The GIP rumble command takes actuator levels from 0 to 100, but xpad_play_effect() converts the 16-bit FF_RUMBLE magnitudes by dividing by 512, which yields levels up to 127. Not all pads handle out-of-range levels. On a PowerA XBX Spectra Enhanced Wired Controller (20d6:4002), levels 1 to 100 produce proportional rumble, 101 produces almost none, and 110 to 127 produce only weak rumble. Any effect above about 78% of full magnitude, including full-strength rumble, is therefore barely noticeable on this pad. Scale the magnitudes to 0-100 instead. This is the range used by SDL's HIDAPI Xbox One driver and by the xbox_gip driver currently under review on linux-input. Intermediate magnitudes now map to proportionally lower levels than before; for example, half magnitude sends 50 instead of 64. The problem was diagnosed and this patch written with the help of an AI coding assistant (Claude). The submitter tested it on the hardware above and reviewed the change. Fixes: 2a6d7527b35c ("Input: xpad - update Xbox One Force Feedback Support") Cc: stable@vger.kernel.org Signed-off-by: Elliott Tyson --- Notes: Testing: - Built as a module against 7.3.4 (Clang, W=1) with no warnings and loaded on that kernel. A GCC/allmodconfig build was not done. - Hardware: PowerA XBX Spectra Enhanced Wired Controller (20d6:4002). FF_RUMBLE effects were uploaded directly with EVIOCSFF. Before the patch, full magnitude on both motors barely rumbled and half magnitude felt stronger; with the patch, rumble strength increases with magnitude all the way to full. - Not tested on a Microsoft-made controller or on other architectures. Behavior change for pads that already worked: below the old maximum, the same magnitude now gives a level about 22% lower (half magnitude sends 50 instead of 64). If you would rather keep those levels, clamping instead (min(strong / 512, 100)) also fixes this pad. I chose rescaling to match SDL and xbox_gip. The pending xbox_gip series (v7, patch 2 "Input: xpad - Remove Xbox One support") removes this code. This fix is meant for current and stable kernels in the meantime; please drop it if that series lands first. AI assistance: Claude Code (Claude Opus 5.5) found the cause using userspace FF_RUMBLE tests, wrote the code change and the commit message, and built the test module. I ran the hardware tests and reviewed the change and the description before submitting. drivers/input/joystick/xpad.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/input/joystick/xpad.c b/drivers/input/joystick/xpad.c index d2eb48d98fea..6cb3be9845fb 100644 --- a/drivers/input/joystick/xpad.c +++ b/drivers/input/joystick/xpad.c @@ -1606,8 +1606,9 @@ static int xpad_play_effect(struct input_dev *dev, void *data, struct ff_effect packet->data[5] = GIP_MOTOR_ALL; packet->data[6] = 0x00; /* left trigger */ packet->data[7] = 0x00; /* right trigger */ - packet->data[8] = strong / 512; /* left actuator */ - packet->data[9] = weak / 512; /* right actuator */ + /* GIP actuator levels range from 0 to 100 */ + packet->data[8] = strong * 100 / 0xffff; /* left actuator */ + packet->data[9] = weak * 100 / 0xffff; /* right actuator */ packet->data[10] = 0xFF; /* on period */ packet->data[11] = 0x00; /* off period */ packet->data[12] = 0xFF; /* repeat count */ base-commit: ab13b5fc808e043a13f861a7c1c2ba8173c22b84 -- 2.56.1