4.7-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Olof Johansson <o...@lixom.net>

commit 1c500840934a138bd6b13556c210516e9301fbee upstream.

In at least one known setup, the chip comes up in a state where reading
the chip ID returns garbage unless it's been reset, due to noise on the
wires during system boot.

All supported chips have the same reset method, and based on the
datasheets they all need 1.3 or 1.8ms to recover after reset. So, do
the conservative thing here and always reset the chip.

Signed-off-by: Olof Johansson <o...@lixom.net>
Reviewed-by: Srinivas Pandruvada <srinivas.pandruv...@linux.intel.com>
Signed-off-by: Jonathan Cameron <ji...@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gre...@linuxfoundation.org>

---
 drivers/iio/accel/bmc150-accel-core.c |   11 +++++++++++
 1 file changed, 11 insertions(+)

--- a/drivers/iio/accel/bmc150-accel-core.c
+++ b/drivers/iio/accel/bmc150-accel-core.c
@@ -67,6 +67,9 @@
 #define BMC150_ACCEL_REG_PMU_BW                0x10
 #define BMC150_ACCEL_DEF_BW                    125
 
+#define BMC150_ACCEL_REG_RESET                 0x14
+#define BMC150_ACCEL_RESET_VAL                 0xB6
+
 #define BMC150_ACCEL_REG_INT_MAP_0             0x19
 #define BMC150_ACCEL_INT_MAP_0_BIT_SLOPE       BIT(2)
 
@@ -1497,6 +1500,14 @@ static int bmc150_accel_chip_init(struct
        int ret, i;
        unsigned int val;
 
+       /*
+        * Reset chip to get it in a known good state. A delay of 1.8ms after
+        * reset is required according to the data sheets of supported chips.
+        */
+       regmap_write(data->regmap, BMC150_ACCEL_REG_RESET,
+                    BMC150_ACCEL_RESET_VAL);
+       usleep_range(1800, 2500);
+
        ret = regmap_read(data->regmap, BMC150_ACCEL_REG_CHIP_ID, &val);
        if (ret < 0) {
                dev_err(dev, "Error: Reading chip id\n");


Reply via email to