On 21/12/2022 20:15, Daniel Lezcano wrote:
On 21/12/2022 19:05, Dmitry Baryshkov wrote:
On 21/12/2022 17:45, Daniel Lezcano wrote:
On 21/12/2022 03:05, Dmitry Baryshkov wrote:
According to the vendor kernels (msm-3.10, 3.14 and 3.18), msm8939
supports only 9 sensors. Remove the rogue sensor's hw_id.
Fixes: 332bc8ebab2c ("thermal: qcom: tsens-v0_1: Add support for
MSM8939")
Cc: Bryan O'Donoghue <bryan.odonoghue@xxxxxxxxxx>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
---
drivers/thermal/qcom/tsens-v0_1.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/thermal/qcom/tsens-v0_1.c
b/drivers/thermal/qcom/tsens-v0_1.c
index 0bc4e5cec184..57ac23f9d9b7 100644
--- a/drivers/thermal/qcom/tsens-v0_1.c
+++ b/drivers/thermal/qcom/tsens-v0_1.c
@@ -605,9 +605,9 @@ static const struct tsens_ops ops_8939 = {
};
struct tsens_plat_data data_8939 = {
- .num_sensors = 10,
+ .num_sensors = 9,
.ops = &ops_8939,
- .hw_ids = (unsigned int []){ 0, 1, 2, 3, 5, 6, 7, 8, 9,
10 },
+ .hw_ids = (unsigned int []){ 0, 1, 2, 3, 5, 6, 7, 8, 9 },
Does not patch 4 says we can drop those hw_ids ?
No. In patch 4 we drop contiguous IDs. For the msm8939 sensor 4 is
omitted from hw_ids, so we can not drop the array.
Ah, yes, indeed.
.num_sensor = ARRAY_SIZE(data_8939.hw_ids); should work
Interesting suggestion, I should give it a thought.
If the hw_ids are ordered, you may consider a bitmask instead of an array
I'm not sure if it's worth doing that. With the current approach it is
easy to get hw_id corresponding to the sensor #i, If we switch to the
bitmask, it wouldn't be that easy. And the saved space isn't that big.
--
With best wishes
Dmitry
