USB Host: integrate class API [1: types] by josuah · Pull Request #94504 · zephyrproject-rtos/zephyr

josuah · 2025-08-14T11:57:23Z

Downstream:

USB Host: integrate class API [2: helpers] #94590

This is a very small, incremental change to help downstream pull requests use a shared base.

I was thinking of doing several such small pull requests with just a few commits, to progressively integrate the code of other pull requests:

Here is how I tested it:

west build -t run -b native_sim samples/subsys/usb/shell/ \
-DEXTRA_CONF_FILE=virtual.conf \
-DDTC_OVERLAY_FILE=virtual.overlay \
-DEXTRA_CONF_FILE=device_and_host_prj.conf

Then in another terminal, connect to the console, such as using picocom /dev/pts/0.

Make the struct name match the device naming for ease of use, although slightly longer name. Propagate the change to the subsystem, includes, tests and samples. Signed-off-by: Josuah Demangeon <josuah.demangeon@nordicsemi.no>

jfischer-no · 2025-09-03T09:30:34Z


 	udev->cfg_desc = k_heap_alloc(&usb_device_heap,
-				      cfg_desc.wTotalLength,
+				      cfg_desc.wTotalLength + sizeof(struct usb_desc_header),


What is the intention? I remember doing the same, but then I dropped it in favor of just using wTotalLength.

This was to add a nil descriptor at the end so that places like this can be sure to never go past the end of the array:

zephyr/subsys/usb/host/usbh_device.c

Line 288 in 11a9096

while ((dhp->bDescriptorType != 0 || dhp->bLength != 0) && (void *)dhp < desc_end) {

Maybe the code above can be modified instead? Such as testing for (void *)dhp < desc_end only and not test for dhp->bDescriptorType != 0 || dhp->bLength != 0?

I don't like the concept of nil descriptor. I think the descriptor sets for classes should just end with NULL pointer, and the check here should be fine with (void *)dhp < desc_end condition.

But of course, this needs to detect bLength == 0 and fail (if it is within wTotalLength) because the data is tainted (coming from external untrusted source).

Thank you I understand the difference now, I am dropping the commit.

I went with this to check dhp before it is de-referenced:

-while ((dhp->bDescriptorType != 0 || dhp->bLength != 0) && (void *)dhp < desc_end) { +while ((void *)dhp < desc_end && (dhp->bDescriptorType != 0 || dhp->bLength != 0)) {

jfischer-no · 2025-09-03T09:59:37Z

+ *
+ * @return true if USB host is in enabled, false otherwise
+ */
+static inline bool usbd_is_enabled(const struct usbh_context *const uhs_ctx)


s/usbd/usbh

However, I am not sure how it could be used. In my approach, I do not explicitly register any class drivers, I simply place them in an iterable section.

I am removing it then, thank you

jfischer-no · 2025-09-03T09:59:39Z

+ *
+ * @return true if USB host is in enabled, false otherwise
+ */
+static inline bool usbd_is_initialized(const struct usbh_context *const uhs_ctx)


s/usbd/usbh

In the while loop parsing descriptors, check that the descriptor is past the end just before dereferencing it to check if it is seemingly valid. Signed-off-by: Josuah Demangeon <josuah.demangeon@nordicsemi.no>

jfischer-no · 2025-09-04T13:17:30Z

+static inline bool usbh_is_initialized(const struct usbh_context *const uhs_ctx)
+{
+	return uhs_ctx->status.initialized;
+}


Sorry, perhaps I was not clear. I think it would be better to remove this commit from the PR. It is not used anywhere. I will comment on your second PR once this one has been merged.

