USB host HID and DFU classes implementation by vgololobov · Pull Request #3 · renesas/zephyr

vgololobov · 2026-07-16T16:08:12Z

PoC stage according to SoW Block 1 (REQ-VND-USB-002)
for Renesas internal review

… 1 (REQ-VND-USB-002) Signed-off-by: Valentyn Gololobov <val@embedd.it>

… 3 (REQ-VND-USB-005) Signed-off-by: Valentyn Gololobov <val@embedd.it>

thenguyenyf

There is an issue with the current software model: the service built on top of the USBH HID is mutually exclusive with the choices made at build time. This model will prevent support of multiple devices with difference protocol (e.g 1 keyboard + 1 HID CFU device in an application).

I will recommend the new software model featuring per-instance protocol selection. The prototype proposal is below (refers from USBD core model and USBD init sample):

DTS hierarchy:

&zephyr_uhc0 {
	/* A specific Logitech keyboard (046d:c31c), forced boot protocol,
	 * routed to the input subsystem by a service in the app.
	 */
	hid_kbd: hid-device-0 {
		compatible = "zephyr,usbh-hid-device";
		protocol = "boot";
		protocol-code = "keyboard";
		in-report-size = <8>;
		in-polling-period-us = <8000>;
	};

	/* Any non-boot HID interface, full report protocol, consumed raw. */
	hid_raw: hid-device-1 {
		compatible = "zephyr,usbh-hid-device";
		match-class;
		protocol = "report";
		protocol-code = "none";
		in-report-size = <64>;
		report-max-fields = <48>;
		report-max-variants = <8>;
	};
};

Device instance binding: USBH <- USBH HID devices

/*
 * Using this macro as a helper to define USBH HID context from application
 * _hid_ctx_name: USBH HID device context name
 * _usb_controller_ctx: USBH controller context name, which defined by USBH_CONTROLLER_DEFINE
 * _node_id: The device node ID of a "zephyr,usbh-hid-device"
 * /
#define USBH_HID_DEFINE(_hid_ctx_name, _usb_controller_ctx, _node_id);

thenguyenyf · 2026-07-23T06:22:38Z

As I understand it, the enumeration flow in the current implementation is
GET_DESC -> SET_PROTOCOL -> SET_IDLE

But when investigate the common enumeration that is applied on HID USB stack of other platforms/vendor SDKs, it's usually
GET_DESC -> SET_IDLE -> GET_IDLE (optional) -> SET_PROTOCOL -> GET_PROTOCOL (optional) -> start interrupt EPs scheduling

The HID 1.11 specifications do not define a fixed enumeration flow; they simply define each request and its constraints. Therefore, I'm not sure if there is any issue with the current enumeration flow in this PR. Just add this comment as a note — it is not a request to be fixed.

drivers: usbh: hid: Added missing description for `in-idle-rate-ms` drivers: usbh: hid: Reverted SET_IDLE failure to non-fatal SET_IDLE is an optional request and an HID device may not support it; its failure should not compromise the probing function. drivers: usbh: hid: Added set_idle/get_idle as API calls Added `usbh_hid_set_idle_rate` and `usbh_hid_get_idle_rate` to the USB host HID driver. Added a test case to verify their implementation. drivers: usbh: hid: Minor fixes - Removed unused includes - Removed Doxygen description of private functions - Removed debug print function - Fixed physical limits default value drivers: usbh: Reverted unintended file change `uhc_mcux_common.c` drivers: usbh: hid: added clearing of IN endpoint in case of stall `hid.h` should only contain common macros; API and typedefes like the ones used by `hid_report.c` belong into `usbh_hid.h`. Moved those and changed namespacing accordingly. drivers: usbh: hid: Reverted formatting of untouched parts of `hid.h` drivers: usbh: hid: Defined DTS layout for HID devices Moved some configuration options for the USB HID host driver from macros to DTS properties. Now HID driver instances are tied to DTS nodes. drivers: usbh: hid: Fixed licensing attribution Signed-off-by: Valentyn Gololobov <val@embedd.it>

…-rtos#3 Review: - Fixed return value in 'getstate' fn - Eliminated test warnings - Used MIN instead of ternary operator - Updated license text Improvements: - Added recovery in case of DFU device internal status error - Updated dfu_dnload shell command, the FW text is passed from CLI - Compile dfu shell sub-commands only when CONFIG_USBH_DFU_CLASS is enabled - Remove unecessary 'dfu_get_status_msg' API - Increase default USBH_DFU_LIMIT_DATA_ALLOC_BYTES to 512 Bytes Signed-off-by: Valentyn Gololobov <val@embedd.it>

Added `usbh_hid_set_protocol` and `usbh_hid_get_protocol` to change the configured protocol of an HID device. Consequently moved the status of `boot_protocol` to RAM in the driver. Signed-off-by: Mattia Maldini <mattia.m@embedd.it>

nhutnguyenkc

Some minor issues

nhutnguyenkc

I try the shell and see some mismatches in help. Could you please check and correct them?

