USB Keyboard and Mouse Unneeded Write (OUT) Endpoint Descriptors? #250

New issue

Open

opened 2026-02-20 08:21:29 -05:00 by deekerman · 2 comments

deekerman commented

2026-02-20 08:21:29 -05:00

Owner

Originally created by @rmschooley on GitHub (May 6, 2025).

The usb keyboard, the usb absolute mouse, and the usb relative mouse devices all seem to employ two endpoint descriptors: 1) a read (IN) endpoint descriptor, and 2) a write (OUT) endpoint descriptor. Most if not all usb keyboard and usb mice only employ one read (IN) endpoint descriptor. Unless there is some reason for these write (OUT) endpoint descriptors to exist, it seems like they should be eliminated to simplify host software processing.

Originally created by @rmschooley on GitHub (May 6, 2025). The usb keyboard, the usb absolute mouse, and the usb relative mouse devices all seem to employ two endpoint descriptors: 1) a read (IN) endpoint descriptor, and 2) a write (OUT) endpoint descriptor. Most if not all usb keyboard and usb mice only employ one read (IN) endpoint descriptor. Unless there is some reason for these write (OUT) endpoint descriptors to exist, it seems like they should be eliminated to simplify host software processing.

deekerman added the

component: usb/hid

component: hid/keyboard

labels

2026-02-20 08:21:29 -05:00

deekerman commented

2026-02-20 08:21:30 -05:00

Author

Owner

@wblondel commented on GitHub (May 6, 2025):

Could this unnecessary OUT endpoint descriptor be the cause of missed key release events / repeated key press states / extreme mouse lags?

@wblondel commented on GitHub (May 6, 2025): Could this unnecessary OUT endpoint descriptor be the cause of missed key release events / repeated key press states / extreme mouse lags?

deekerman commented

2026-02-20 08:21:30 -05:00

Author

Owner

@rmschooley commented on GitHub (May 6, 2025):

No, I don't believe so. Unless the device is using these endpoints for some special/unusual reason, I don't think they will ever even be used. With perfect host software, I don't think these extra endpoints would ever cause an issue. But, I think we all know that usb device handling is not always perfect within the bios environment of desktops/notebooks/servers. So, I definitely think they should be removed if at all possible to eliminate any unnecessary complication.

Someday (hopefully soon), I hope to put a software usb bus analyzer or better yet a hardware usb bus analyzer on this device and see what is happening during these crazy key events -- which I have indeed seen as well especially when typing fast. Unfortunately, I don't have access to my hardware analyzer right now.

@rmschooley commented on GitHub (May 6, 2025): No, I don't believe so. Unless the device is using these endpoints for some special/unusual reason, I don't think they will ever even be used. With perfect host software, I don't think these extra endpoints would ever cause an issue. But, I think we all know that usb device handling is not always perfect within the bios environment of desktops/notebooks/servers. So, I definitely think they should be removed if at all possible to eliminate any unnecessary complication. Someday (hopefully soon), I hope to put a software usb bus analyzer or better yet a hardware usb bus analyzer on this device and see what is happening during these crazy key events -- which I have indeed seen as well especially when typing fast. Unfortunately, I don't have access to my hardware analyzer right now.