ rc = check_resolved_cores(priv);
+ if (!rc) {
+ for (i = 0; i < CORE_NUMS_MAX; i++) {
+ if (priv->core_mask & BIT(i)) {
+ rc = create_core_temp_group(priv, i);
+ if (rc)
+ return rc;
+ }
+ }
+
+ rc = sysfs_create_groups(&priv->hwmon_dev->kobj,
+ priv->core_attr_groups);
+ }
+
+ return rc;
+}
+
+/* DIMM temperature */
+static ssize_t (*const dimm_show_fn[DIMM_TEMP_ATTRS]) (struct device *dev,
+ struct device_attribute *devattr, char *buf) = {
+ show_dimm_label,
+ show_dimm_temp,
+};
+
+static const char *const dimm_suffix[DIMM_TEMP_ATTRS] = {
+ "label",
+ "input",
+};
+
+static int check_populated_dimms(struct peci_hwmon *priv)
+{
+ struct peci_rd_pkg_cfg_msg msg;
+ int i, rc, pass = 0;
+
+do_scan:
+ for (i = 0; i < (DIMM_SLOT_NUMS_MAX / 2); i++) {
+ msg.addr = priv->addr;
+ msg.index = MBX_INDEX_DDR_DIMM_TEMP;
+ msg.param = i; /* channel */
+ msg.rx_len = 4;
+
+ rc = send_peci_cmd(priv, PECI_CMD_RD_PKG_CFG, (void *)&msg);
+ if (rc < 0)
+ return rc;
+
+ if (msg.pkg_config[0]) /* DIMM #0 on the channel */
+ priv->dimm_mask |= BIT(i);
+
+ if (msg.pkg_config[1]) /* DIMM #1 on the channel */
+ priv->dimm_mask |= BIT(i + 1);

Each loop sets overlapping bits in dimm_mask. The first loop sets
bit 0 and 1, the second sets bit 1 and 2, and so on. I _think_ this
should probably set bits (i*2) and (i*2+1). If so, I would suggest to
test the code in a system with more than one DIMM in more than one bank.


Thanks for your pointing out my mistake. It has to be changed to (i*2) and
(i*2+1) as you suggested. I'll fix it with taking enough test on a various
DIMM setting environment.

+ }
+
+ /* Do 2-pass scanning */
+ if (priv->dimm_mask && pass == 0) {
+ pass++;
+ goto do_scan;

This goto is only used to avoid a nested loops. Please don't do that.
If you want to avoid indentation levels, add another function.

Also, this will require an explanation why the loop is executed if
and only if a dimm is found the first time around.


The reason why is needs 2-pass scanning is, we function can be called in the
middle of the timing when cient BIOS is updating the values. I'll add this
as a comment and rewrite this code without using the goto.


How would that be different during the 2nd scan ? If there is concern about
concurrency, the loop would have to be repeated until there are no more changes.
Even then there could be update during the last scan (which the code did not
catch).


Okay. I'll change it to single scan logic after taking some tests.

+ }
+
+ if (!priv->dimm_mask)
+ return -EAGAIN;
+
+ dev_dbg(priv->dev, "Scanned populated DIMMs: 0x%x\n", priv->dimm_mask);
+ return 0;
+}
+
+static int create_dimm_temp_group(struct peci_hwmon *priv, int dimm_no)
+{
+ struct dimm_temp_group *data;
+ int i;
+
+ data = devm_kzalloc(priv->dev, sizeof(struct dimm_temp_group),
+ GFP_KERNEL);
+ if (!data)
+ return -ENOMEM;
+
+ for (i = 0; i < DIMM_TEMP_ATTRS; i++) {
+ snprintf(data->attr_name[i], ATTR_NAME_LEN,
+ "temp%d_%s", priv->global_idx, dimm_suffix[i]);
+ sysfs_attr_init(&data->sd_attrs[i].dev_attr.attr);
+ data->sd_attrs[i].dev_attr.attr.name = data->attr_name[i];
+ data->sd_attrs[i].dev_attr.attr.mode = 0444;
+ data->sd_attrs[i].dev_attr.show = dimm_show_fn[i];
+ data->sd_attrs[i].index = dimm_no;
+ data->attrs[i] = &data->sd_attrs[i].dev_attr.attr;
+ }
+
+ data->attr_group.attrs = data->attrs;
+ priv->dimm_attr_groups[priv->dimm_idx++] = &data->attr_group;
+ priv->global_idx++;
+
+ return 0;
+}
+
+static int create_dimm_temp_groups(struct peci_hwmon *priv)
+{
+ int rc, i;
+
+ rc = check_populated_dimms(priv);
+ if (!rc) {
+ for (i = 0; i < DIMM_SLOT_NUMS_MAX; i++) {
+ if (priv->dimm_mask & BIT(i)) {
+ rc = create_dimm_temp_group(priv, i);
+ if (rc)
+ return rc;
+ }
+ }
+
+ rc = sysfs_create_groups(&priv->hwmon_dev->kobj,
+ priv->dimm_attr_groups);
+ if (!rc)
+ dev_dbg(priv->dev, "Done DIMM temp group creation\n");
+ } else if (rc == -EAGAIN) {
+ queue_delayed_work(priv->work_queue, &priv->work_handler,
+ DIMM_MASK_CHECK_DELAY);
+ dev_dbg(priv->dev, "Diferred DIMM temp group creation\n");

