USB HID device stays in infinite loop when key pressed, when there is no USB support...

[matsobdev]

...when plugged in to powered off computer (or restarting from working both computer and USB stack). After computer startup, stack initialises at BIOS USB and repetitive key pressing causes:

while (hid_state == HID_STATE_BUSY) {
}

in hid_keyboard_template.c to loop to infinity and beyond when key pressed between BIOS and OS loading, where is no USB host. Changing it to:

void send_report(uint8_t rep_id, uint8_t modifier, uint8_t key_code)
{
   uint8_t sendbuffer[9] = {rep_id, modifier, 0x00, key_code, 0x00, 0x00, 0x00, 0x00, 0x00};
   int ret = usbd_ep_start_write(HID_INT_EP, sendbuffer, 9);
   if (ret < 0) {
      return;
   }
   hid_state = HID_STATE_BUSY;
   // enough is enough - max time to proceed
   uint64_t start;
   uint64_t duration = 0;
   start = time_us_64();
   while (hid_state == HID_STATE_BUSY && duration < 100000) {
      duration = time_us_64() - start;
   }
}

solves the issue. And it happens on Pico or Pico W.

[sakumisu]

It is user code, i just give a simple template.

[matsobdev]

Yes, it's modified for that purpose of dealing with an issue (only after comment). Everything's great, works nice but it locks in described circumstances. Talking about unmodified hid_keyboard_template.c. hid_state will be always HID_STATE_BUSY then, so bypassing it to try again when if occurs too long - I guess unnaturally long for HID keyboard.

[sakumisu]

Yes, it's modified for that purpose of dealing with an issue (only after comment). Everything's great, works nice but it locks in described circumstances. Talking about unmodified hid_keyboard_template.c. hid_state will be always HID_STATE_BUSY then, so bypassing it to try again when if occurs too long - I guess unnaturally long for HID keyboard.

This should be modified by users, i just tell that the api is ok, how to call the api is user work.