- Added README for test cases - Fixed HID tests so that they can run on ek_ra8m2 - Added a test for the `set_protocol`/`get_protocol` API. Signed-off-by: Mattia Maldini <mattia.m@embedd.it> Signed-off-by: Valentyn Gololobov <val@embedd.it>

- Remove unnecessary includes - Update Readme and shell help for DFU commands - Use sizeof(struct) instead of numeric constant - Fix typos in comments Signed-off-by: Marian Cingel <marian.c@embedd.it> Signed-off-by: Valentyn Gololobov <val@embedd.it>

…concurrent deallocation and MISRA VLA rule - Now the `tests/subsys/usb/hid` test also checks that the data provided to the user callback is correct. - Removed Doxygen comments from internal methods. - If a transfer is cancelled it may still be needed by the underlying UHC driver, so it must only be allocated by the callback. - A couple of APIs had the `data[data_length]` notation, fixed. - Verified functionality with MAX3421E, aligned the renesas uhc driver to use `-EPIPE` error code for stall events. Signed-off-by: Mattia Maldini <mattia.m@embedd.it> Signed-off-by: Valentyn Gololobov <val@embedd.it>

- `hid_report` test now checks multiple examples of erroneous report descriptors. - `hid` test now checks invalid parameters and a non responsive device. - Fixes some deallocations of uninitialized pointers. Signed-off-by: Mattia Maldini <mattia.m@embedd.it>

z_arm64_sync_exc() demuxes SVC exceptions by the low 8 bits of the SVC immediate (svc #1: IRQ offload, svc zephyrproject-rtos#2: runtime except, svc #3: system call) without checking which exception level the SVC was taken from. On ARM64, SVC from EL0 and EL1 both report ESR EC 0x15, so hardware does not distinguish them. With CONFIG_USERSPACE=y and CONFIG_IRQ_OFFLOAD=y, any user thread can execute a raw svc #1 with x0/x1 pointing at an arbitrary kernel function and argument; the handler then runs 'blr x1' at EL1, i.e. privilege escalation / arbitrary code execution in kernel mode. The RISC-V port avoids this class of bug by distinguishing ECALLU/ECALLM via mcause. Note that CONFIG_IRQ_OFFLOAD depends on TEST (arch/Kconfig), so this is not reachable in a production build; only test builds that enable irq_offload() are exposed. This distinction matters for backport decisions. Read the saved SPSR from the ESF and route any SVC taken from EL0 to a user-only path that accepts just the system call and runtime-except immediates. The runtime-except entry must stay reachable from EL0 because user threads raise stack-check failures and oopses through it; z_arm64_do_kernel_oops() sanitizes the user-supplied reason, so this is not a privilege boundary. All other EL0 SVCs raise a fatal CPU exception. Kernel-mode SVCs keep their previous behavior, except that svc #3 from EL1 is no longer dispatched to the syscall handler, which kernel code never uses. Fixes zephyrproject-rtos#116254 Signed-off-by: Hongquan Li <hongquan.li@processmission.com>

xudongzheng · 2026-09-11T04:13:35Z

Is this intended for submission to upstream Zephyr? FYI zephyrproject-rtos#118758

USB host HID class implementation at PoC stage according to SoW Block…

02b08d9

… 1 (REQ-VND-USB-002) Signed-off-by: Valentyn Gololobov <val@embedd.it>

vgololobov force-pushed the support_renesas_ra_uhc branch from d5f9154 to 02b08d9 Compare July 16, 2026 16:46

vgololobov changed the title ~~USB host HID class implementation~~ USB host HID and DFU classes implementation Jul 16, 2026

vgololobov force-pushed the support_renesas_ra_uhc branch from 626c88f to 327b369 Compare July 16, 2026 16:58

USB host DFU class implementation at PoC stage according to SoW Block…

5c566f9

… 3 (REQ-VND-USB-005) Signed-off-by: Valentyn Gololobov <val@embedd.it>

vgololobov force-pushed the support_renesas_ra_uhc branch from 327b369 to 5c566f9 Compare July 16, 2026 23:12

nhutnguyenkc reviewed Jul 21, 2026

View reviewed changes

thenguyenyf reviewed Jul 23, 2026

View reviewed changes

vgololobov and others added 3 commits July 24, 2026 16:24

nhutnguyenkc reviewed Jul 28, 2026

View reviewed changes

Comment thread include/zephyr/usb/class/usbh_dfu.h

Comment thread subsys/usb/host/class/Kconfig.dfu Outdated

Comment thread subsys/usb/host/class/usbh_dfu.c Outdated

sample test applications for testing USB host HID and DFU classes

349cb11

KhiemNguyenT self-assigned this Jul 28, 2026

nhutnguyenkc reviewed Jul 29, 2026

View reviewed changes

Comment thread subsys/usb/host/usbh_shell.c Outdated

Comment thread subsys/usb/host/usbh_shell.c Outdated

Comment thread subsys/usb/host/usbh_shell.c Outdated

vgololobov and others added 4 commits July 30, 2026 15:27

drivers: usbh: hid:

cb893af

- Added README for test cases - Fixed HID tests so that they can run on ek_ra8m2 - Added a test for the `set_protocol`/`get_protocol` API. Signed-off-by: Mattia Maldini <mattia.m@embedd.it> Signed-off-by: Valentyn Gololobov <val@embedd.it>

DFU host: review update

cfa0e19

- Remove unnecessary includes - Update Readme and shell help for DFU commands - Use sizeof(struct) instead of numeric constant - Fix typos in comments Signed-off-by: Marian Cingel <marian.c@embedd.it> Signed-off-by: Valentyn Gololobov <val@embedd.it>

Uh oh!

Conversation

vgololobov commented Jul 16, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

thenguyenyf left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

thenguyenyf Jul 23, 2026

Choose a reason for hiding this comment

Uh oh!

nhutnguyenkc left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

nhutnguyenkc left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

xudongzheng commented Sep 11, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

nhutnguyenkc left a comment •

edited

Loading