* [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom
@ 2026-11-10 2:06 Siarhei Vishniakou
2026-11-10 2:06 ` [PATCH 1/3] Input: add ABS_TWIST event code Siarhei Vishniakou
` (3 more replies)
0 siblings, 4 replies; 8+ messages in thread
From: Siarhei Vishniakou @ 2026-11-10 2:06 UTC (permalink / raw)
To: Dmitry Torokhov, Jiri Kosina, Benjamin Tissoires, Ping Cheng,
Jason Gerecke
Cc: Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc,
linux-kernel, Siarhei Vishniakou
Some styli can report rotation along their long ("barrel") axis,
commonly referred to as "twist". While the HID specification defines
this usage as HID_DG_TWIST (0x000d0041) in the Digitizers page, there is
currently no corresponding evdev axis code to expose it to userspace.
Because evdev has no MSC_SCAN equivalent for axes, drivers and OEMs
currently work around this by mapping stylus twist onto unrelated axes
such as ABS_Z or ABS_RZ. As a result, userspace input stacks must
maintain device-specific workarounds to know which axis carries twist.
This series introduces a dedicated ABS_TWIST (0x1d) axis for single-tool
devices such as styli and wires it up in the HID subsystem:
1. Input: Define ABS_TWIST (0x1d) in input-event-codes.h and document it
in event-codes.rst with resolution in units/radian.
2. HID: Map HID_DG_TWIST (0x000d0041) to ABS_TWIST in hid-input, compute
its resolution in units/radian, and add KUnit test coverage.
3. HID: wacom: Report pen twist on ABS_TWIST in addition to ABS_Z so
that userspace can use the standard axis while maintaining backward
compatibility with existing applications that expect ABS_Z.
Siarhei Vishniakou (3):
Input: add ABS_TWIST event code
HID: input: map Digitizers Twist usage to ABS_TWIST
HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist
Documentation/input/event-codes.rst | 6 ++
drivers/hid/hid-debug.c | 1 +
drivers/hid/hid-input-test.c | 87 ++++++++++++++++++++++++++
drivers/hid/hid-input.c | 5 ++
drivers/hid/wacom_wac.c | 35 +++++++++--
include/uapi/linux/input-event-codes.h | 1 +
6 files changed, 129 insertions(+), 6 deletions(-)
--
2.56.0.455.g6c5a503b38-goog
^ permalink raw reply [flat|nested] 8+ messages in thread* [PATCH 1/3] Input: add ABS_TWIST event code 2026-11-10 2:06 [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Siarhei Vishniakou @ 2026-11-10 2:06 ` Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 2/3] HID: input: map Digitizers Twist usage to ABS_TWIST Siarhei Vishniakou ` (2 subsequent siblings) 3 siblings, 0 replies; 8+ messages in thread From: Siarhei Vishniakou @ 2026-11-10 2:06 UTC (permalink / raw) To: Dmitry Torokhov, Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke Cc: Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel, Siarhei Vishniakou Some styli can report the rotation along their long ("barrel") axis. This is typically referred to as "twist". In the HID usage pages, twist is already supported as HID_DG_TWIST. However, this usage is not reported to the userspace via evdev interface today, since there's no MSC_SCAN equivalent for axes. Peripheral OEMs get around this today by mapping this axis to other axes like ABS_Z or ABS_RZ. This is not a valid way to report twist. As a result, the userspace programs that would like to support twist must account for the various ways specific OEMs map this, since there's no generic axis for this. Unify this behaviour by introducing a new axis, ABS_TWIST. This axis will only apply to the single-pointer touch devices (stylus), as opposed to "per-slot" multi-touch devices (which would then suggest something like ABS_MT_TWIST). Here we define the ABS_TWIST absolute axis code (0x1d) to report the rotation of a tool (such as a stylus) around its major (barrel) axis, and document it in event-codes.rst. The device can specify its twist resolution as units/radian. Signed-off-by: Siarhei Vishniakou <svv@google.com> --- Documentation/input/event-codes.rst | 6 ++++++ drivers/hid/hid-debug.c | 1 + include/uapi/linux/input-event-codes.h | 1 + 3 files changed, 8 insertions(+) diff --git a/Documentation/input/event-codes.rst b/Documentation/input/event-codes.rst index b4557462edd7..f1f7fd502081 100644 --- a/Documentation/input/event-codes.rst +++ b/Documentation/input/event-codes.rst @@ -235,6 +235,12 @@ A few EV_ABS codes have special meanings: BTN_TOOL_<name> signals the type of tool that is currently detected by the hardware and is otherwise independent of ABS_DISTANCE and/or BTN_TOUCH. +* ABS_TWIST: + + - Used to describe the rotation of a tool (such as a stylus) around its + major (barrel) axis. If the resolution is non-zero, the twist data is + in units/radian. + * ABS_PROFILE: - Used to describe the state of a multi-value profile switch. An event is diff --git a/drivers/hid/hid-debug.c b/drivers/hid/hid-debug.c index 15e35702773c..594cf5248163 100644 --- a/drivers/hid/hid-debug.c +++ b/drivers/hid/hid-debug.c @@ -1015,6 +1015,7 @@ static const char *absolutes[ABS_CNT] = { [ABS_HAT3Y] = "Hat 3Y", [ABS_PRESSURE] = "Pressure", [ABS_DISTANCE] = "Distance", [ABS_TILT_X] = "XTilt", [ABS_TILT_Y] = "YTilt", [ABS_TOOL_WIDTH] = "ToolWidth", + [ABS_TWIST] = "Twist", [ABS_VOLUME] = "Volume", [ABS_PROFILE] = "Profile", [ABS_MISC] = "Misc", [ABS_MT_TOUCH_MAJOR] = "MTMajor", diff --git a/include/uapi/linux/input-event-codes.h b/include/uapi/linux/input-event-codes.h index f410c22e080d..cdc718fc1c4a 100644 --- a/include/uapi/linux/input-event-codes.h +++ b/include/uapi/linux/input-event-codes.h @@ -862,6 +862,7 @@ #define ABS_TILT_X 0x1a #define ABS_TILT_Y 0x1b #define ABS_TOOL_WIDTH 0x1c +#define ABS_TWIST 0x1d #define ABS_VOLUME 0x20 #define ABS_PROFILE 0x21 -- 2.56.0.455.g6c5a503b38-goog ^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 2/3] HID: input: map Digitizers Twist usage to ABS_TWIST 2026-11-10 2:06 [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 1/3] Input: add ABS_TWIST event code Siarhei Vishniakou @ 2026-11-10 2:06 ` Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 3/3] HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist Siarhei Vishniakou 2026-11-10 2:25 ` [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Dmitry Torokhov 3 siblings, 0 replies; 8+ messages in thread From: Siarhei Vishniakou @ 2026-11-10 2:06 UTC (permalink / raw) To: Dmitry Torokhov, Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke Cc: Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel, Siarhei Vishniakou The axis ABS_TWIST is used to report the "twist" (or barrel) rotation of a stylus. Map the HID Digitizers Twist usage (0x000d0041) to ABS_TWIST in hid-input, and calculate its axis resolution in units/radian. Signed-off-by: Siarhei Vishniakou <svv@google.com> --- drivers/hid/hid-input-test.c | 87 ++++++++++++++++++++++++++++++++++++ drivers/hid/hid-input.c | 5 +++ 2 files changed, 92 insertions(+) diff --git a/drivers/hid/hid-input-test.c b/drivers/hid/hid-input-test.c index 6f5c71660d82..6f1717c0cbe0 100644 --- a/drivers/hid/hid-input-test.c +++ b/drivers/hid/hid-input-test.c @@ -62,9 +62,96 @@ static void hid_test_input_get_battery_property(struct kunit *test) KUNIT_EXPECT_EQ(test, val.intval, POWER_SUPPLY_STATUS_DISCHARGING); } +static void hid_test_input_calc_abs_res_twist(struct kunit *test) +{ + struct hid_field field = { + .logical_minimum = 0, + .logical_maximum = 3599, + .physical_minimum = 0, + .physical_maximum = 3599, + .unit = 0x14, /* Degrees */ + .unit_exponent = -1, + }; + + /* 0..3599 over 0..359.9 degrees -> ~573 units/radian */ + KUNIT_EXPECT_EQ(test, hidinput_calc_abs_res(&field, ABS_TWIST), 573); + + /* 0..359 over 0..359 degrees -> ~57 units/radian */ + field.logical_maximum = 359; + field.physical_maximum = 359; + field.unit_exponent = 0; + KUNIT_EXPECT_EQ(test, hidinput_calc_abs_res(&field, ABS_TWIST), 57); + + /* 0..628 over 0..6.28 radians -> 100 units/radian */ + field.logical_maximum = 628; + field.physical_maximum = 628; + field.unit = 0x12; /* Radians */ + field.unit_exponent = -2; + KUNIT_EXPECT_EQ(test, hidinput_calc_abs_res(&field, ABS_TWIST), 100); + + /* Linear unit (centimeters) is invalid for ABS_TWIST */ + field.unit = 0x11; + KUNIT_EXPECT_EQ(test, hidinput_calc_abs_res(&field, ABS_TWIST), 0); +} + +static void hid_test_input_configure_twist_usage(struct kunit *test) +{ + struct hid_device *dev; + struct hid_driver *drv; + struct hid_input *hidinput; + struct input_dev *input; + struct hid_field *field; + struct hid_usage *usage; + + dev = kunit_kzalloc(test, sizeof(*dev), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, dev); + + drv = kunit_kzalloc(test, sizeof(*drv), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, drv); + dev->driver = drv; + + input = kunit_kzalloc(test, sizeof(*input), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, input); + input->absinfo = kunit_kcalloc(test, ABS_CNT, sizeof(*input->absinfo), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, input->absinfo); + input_set_drvdata(input, dev); + + hidinput = kunit_kzalloc(test, sizeof(*hidinput), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, hidinput); + hidinput->input = input; + + field = kunit_kzalloc(test, sizeof(*field), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, field); + field->report_count = 1; + field->logical_minimum = 0; + field->logical_maximum = 3599; + field->physical_minimum = 0; + field->physical_maximum = 3599; + field->unit = 0x14; /* Degrees */ + field->unit_exponent = -1; + field->usages_priorities = kunit_kzalloc(test, sizeof(int), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, field->usages_priorities); + + usage = kunit_kzalloc(test, sizeof(*usage), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, usage); + usage->hid = HID_DG_TWIST; + + hidinput_configure_usage(hidinput, field, usage, 0); + + KUNIT_EXPECT_EQ(test, usage->type, EV_ABS); + KUNIT_EXPECT_EQ(test, usage->code, ABS_TWIST); + KUNIT_EXPECT_TRUE(test, test_bit(EV_ABS, input->evbit)); + KUNIT_EXPECT_TRUE(test, test_bit(ABS_TWIST, input->absbit)); + KUNIT_EXPECT_EQ(test, input_abs_get_min(input, ABS_TWIST), 0); + KUNIT_EXPECT_EQ(test, input_abs_get_max(input, ABS_TWIST), 3599); + KUNIT_EXPECT_EQ(test, input_abs_get_res(input, ABS_TWIST), 573); +} + static struct kunit_case hid_input_tests[] = { KUNIT_CASE(hid_test_input_update_battery_charge_status), KUNIT_CASE(hid_test_input_get_battery_property), + KUNIT_CASE(hid_test_input_calc_abs_res_twist), + KUNIT_CASE(hid_test_input_configure_twist_usage), { } }; diff --git a/drivers/hid/hid-input.c b/drivers/hid/hid-input.c index 6a1651f572e2..dc4bd2a579d2 100644 --- a/drivers/hid/hid-input.c +++ b/drivers/hid/hid-input.c @@ -291,6 +291,7 @@ __s32 hidinput_calc_abs_res(const struct hid_field *field, __u16 code) case ABS_WHEEL: case ABS_TILT_X: case ABS_TILT_Y: + case ABS_TWIST: if (field->unit == 0x14) { /* If degrees */ /* Convert to radians */ prev = logical_extents; @@ -1018,6 +1019,10 @@ static void hidinput_configure_usage(struct hid_input *hidinput, struct hid_fiel map_abs_clear(ABS_TILT_Y); break; + case 0x41: /* Twist */ + map_abs_clear(ABS_TWIST); + break; + case 0x33: /* Touch */ case 0x42: /* TipSwitch */ case 0x43: /* TipSwitch2 */ -- 2.56.0.455.g6c5a503b38-goog ^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 3/3] HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist 2026-11-10 2:06 [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 1/3] Input: add ABS_TWIST event code Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 2/3] HID: input: map Digitizers Twist usage to ABS_TWIST Siarhei Vishniakou @ 2026-11-10 2:06 ` Siarhei Vishniakou 2026-11-13 23:29 ` Jason Gerecke 2026-11-10 2:25 ` [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Dmitry Torokhov 3 siblings, 1 reply; 8+ messages in thread From: Siarhei Vishniakou @ 2026-11-10 2:06 UTC (permalink / raw) To: Dmitry Torokhov, Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke Cc: Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel, Siarhei Vishniakou Historically, Wacom tablets reported stylus barrel rotation (twist) via ABS_Z because a dedicated evdev axis did not exist. Now that we have this axis, report pen twist on ABS_TWIST in addition to ABS_Z. Keep reporting ABS_Z to maintain backward compatibility for existing applications that support Wacom devices. Signed-off-by: Siarhei Vishniakou <svv@google.com> --- drivers/hid/wacom_wac.c | 35 +++++++++++++++++++++++++++++------ 1 file changed, 29 insertions(+), 6 deletions(-) diff --git a/drivers/hid/wacom_wac.c b/drivers/hid/wacom_wac.c index 614f2adab563..7676bc26a027 100644 --- a/drivers/hid/wacom_wac.c +++ b/drivers/hid/wacom_wac.c @@ -802,8 +802,10 @@ static void wacom_exit_report(struct wacom_wac *wacom) input_report_key(input, BTN_STYLUS2, 0); input_report_key(input, BTN_TOUCH, 0); input_report_abs(input, ABS_WHEEL, 0); - if (features->type >= INTUOS3S) + if (features->type >= INTUOS3S) { input_report_abs(input, ABS_Z, 0); + input_report_abs(input, ABS_TWIST, 0); + } } input_report_key(input, wacom->tool[idx], 0); input_report_abs(input, ABS_MISC, 0); /* reset tool id */ @@ -981,6 +983,7 @@ static int wacom_intuos_general(struct wacom_wac *wacom) t = (data[7] & 0x20) ? ((t > 900) ? ((t-1) / 2 - 1350) : ((t-1) / 2 + 450)) : (450 - t / 2) ; input_report_abs(input, ABS_Z, t); + input_report_abs(input, ABS_TWIST, t); } else { /* 4D mouse 2nd packet */ t = (data[6] << 3) | ((data[7] >> 5) & 7); @@ -1411,6 +1414,7 @@ static void wacom_intuos_pro2_bt_pen(struct wacom_wac *wacom) input_report_abs(pen_input, ABS_TILT_Y, (signed char)frame[8]); input_report_abs(pen_input, ABS_Z, rotation); + input_report_abs(pen_input, ABS_TWIST, rotation); input_report_abs(pen_input, ABS_WHEEL, get_unaligned_le16(&frame[11])); } @@ -1935,7 +1939,7 @@ static void wacom_map_usage(struct input_dev *input, struct hid_usage *usage, int resolution; if (equivalent_usage == HID_DG_TWIST) { - resolution_code = ABS_RZ; + resolution_code = ABS_TWIST; } resolution = hidinput_calc_abs_res(field, resolution_code); @@ -2332,6 +2336,7 @@ static void wacom_wac_pen_usage_mapping(struct hid_device *hdev, break; case HID_DG_TWIST: wacom_map_usage(input, usage, field, EV_ABS, ABS_Z, 0); + wacom_map_usage(input, usage, field, EV_ABS, ABS_TWIST, 0); break; case HID_DG_ERASER: input_set_capability(input, EV_KEY, BTN_TOOL_RUBBER); @@ -2522,10 +2527,21 @@ static void wacom_wac_pen_event(struct hid_device *hdev, struct hid_field *field return; /* send pen events only when the pen is in range */ - if (wacom_wac->hid_data.inrange_state) - input_event(input, usage->type, usage->code, value); - else if (wacom_wac->shared->stylus_in_proximity && !wacom_wac->hid_data.sense_state) - input_event(input, usage->type, usage->code, 0); + if (!wacom_wac->hid_data.inrange_state) { + if (!wacom_wac->shared->stylus_in_proximity || + wacom_wac->hid_data.sense_state) + return; + value = 0; + } + + input_event(input, usage->type, usage->code, value); + /* + * ABS_Z was used to report twist data, because ABS_TWIST was not yet + * defined at the time. For compatibility purposes, also report twist + * via ABS_Z. + */ + if (equivalent_usage == HID_DG_TWIST) + input_report_abs(input, ABS_Z, value); } static void wacom_wac_pen_pre_report(struct hid_device *hdev, @@ -3741,6 +3757,7 @@ void wacom_setup_device_quirks(struct wacom *wacom) __clear_bit(BTN_TOOL_PENCIL, wacom_wac->pen_input->keybit); __clear_bit(BTN_TOOL_AIRBRUSH, wacom_wac->pen_input->keybit); __clear_bit(ABS_Z, wacom_wac->pen_input->absbit); + __clear_bit(ABS_TWIST, wacom_wac->pen_input->absbit); __clear_bit(ABS_DISTANCE, wacom_wac->pen_input->absbit); __clear_bit(ABS_TILT_X, wacom_wac->pen_input->absbit); __clear_bit(ABS_TILT_Y, wacom_wac->pen_input->absbit); @@ -3823,6 +3840,8 @@ int