scottk/OpenCellular
1
0
Fork 0
mirror of https://github.com/Telecominfraproject/OpenCellular.git synced 2025-12-29 01:50:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

motion_lid: Increase precision in noisy mag check.

Browse Source 
This commit increases the precision used in the noisy magnitude
deviation check by multiplying the scaled sensor data by 8 in the
intermediate calculations.  Prior to this, due to some bits being lost,
certain devices would determine the lid angle as unreliable in specific
angles, even though the device was at rest.

BUG=b:63148973
BRANCH=eve,gru,reef
TEST=Flash bob, set DUT on desk, run `while true; do ectool motionsense
lid_angle; sleep 0.1; done`, slowly move the lid from 15 degrees until
~350.  Verify that no particular angle results in a unreliable lid angle
reading.
TEST=Run `evtest` and examing cros-ec-buttons, fold screen all the way
back to make tablet mode, shake device for at least 30s.  Verify that
there are no spurious transitions of the tablet mode switch.
TEST=Repeat above tests on kevin.

Change-Id: Iff06c1df2dd33c60e26a59183f62f29b71548729
Signed-off-by: Aseda Aboagye <aaboagye@google.com>
Reviewed-on: https://chromium-review.googlesource.com/567050
Commit-Ready: Aseda Aboagye <aaboagye@chromium.org>
Tested-by: Aseda Aboagye <aaboagye@chromium.org>
Reviewed-by: Gwendal Grignou <gwendal@chromium.org>
Reviewed-by: Shawn N <shawnn@chromium.org>
This commit is contained in:
Aseda Aboagye
2017-07-11 14:44:24 -07:00
committed by chrome-bot
parent 1b9a553ece
commit df09bc2c83
1 changed files with 20 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
common/motion_lid.c
View File

@@ -256,16 +256,28 @@ static int calculate_lid_angle(const vector_3_t base, const vector_3_t lid,
lid_range = accel_lid->drv->get_range(accel_lid);
for (i = X; i <= Z; i++) {
scaled_base[i] = base[i] * base_range * 10 / (1 << 15);
scaled_lid[i] = lid[i] * lid_range * 10 / (1 << 15);
/*
* To increase precision, we'll use 8x the sensor data in the
* intermediate calculation. We would normally divide by 2^15.
*
* This is safe because even at a range of 8g, calculating the
* magnitude squared should still be less than the max of a
* 32-bit signed integer.
*
* The max that base[i] could be is 32768, resulting in a max
* value for scaled_base[i] of 640 @ 8g range and force.
* Typically our range is set to 2g.
*/
scaled_base[i] = base[i] * base_range * 10 >> 12;
scaled_lid[i] = lid[i] * lid_range * 10 >> 12;
}
base_magnitude2 = scaled_base[X] * scaled_base[X] +
scaled_base[Y] * scaled_base[Y] +
scaled_base[Z] * scaled_base[Z];
lid_magnitude2 = scaled_lid[X] * scaled_lid[X] +
scaled_lid[Y] * scaled_lid[Y] +
scaled_lid[Z] * scaled_lid[Z];
base_magnitude2 = (scaled_base[X] * scaled_base[X] +
scaled_base[Y] * scaled_base[Y] +
scaled_base[Z] * scaled_base[Z]) >> 6;
lid_magnitude2 = (scaled_lid[X] * scaled_lid[X] +
scaled_lid[Y] * scaled_lid[Y] +
scaled_lid[Z] * scaled_lid[Z]) >> 6;
/*
* Check to see if they differ than more than NOISY_MAGNITUDE_DEVIATION.