Thread (3 messages) 3 messages, 2 authors, 2015-10-08

[PATCH v2 2/9] thermal: qcom: tsens-8916: Add support for 8916 family of SoCs

From: Punit Agrawal <hidden>
Date: 2015-09-16 13:47:17
Also in: linux-arm-msm, linux-pm

Hi Rajendra,

One comment below.

Rajendra Nayak [off-list ref] writes:
Add support to calibrate sensors on 8916 family and also add common
functions to read temperature from sensors (This can be reused on
other SoCs having similar TSENS device)
The calibration data is read from eeprom using the generic nvmem
framework apis.

Based on the original code by Siddartha Mohanadoss and Stephen Boyd.

Signed-off-by: Rajendra Nayak <redacted>
---
 drivers/thermal/qcom/Makefile       |   2 +-
 drivers/thermal/qcom/tsens-8916.c   | 107 +++++++++++++++++++++++++++++
 drivers/thermal/qcom/tsens-common.c | 130 ++++++++++++++++++++++++++++++++++++
 drivers/thermal/qcom/tsens.c        |   1 +
 drivers/thermal/qcom/tsens.h        |  11 +++
 5 files changed, 250 insertions(+), 1 deletion(-)
 create mode 100644 drivers/thermal/qcom/tsens-8916.c
 create mode 100644 drivers/thermal/qcom/tsens-common.c
[...]
quoted hunk
diff --git a/drivers/thermal/qcom/tsens-common.c b/drivers/thermal/qcom/tsens-common.c
new file mode 100644
index 0000000..4acf52c
--- /dev/null
+++ b/drivers/thermal/qcom/tsens-common.c
@@ -0,0 +1,130 @@
+/*
+ * Copyright (c) 2015, The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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/platform_device.h>
+#include <linux/nvmem-consumer.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/of_address.h>
+#include <linux/regmap.h>
+#include "tsens.h"
+
+#define S0_ST_ADDR		0x1030
+#define SN_ADDR_OFFSET		0x4
+#define SN_ST_TEMP_MASK		0x3ff
+#define CAL_DEGC_PT1		30
+#define CAL_DEGC_PT2		120
+#define SLOPE_FACTOR		1000
+
+char *qfprom_read(struct device *dev, const char *cname)
+{
+	struct nvmem_cell *cell;
+	ssize_t data;
+	char *ret;
+
+	cell = nvmem_cell_get(dev, cname);
+	if (IS_ERR(cell))
+		return ERR_CAST(cell);
+
+	ret = nvmem_cell_read(cell, &data);
+	nvmem_cell_put(cell);
+	return ret;
+}
+
+void compute_intercept_slope(struct tsens_device *tmdev, u32 *p1,
+			     u32 *p2, u32 mode)
+{
+	int i;
+	int num, den;
+
+	for (i = 0; i < tmdev->num_sensors; i++) {
+		dev_dbg(tmdev->dev,
+			"sensor%d - data_point1:%#x data_point2:%#x\n",
+			i, p1[i], p2[i]);
+
+		if (mode == TWO_PT_CALIB) {
+			/*
+			 * slope (m) = adc_code2 - adc_code1 (y2 - y1)/
+			 *	temp_120_degc - temp_30_degc (x2 - x1)
+			 */
+			num = p2[i] - p1[i];
+			num *= SLOPE_FACTOR;
+			den = CAL_DEGC_PT2 - CAL_DEGC_PT1;
+			tmdev->sensor[i].slope = num / den;
+		}
+
+		tmdev->sensor[i].offset = (p1[i] * SLOPE_FACTOR) -
+				(CAL_DEGC_PT1 *
+				tmdev->sensor[i].slope);
+		dev_dbg(tmdev->dev, "offset:%d\n", tmdev->sensor[i].offset);
+	}
+}
+
+static inline int code_to_degc(u32 adc_code, const struct tsens_sensor *s)
+{
+	int degc, num, den;
+
+	num = (adc_code * SLOPE_FACTOR) - s->offset;
+	den = s->slope;
+
+	if (num > 0)
+		degc = num + (den / 2);
+	else if (num < 0)
+		degc = num - (den / 2);
+	else
+		degc = num;
+
+	degc /= den;
+
+	return degc;
+}
+
+int get_temp_common(struct tsens_device *tmdev, int id, int *temp)
+{
+	struct tsens_sensor *s = &tmdev->sensor[id];
+	u32 code;
+	unsigned int sensor_addr;
+	int last_temp = 0, ret;
+
+	sensor_addr = S0_ST_ADDR + s->hw_id * SN_ADDR_OFFSET;
+	ret = regmap_read(tmdev->map, sensor_addr, &code);
+	if (ret)
+		return ret;
+	last_temp = code & SN_ST_TEMP_MASK;
+
+	*temp = code_to_degc(last_temp, s) * 1000;
+
+	return 0;
+}
The way this function is coded the temperature will only ever change by
1C (i.e., 1000mC). Is there a chance that you're losing precision in the
code_to_degc conversion that could be preserved perhaps?

This is useful in scenarios where you use the power allocator governor
and the greater precision allows better power budget estimation.
+
+static const struct regmap_config tsens_config = {
+	.reg_bits	= 32,
+	.val_bits	= 32,
+	.reg_stride	= 4,
+};
+
+int init_common(struct tsens_device *tmdev)
+{
+	void __iomem *base;
+
+	base = of_iomap(tmdev->dev->of_node, 0);
+	if (IS_ERR(base))
+		return -EINVAL;
+
+	tmdev->map = devm_regmap_init_mmio(tmdev->dev, base, &tsens_config);
+	if (!tmdev->map)
+		return -ENODEV;
+
+	return 0;
+}
[...]
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help