Hide the mutex lock details inside a wrapper call, like done for the device stack API. This is not used for the per-device lock of the host stack which use a different mutex lock. Signed-off-by: Josuah Demangeon <josuah.demangeon@nordicsemi.no>

Update the USB shell sample to the new syntax. This is based on tests done with native_sim, along with the virtual.conf, virtual.overlay and device_and_host_prj.conf extra config files. Signed-off-by: Josuah Demangeon <josuah.demangeon@nordicsemi.no>

sonarqubecloud · 2025-09-04T16:05:45Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

josuah · 2025-09-04T16:22:26Z

Well it works much better when testing things before pushing...

kartben · 2025-09-05T11:46:02Z

@@ -53,7 +53,7 @@ struct usbh_contex {

 #define USBH_CONTROLLER_DEFINE(device_name, uhc_dev)			\


please consider adding doxygen docs at some point in the future :) thanks!

usb: host: rename "struct usbh_contex" to "usbh_context"

0054ae9

Make the struct name match the device naming for ease of use, although slightly longer name. Propagate the change to the subsystem, includes, tests and samples. Signed-off-by: Josuah Demangeon <josuah.demangeon@nordicsemi.no>

josuah added the area: USB Universal Serial Bus label Aug 14, 2025

josuah marked this pull request as ready for review August 14, 2025 12:57

zephyrbot added the area: Samples Samples label Aug 14, 2025

zephyrbot requested review from jfischer-no, kartben, nashif and tmon-nordic August 14, 2025 12:58

zephyrbot assigned jfischer-no Aug 14, 2025

josuah mentioned this pull request Aug 14, 2025

usb: host: class: support for usb host video class #94085

Merged

josuah changed the title ~~Start refactoring the Host class API~~ USB Host: integrate class API [1] Aug 17, 2025

josuah mentioned this pull request Aug 17, 2025

USB Host: integrate class API [2: helpers] #94590

Merged

josuah changed the title ~~USB Host: integrate class API [1]~~ USB Host: integrate class API [1: types] Aug 17, 2025

josuah mentioned this pull request Aug 17, 2025

USB Host: integrate class API [3: UVC] #94591

Merged

jfischer-no reviewed Sep 3, 2025

View reviewed changes

josuah force-pushed the pr_usb_host_class_api branch from 2a7863a to 4b8314c Compare September 3, 2025 12:13

zephyrbot requested a review from JarmouniA September 3, 2025 12:14

usb: host: check that descriptor is in bound before dereferencing

9f1d92a

In the while loop parsing descriptors, check that the descriptor is past the end just before dereferencing it to check if it is seemingly valid. Signed-off-by: Josuah Demangeon <josuah.demangeon@nordicsemi.no>

josuah force-pushed the pr_usb_host_class_api branch from 4b8314c to 4c8c486 Compare September 3, 2025 14:40

josuah mentioned this pull request Sep 4, 2025

drivers: uhc_dwc2: Initial support #94266

Closed

jfischer-no reviewed Sep 4, 2025

View reviewed changes

josuah force-pushed the pr_usb_host_class_api branch 3 times, most recently from d78e394 to d769180 Compare September 4, 2025 15:48

Josuah Demangeon added 2 commits September 4, 2025 15:49

josuah force-pushed the pr_usb_host_class_api branch from d769180 to 8a40c80 Compare September 4, 2025 15:49

josuah requested a review from jfischer-no September 4, 2025 16:24

jfischer-no approved these changes Sep 4, 2025

View reviewed changes

tmon-nordic approved these changes Sep 5, 2025

View reviewed changes

kartben merged commit 190cde1 into zephyrproject-rtos:main Sep 5, 2025
26 checks passed

kartben reviewed Sep 5, 2025

View reviewed changes

josuah deleted the pr_usb_host_class_api branch September 7, 2025 15:41

This was referenced Nov 13, 2025

subsys: usb: host: Add USB Host CDC ECM Class Support for Ethernet #99097

Open

subsys: usb: host: Add USB Host Class Support #99334

Closed

Girinandha-M mentioned this pull request Nov 13, 2025

subsys: usb: host: Add USB Host Vendor specific serial class support #99173

Open

AidenHu mentioned this pull request Nov 20, 2025

usb: host: class: support for usb host hub class #99735

Open

		@@ -53,7 +53,7 @@ struct usbh_contex {

		#define USBH_CONTROLLER_DEFINE(device_name, uhc_dev) \

Conversation

josuah commented Aug 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

josuah Sep 3, 2025 • 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sonarqubecloud Bot commented Sep 4, 2025

Quality Gate passed

Uh oh!

josuah commented Sep 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

josuah commented Aug 14, 2025 •

edited

Loading

josuah Sep 3, 2025 •

edited

Loading

josuah commented Sep 4, 2025 •

edited

Loading