fix(os_detection): don't classify macOS sequences as Windows
macOS 26.x (ChibiOS) can send a late wLength=0x04 packet after the two
0xFF packets that already established OS_MACOS. The Windows check
(cnt_ff >= 2 && cnt_04 >= 1) fired on this combination and overwrites
the correct result.
Real-world Windows sequences always start with 0xFF packets (cnt_02 = 0);
macOS sequences always open with at least two 0x02 packets (cnt_02 >= 2).
Adding && setups_data.cnt_02 < 2 to the Windows condition is sufficient
to separate the two populations without affecting any existing detection.
Adds a regression test for the affected sequence.
* usb_device_state: add idle_rate, led and protocol
Previously all usb drivers and platform implementations (expect for our
oddball atsam) tracked the same two global variables:
- keyboard_protocol: to indicate if we are in report or boot protocol
- keyboard_idle: for the idle_rate of the keyboard endpoint
And a local variable that was exposed trough some indirection:
- keyboard_led_state: for the currently set indicator leds (caps lock etc.)
These have all been moved into the usb_device_state struct wich is
accessible by getters and setters.
This reduces code duplication and centralizes the state management
across platforms and drivers.
Signed-off-by: Stefan Kerkmann <karlk90@pm.me>
* usb_device_state: reset protocol on reset
The usb hid specification section 7.2.6 states:
When initialized, all devices default to report protocol. However the
host should not make any assumptions about the device’s state and should
set the desired protocol whenever initializing a device.
Thus on reset we should always do exactly that.
Signed-off-by: Stefan Kerkmann <karlk90@pm.me>
* keyboards: fix oversize warnings
Signed-off-by: Stefan Kerkmann <karlk90@pm.me>
---------
Signed-off-by: Stefan Kerkmann <karlk90@pm.me>
* Default OS_DETECTION_DEBOUNCE bumped from 200ms to 250ms
* Add OS_DETECTION_SINGLE_REPORT to prevent undesired multiple reports
* Prevents random stability issues on ARM MacBooks after switching via KVM
* Works for every device I could test, including ARM MacBooks
* Disabled by default to keep current behavior
* Add Troubleshooting section on documentation
* Tweak reset logic to prevent a freeze with some KVMs
The USB stack on ARM MacBooks is more similar to that of iOS and,
for some reason, it seems to like sending packets that influence
the OS detection and results in a second OS_MACOS report being sent
at a random period of time after plugging the keyboard back. This
does not always happen and the consequences of this vary based on
what the user is doing in the callback, but since this is not
obvious and it's hard to debug, I've decided to add a flag for
those affected by such issue. The stability issue I had in mine was
a combination of factors and I found the actual cause being my own
bad math when changing the default layer, but this change alone is
also confirmed to fix it. Lastly, soem KVMs seem to leave the USB
controlled in a suspended state when cold-booting Windows, meaning
the keyboard would hang and the reset logic would not work. This
tunes it so that it can get out of such state. Also retested for
compatibility with my old KVM to ensure the logic works for both.
Jens van Almsick