wacom_setup_pen_input_capabilities(struct input_dev *input_dev, case CINTIQ_COMPANION_2: input_set_abs_params(input_dev, ABS_Z, -900, 899, 0, 0); input_abs_set_res(input_dev, ABS_Z, 287); + input_set_abs_params(input_dev, ABS_TWIST, -900, 899, 0, 0); + input_abs_set_res(input_dev, ABS_TWIST, 287); wacom_setup_cintiq(wacom_wac); break; @@ -3835,6 +3854,8 @@ int wacom_setup_pen_input_capabilities(struct input_dev *input_dev, case INTUOS4S: input_set_abs_params(input_dev, ABS_Z, -900, 899, 0, 0); input_abs_set_res(input_dev, ABS_Z, 287); + input_set_abs_params(input_dev, ABS_TWIST, -900, 899, 0, 0); + input_abs_set_res(input_dev, ABS_TWIST, 287); fallthrough; case INTUOS: @@ -3855,6 +3876,8 @@ int wacom_setup_pen_input_capabilities(struct input_dev *input_dev, input_set_abs_params(input_dev, ABS_Z, -900, 899, 0, 0); input_abs_set_res(input_dev, ABS_Z, 287); + input_set_abs_params(input_dev, ABS_TWIST, -900, 899, 0, 0); + input_abs_set_res(input_dev, ABS_TWIST, 287); wacom_setup_intuos(wacom_wac); break; -- 2.56.0.455.g6c5a503b38-goog ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH 3/3] HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist 2026-11-10 2:06 ` [PATCH 3/3] HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist Siarhei Vishniakou @ 2026-11-13 23:29 ` Jason Gerecke 0 siblings, 0 replies; 8+ messages in thread From: Jason Gerecke @ 2026-11-13 23:29 UTC (permalink / raw) To: Siarhei Vishniakou Cc: Dmitry Torokhov, Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke, Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel On Mon, Nov 9, 2026 at 6:21 PM Siarhei Vishniakou <svv@google.com> wrote: > > Historically, Wacom tablets reported stylus barrel rotation (twist) > via ABS_Z because a dedicated evdev axis did not exist. Now that we > have this axis, report pen twist on ABS_TWIST in addition to ABS_Z. > > Keep reporting ABS_Z to maintain backward compatibility for existing > applications that support Wacom devices. > > Signed-off-by: Siarhei Vishniakou <svv@google.com> > --- > drivers/hid/wacom_wac.c | 35 +++++++++++++++++++++++++++++------ > 1 file changed, 29 insertions(+), 6 deletions(-) > > diff --git a/drivers/hid/wacom_wac.c b/drivers/hid/wacom_wac.c > index 614f2adab563..7676bc26a027 100644 > --- a/drivers/hid/wacom_wac.c > +++ b/drivers/hid/wacom_wac.c > @@ -802,8 +802,10 @@ static void wacom_exit_report(struct wacom_wac *wacom) > input_report_key(input, BTN_STYLUS2, 0); > input_report_key(input, BTN_TOUCH, 0); > input_report_abs(input, ABS_WHEEL, 0); > - if (features->type >= INTUOS3S) > + if (features->type >= INTUOS3S) { > input_report_abs(input, ABS_Z, 0); > + input_report_abs(input, ABS_TWIST, 0); > + } > } > input_report_key(input, wacom->tool[idx], 0); > input_report_abs(input, ABS_MISC, 0); /* reset tool id */ > @@ -981,6 +983,7 @@ static int wacom_intuos_general(struct wacom_wac *wacom) > t = (data[7] & 0x20) ? ((t > 900) ? ((t-1) / 2 - 1350) : > ((t-1) / 2 + 450)) : (450 - t / 2) ; > input_report_abs(input, ABS_Z, t); > + input_report_abs(input, ABS_TWIST, t); > } else { > /* 4D mouse 2nd packet */ > t = (data[6] << 3) | ((data[7] >> 5) & 7); > @@ -1411,6 +1414,7 @@ static void wacom_intuos_pro2_bt_pen(struct wacom_wac *wacom) > input_report_abs(pen_input, ABS_TILT_Y, > (signed char)frame[8]); > input_report_abs(pen_input, ABS_Z, rotation); > + input_report_abs(pen_input, ABS_TWIST, rotation); > input_report_abs(pen_input, ABS_WHEEL, > get_unaligned_le16(&frame[11])); > } > @@ -1935,7 +1939,7 @@ static void wacom_map_usage(struct input_dev *input, struct hid_usage *usage, > int resolution; > > if (equivalent_usage == HID_DG_TWIST) { > - resolution_code = ABS_RZ; > + resolution_code = ABS_TWIST; > } Requesting a comment here just to make sure that anyone that is only casually reading doesn't think this code is dead. Setting *some* kind of radial axis here (be it ABS_RZ or ABS_TWIST) is critical to ensuring HID properly calculates