s/Diferred/Deferred/


Will fix this typo.

If PECI never reports any DIMMS, this will be repeated forever until
it finds at least one group. Is this intentional ? If so, I would expect
some detailed explanation of the rationale. As it is, the DIMM temperatures
can show up randomly after some hours of runtime, which isn't exactly
deterministic. Maybe that does make sense, but it will need to be explained.


In general, a client CPU will report DIMM population info just after the
client CPU completes memory training and testing at the very beginning of
BIOS boot. The time varies depends on the client system but it would be less
than 5 minutes. I'll add a timeout logic.

But it is not complete by the time Linux boots, and there is no "incomplete"
message ?

This sounds racy; how is it guaranteed that any reading is complete and that
no additional DIMMs will show up some arbitrary time after the first DIMM
was reported ?


This hwmon driver is for BMC side kernel which is for monitoring remote CPUs, not a local CPU. So it needs to consider the remote CPU's booting from the remote server's power on.

+ }
+
+ return rc;
+}
+
+static void create_dimm_temp_groups_delayed(struct work_struct *work)
+{
+ struct delayed_work *dwork = to_delayed_work(work);
+ struct peci_hwmon *priv = container_of(dwork, struct peci_hwmon,
+ work_handler);
+ int rc;
+
+ rc = create_dimm_temp_groups(priv);
+ if (rc && rc != -EAGAIN)
+ dev_dbg(priv->dev, "Skipped to creat DIMM temp groups\n");
+}
+
+static int peci_hwmon_probe(struct peci_client *client)
+{
+ struct device *dev = &client->dev;
+ struct peci_hwmon *priv;
+ int rc;
+
+ if ((client->adapter->cmd_mask &
+ (BIT(PECI_CMD_GET_TEMP) | BIT(PECI_CMD_RD_PKG_CFG))) !=
+ (BIT(PECI_CMD_GET_TEMP) | BIT(PECI_CMD_RD_PKG_CFG))) {
+ dev_err(dev, "Client doesn't support temperature monitoring\n");
+ return -EINVAL;
+ }
+
+ priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+ if (!priv)
+ return -ENOMEM;
+
+ dev_set_drvdata(dev, priv);
+ priv->client = client;
+ priv->dev = dev;
+ priv->addr = client->addr;
+ priv->cpu_no = priv->addr - PECI_BASE_ADDR;
+
+ snprintf(priv->name, PECI_NAME_SIZE, "peci_hwmon.cpu%d", priv->cpu_no);
+
+ priv->work_queue = create_singlethread_workqueue(priv->name);
+ if (!priv->work_queue)
+ return -ENOMEM;
+
+ priv->hwmon_dev = hwmon_device_register_with_groups(priv->dev,
+ priv->name,
+ priv,
+ default_attr_groups);
+
I'll expect a detailed explanation why using hwmon_device_register_with_info()
does not work for this driver, and why it would make sense to ever register
the hwmon device before its attributes are available. From my perspective,
the driver should delay registration entirely until all attributes are
available. The hwmon ABI implicitly assumes that all sensors are available
at the time of hwmon device registration. Anything else can result in
unexpected behavior.


