fix(hid-rp): gamepad input report data offset for byte-wide axes

[owenpkent]

Problem

GamepadInputReport::get_report() and set_data() skip a fixed two bytes at the front of the object. That is correct for the default uint16_t axes (the one-byte report id plus one byte of alignment padding before joystick_axes), but a report with uint8_t axes has no padding, so the copy starts one byte late:

• X is dropped and every later field moves down one slot (the host sees Y on X, Z on Y, and so on)
• RZ reads the brake value, 0 at rest, which pins the right stick to one edge
• the last byte is read from past the end of the object, and set_data() writes one byte past it

With REPORT_ID == 0 there is no id byte at all, so the offset is wrong there too.

Object dump of GamepadInputReport<12, uint8_t, uint8_t, 0, 255, 0, 255, 1> with sentinel values (01 is the id, then X Y Z RZ brake accelerator hat buttons consumer):

object bytes:  01 A1 A2 A3 A4 B1 B2 01 00 00 00
get_report():  A2 A3 A4 B1 B2 01 00 00 00 C0   before (last byte is whatever follows the object)
get_report():  A1 A2 A3 A4 B1 B2 01 00 00 00   after

We hit this building an 8-bit joystick for the Xbox Adaptive Controller: the stick sat at one edge and only moved on one axis.

Fix

Derive the offset from the layout instead of hard-coding it: data_offset = REPORT_ID != 0 ? alignof(JOYSTICK_TYPE) : 0, used by both get_report() and set_data(). The default uint16_t report is unchanged (its offset is still 2).

Test

components/hid-rp/test/hid_rp_gamepad_host_test.cpp, in the same standalone style as the other hid-rp host tests. It checks the byte layout of the default report, a byte-wide report, a byte-wide report with no id, the neutral (centered) report, and the set_data() round trip. Against the current header the byte-wide cases fail on every field; with this change everything passes. hid_rp_3dconnexion_host_test and hid_rp_report_map_host_test still pass.

Not touched here, since no current report uses it: mixed widths (say uint8_t axes with uint16_t triggers) would also get padding between joystick_axes and trigger_axes, which num_data_bytes does not account for.

[github-actions]

✅Static analysis result - no issues found! ✅

[Copilot]

Copilot review overview

🟢 Approval recommended

The offset correction matches the supported object layouts and is covered by focused regression tests.

Review effort: Balanced
Findings: None

What changed in this PR

Fixes gamepad report serialization offsets for byte-wide axes and reports without IDs.

Changes:

• Derives payload offset from report ID presence and joystick alignment.
• Adds host tests for default, byte-wide, no-ID, neutral, and round-trip cases.

File	Description
components/​hid-rp/​include/​hid-rp-gamepad.hpp	Corrects payload offsets in serialization and deserialization.
components/​hid-rp/​test/​hid_rp_gamepad_host_test.cpp	Adds regression coverage for affected layouts.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.