the resolution for ABS_Z, and that might not be obvious when you see HID_DG_TWIST forcing the resolution_code to ABS_TWIST. > > resolution = hidinput_calc_abs_res(field, resolution_code); > @@ -2332,6 +2336,7 @@ static void wacom_wac_pen_usage_mapping(struct hid_device *hdev, > break; > case HID_DG_TWIST: Also requesting a comment for this case that explains that the usage here is mapped twice for backwards-compatibility, and that the order matters since the latter call will override `usage->type` and `usage->code`, preventing wacom_wac_pen_event from automatically emitting an ABS_Z event. > wacom_map_usage(input, usage, field, EV_ABS, ABS_Z, 0); > + wacom_map_usage(input, usage, field, EV_ABS, ABS_TWIST, 0); > break; > case HID_DG_ERASER: > input_set_capability(input, EV_KEY, BTN_TOOL_RUBBER); > @@ -2522,10 +2527,21 @@ static void wacom_wac_pen_event(struct hid_device *hdev, struct hid_field *field > return; > > /* send pen events only when the pen is in range */ > - if (wacom_wac->hid_data.inrange_state) > - input_event(input, usage->type, usage->code, value); > - else if (wacom_wac->shared->stylus_in_proximity && !wacom_wac->hid_data.sense_state) > - input_event(input, usage->type, usage->code, 0); > + if (!wacom_wac->hid_data.inrange_state) { > + if (!wacom_wac->shared->stylus_in_proximity || > + wacom_wac->hid_data.sense_state) > + return; > + value = 0; > + } > + > + input_event(input, usage->type, usage->code, value); > + /* > + * ABS_Z was used to report twist data, because ABS_TWIST was not yet > + * defined at the time. For compatibility purposes, also report twist > + * via ABS_Z. > + */ I think this comment could be a little more clear, perhaps referencing the double-mapping in wacom_wac_pen_usage_mapping (and how we need to send data out of both ABS_TWIST and ABS_Z but that `usage->type` and `usage->code` can only be set to one of those two) Jason (she/they) --- Now instead of four in the eights place / you’ve got three, ‘Cause you added one / (That is to say, eight) to the two, / But you can’t take seven from three, / So you look at the sixty-fours.... > + if (equivalent_usage == HID_DG_TWIST) > + input_report_abs(input, ABS_Z, value); > } > > static void wacom_wac_pen_pre_report(struct hid_device *hdev, > @@ -3741,6 +3757,7 @@ void wacom_setup_device_quirks(struct wacom *wacom) > __clear_bit(BTN_TOOL_PENCIL, wacom_wac->pen_input->keybit); > __clear_bit(BTN_TOOL_AIRBRUSH, wacom_wac->pen_input->keybit); > __clear_bit(ABS_Z, wacom_wac->pen_input->absbit); > + __clear_bit(ABS_TWIST, wacom_wac->pen_input->absbit); > __clear_bit(ABS_DISTANCE, wacom_wac->pen_input->absbit); > __clear_bit(ABS_TILT_X, wacom_wac->pen_input->absbit); > __clear_bit(ABS_TILT_Y, wacom_wac->pen_input->absbit); > @@ -3823,6 +3840,8 @@ int wacom_setup_pen_input_capabilities(struct input_dev *input_dev, > case CINTIQ_COMPANION_2: > input_set_abs_params(input_dev, ABS_Z, -900, 899, 0, 0); > input_abs_set_res(input_dev, ABS_Z, 287); > + input_set_abs_params(input_dev, ABS_TWIST, -900, 899, 0, 0); > + input_abs_set_res(input_dev, ABS_TWIST, 287); > wacom_setup_cintiq(wacom_wac); > break; > > @@ -3835,6 +3854,8 @@ int wacom_setup_pen_input_capabilities(struct input_dev *input_dev, > case INTUOS4S: > input_set_abs_params(input_dev, ABS_Z, -900, 899, 0, 0); > input_abs_set_res(input_dev, ABS_Z, 287); > + input_set_abs_params(input_dev, ABS_TWIST, -900, 899, 0, 0); > + input_abs_set_res(input_dev, ABS_TWIST, 287); > fallthrough; > > case INTUOS: > @@ -3855,6 +3876,8 @@ int wacom_setup_pen_input_capabilities(struct input_dev *input_dev, > > input_set_abs_params(input_dev, ABS_Z, -900, 899, 0, 0); > input_abs_set_res(input_dev, ABS_Z, 287); > + input_set_abs_params(input_dev, ABS_TWIST, -900, 899, 0, 0); > + input_abs_set_res(input_dev, ABS_TWIST, 287); > > wacom_setup_intuos(wacom_wac); > break; > -- > 2.56.0.455.g6c5a503b38-goog > > ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom 2026-11-10 2:06 [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Siarhei Vishniakou ` (2 preceding siblings ...) 