AFAIK, hwmon_device_register_with_info is for adding a non-standard

This is wrong.

hwmon_device_register_with_info() has an _option_ to provide a set of
additional non-standard attributes as its last parameter. Its purpose
is to simplify drivers by moving sysfs attribute handling into the
hwmon core. All new drivers should use that API unless there is a
compelling reason not to do so.

attribute so hwmon_device_register_with_group is correct in this case. Also,

This is wrong. Quoting from the documentation.

"hwmon_device_register_with_info is the most comprehensive and preferred means
to register a hardware monitoring device. It creates the standard sysfs
attributes in the hardware monitoring core, letting the driver focus on reading
from and writing to the chip instead of having to bother with sysfs attributes.
Its parameters are described in more detail below."

this delayed additional registration case is being used in core-temp.c using
a similar way.

That doesn't make it better.

+ rc = PTR_ERR_OR_ZERO(priv->hwmon_dev);
+ if (rc) {
+ dev_err(dev, "Failed to register peci hwmon\n");
+ return rc;
+ }
+
+ priv->global_idx = DEFAULT_ATTR_GRP_NUMS + 1;
+
+ rc = create_core_temp_groups(priv);
+ if (rc) {
+ dev_err(dev, "Failed to create core groups\n");
+ return rc;
+ }

This should be done before registering the hwmon device (or be left
to the hwmon core by using the _info API). And it should definitely
not return an error while keeping the hwmon device around.


As I answers above, hwmon_device_register_with_info is for adding a
non-standard attribute and this kind of way is being used in core-temp.c

Again, this is wrong.

using a similar way.

Two wrongs don't make it right. Besides, the coretemp driver handles
(or tries to handle) dynamic CPU insertion and removal, which is not
the case here.

Guenter


Thanks for your kind explanation. My understanding is, it should use hwmon_device_register_with_info with multiple hwmon_channel_info parameter as an array for adding standard attributes and don't use the groups parameter of the API because the groups parameter of this API is for adding a non-standard attribute, right? But even with this change, it still needs to use delayed creation because BMC side kernel doesn't know how many DIMMs are populated on a remote server before the remote server completes its memory training and testing in BIOS, but it needs to check the remote server's CPU temperature as immediate as possible to make appropriate thermal control based on the remote CPU's temperature to avoid any critical thermal issue. What would be a better solution in this case?

Thanks a lot,
Jae

+
+ INIT_DELAYED_WORK(&priv->work_handler, create_dimm_temp_groups_delayed);
+
+ rc = create_dimm_temp_groups(priv);
+ if (rc && rc != -EAGAIN)
+ dev_dbg(dev, "Skipped to creat DIMM temp groups\n");
+
Not that it should be there in the first place, but "creat" is not a word.


Please check my above answers. I'll fix the typo.

Again, Thanks a lot for sharing your time to review it. I really appreciate
it.

BR,
Jae

+ dev_dbg(dev, "peci hwmon for CPU at 0x%x registered\n", priv->addr);
+
+ return 0;
+}
+
+static int peci_hwmon_remove(struct peci_client *client)
+{
+ struct peci_hwmon *priv = dev_get_drvdata(&client->dev);
+
+ cancel_delayed_work(&priv->work_handler);
+ destroy_workqueue(priv->work_queue);
+ sysfs_remove_groups(&priv->hwmon_dev->kobj, priv->core_attr_groups);
+ sysfs_remove_groups(&priv->hwmon_dev->kobj, priv->dimm_attr_groups);
+ hwmon_device_unregister(priv->hwmon_dev);
+
+ return 0;
+}
+
+static const struct of_device_id peci_of_table[] = {
+ { .compatible = "intel,peci-hwmon", },
+ { }
+};
+MODULE_DEVICE_TABLE(of, peci_of_table);
+
+static struct peci_driver peci_hwmon_driver = {
+ .probe = peci_hwmon_probe,
+ .remove = peci_hwmon_remove,
+ .driver = {
+ .name = "peci-hwmon",
+ .of_match_table = of_match_ptr(peci_of_table),
+ },
+};
+module_peci_driver(peci_hwmon_driver);
+
+MODULE_AUTHOR("Jae Hyun Yoo <jae.hyun.yoo@xxxxxxxxxxxxxxx>");
+MODULE_DESCRIPTION("PECI hwmon driver");
+MODULE_LICENSE("GPL v2");
--
2.16.1

--
To unsubscribe from this list: send the line "unsubscribe linux-hwmon" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html