2026-11-10 2:06 ` [PATCH 3/3] HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist Siarhei Vishniakou @ 2026-11-10 2:25 ` Dmitry Torokhov 2026-11-10 18:57 ` Siarhei Vishniakou 3 siblings, 1 reply; 8+ messages in thread From: Dmitry Torokhov @ 2026-11-10 2:25 UTC (permalink / raw) To: Siarhei Vishniakou Cc: Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke, Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel Hi Siarhei, On Mon, Nov 09, 2026 at 06:06:29PM -0800, Siarhei Vishniakou wrote: > Some styli can report rotation along their long ("barrel") axis, > commonly referred to as "twist". While the HID specification defines > this usage as HID_DG_TWIST (0x000d0041) in the Digitizers page, there is > currently no corresponding evdev axis code to expose it to userspace. > > Because evdev has no MSC_SCAN equivalent for axes, drivers and OEMs > currently work around this by mapping stylus twist onto unrelated axes > such as ABS_Z or ABS_RZ. As a result, userspace input stacks must > maintain device-specific workarounds to know which axis carries twist. How is it different from orientation? Thanks. -- Dmitry ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom 2026-11-10 2:25 ` [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Dmitry Torokhov @ 2026-11-10 18:57 ` Siarhei Vishniakou 2026-11-22 22:13 ` Dmitry Torokhov 0 siblings, 1 reply; 8+ messages in thread From: Siarhei Vishniakou @ 2026-11-10 18:57 UTC (permalink / raw) To: Dmitry Torokhov Cc: Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke, Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel пн, 9 нояб. 2026 г. в 18:25, Dmitry Torokhov <dmitry.torokhov@gmail.com>: > > Hi Siarhei, > > On Mon, Nov 09, 2026 at 06:06:29PM -0800, Siarhei Vishniakou wrote: > > Some styli can report rotation along their long ("barrel") axis, > > commonly referred to as "twist". While the HID specification defines > > this usage as HID_DG_TWIST (0x000d0041) in the Digitizers page, there is > > currently no corresponding evdev axis code to expose it to userspace. > > > > Because evdev has no MSC_SCAN equivalent for axes, drivers and OEMs > > currently work around this by mapping stylus twist onto unrelated axes > > such as ABS_Z or ABS_RZ. As a result, userspace input stacks must > > maintain device-specific workarounds to know which axis carries twist. > > How is it different from orientation? > > Thanks. > > -- > Dmitry ABS_MT_ORIENTATION describes the 2D angle of the contact ellipse (or tool projection) on the surface of the touchscreen. Looking strictly at that initial sentence in multi-touch-protocol.rst, it sounds promising: > ABS_MT_ORIENTATION ... The value should describe a signed quarter of a > revolution clockwise around the touch center. However, the rest of the section clarifies that this is the alignment of the contact ellipse with the X/Y axes, not the "twist" of the finger itself. In fact, the docs note that orientation can be omitted if the contact area is circular - even though a circular finger or stylus tip could still spin in place, and that rotation would not be conveyed. For a stylus, orientation is equivalent to the tilt projected onto the touchscreen plane (which single-tool stylus drivers already capture via ABS_TILT_X / ABS_TILT_Y). By contrast, the proposed ABS_TWIST can be thought of as keeping the stylus at a fixed position and tilt angle, and using two fingers to roll it in place like a pencil. This also matches the HID Digitizers specification, which defines Azimuth (0x3f, rotation of the tool projection around the surface Z axis) separately from Twist (0x41, clockwise rotation of the tool around its own major axis), and hid-multitouch already maps HID_DG_AZIMUTH to ABS_MT_ORIENTATION. Since styli can report tilt/azimuth and twist at the same time (and single-tool pen devices don't use ABS_MT_* events), twist needs a dedicated axis. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom 2026-11-10 18:57 ` Siarhei Vishniakou @ 2026-11-22 22:13 ` Dmitry Torokhov 0 siblings, 0 replies; 8+ messages in thread From: Dmitry Torokhov @ 2026-11-22 22:13 UTC (permalink / raw) To: Siarhei Vishniakou Cc: Jiri Kosina, Benjamin Tissoires, Ping Cheng, Jason Gerecke, Jonathan Corbet, Henrik Rydberg, linux-input, linux-doc, linux-kernel On Tue, Nov 10, 2026 at 10:57:31AM -0800, Siarhei Vishniakou wrote: > пн, 9 нояб. 2026 г. в 18:25, Dmitry Torokhov <dmitry.torokhov@gmail.com>: > > > > Hi Siarhei, > > > > On Mon, Nov 09, 2026 at 06:06:29PM -0800, Siarhei Vishniakou wrote: > > > Some styli can report rotation along their long ("barrel") axis, > > > commonly referred to as "twist". While the HID specification defines > > > this usage as HID_DG_TWIST (0x000d0041) in the Digitizers page, there is > > > currently no corresponding evdev axis code to expose it to userspace. > > > > > > Because evdev has no MSC_SCAN equivalent for axes, drivers and OEMs > > > currently work around this by mapping stylus twist onto unrelated axes > > > such as ABS_Z or ABS_RZ. As a result, userspace input stacks must > > > maintain device-specific workarounds to know which axis carries twist. > > > > How is it different from orientation? > > > > Thanks. > > > > -- > > Dmitry > > ABS_MT_ORIENTATION describes the 2D angle of the contact ellipse (or > tool projection) on the surface of the touchscreen. Looking strictly at that > initial sentence in multi-touch-protocol.rst, it sounds promising: > > > ABS_MT_ORIENTATION ... The value should describe a signed quarter of a > > revolution clockwise around the touch center. > > However, the rest of the section clarifies that this is the alignment of > the contact ellipse with the X/Y axes, not the "twist" of the finger > itself. In fact, the docs note that orientation can be omitted if the > contact area is circular - even though a circular finger or stylus tip > could still spin in place, and that rotation would not be conveyed. > > For a stylus, orientation is equivalent to the tilt > projected onto the touchscreen plane (which single-tool stylus drivers already > capture via ABS_TILT_X / ABS_TILT_Y). > By contrast, the proposed ABS_TWIST can be thought of as keeping the > stylus at a fixed position and tilt angle, and using two fingers to roll > it in place like a pencil. > > This also matches the HID Digitizers specification, which defines > Azimuth (0x3f, rotation of the tool projection around the surface Z > axis) separately from Twist (0x41, clockwise rotation of the tool around > its own major axis), and hid-multitouch already maps HID_DG_AZIMUTH to > ABS_MT_ORIENTATION. Since styli can report tilt/azimuth and twist at the > same time (and single-tool pen devices don't use ABS_MT_* events), twist > needs a dedicated axis. OK, let's add it then. But we need a more precise definition of what the values mean (i.e. is rotation clock- or counterclockwise, only positive or negative, what 0 means). Can we also move it below ABS_SND_PROFILE in the doc? Thanks. -- Dmitry ^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2026-11-22 22:13 UTC | newest] Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2026-11-10 2:06 [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 1/3] Input: add ABS_TWIST event code Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 2/3] HID: input: map Digitizers Twist usage to ABS_TWIST Siarhei Vishniakou 2026-11-10 2:06 ` [PATCH 3/3] HID: wacom: report ABS_TWIST alongside ABS_Z for pen twist Siarhei Vishniakou 2026-11-13 23:29 ` Jason Gerecke 2026-11-10 2:25 ` [PATCH 0/3] Input: add ABS_TWIST and support it in HID and Wacom Dmitry Torokhov 2026-11-10 18:57 ` Siarhei Vishniakou 2026-11-22 22:13 ` Dmitry Torokhov
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®