Realforce RGB (AEAX01) startup problem

[tmk]

Problem:
With old firmware Realforce RGB (AEAX01) fails when connecting the keyboard to USB-USB converter first, then plug it into USB port.

https://geekhack.org/index.php?topic=69169.msg2414679#msg2414679

Confirmed the problem with tmk/tmk_keyboard@5132607 (2016-10-18) at least.
NOT confirmed with tmk/tmk_keyboard@cf9f157 (2017-05-11)

Workaround:
https://geekhack.org/index.php?topic=69169.msg2449152#msg2449152

EDIT: for future reference,
at current time workaroudn is plugin converter first, then the keyboard in order.

This doesn't work with tmk/tmk_keyboard@51326077b08.
And tmk/tmk_keyboard@cf9f157a7f6a70 doens't need the workaround at all.

Solution:
This commit doesn't seem to be related regardless of its commit comment.
227f3aa

commit 227f3aab715bf7f04fc616558b80e7fa69d2b914 (origin/upstream_fix_bus_detection, upstream_fix_bus_detection)
Author: Jun Wako <wakojun@gmail.com>
Date:   Sun Jul 14 10:34:05 2019 +0900

    Disable connect detection during settling time
    
    One device has glitch(30us width pulse) on pull-up'd D+ line after
    pulg-in, it can deceive device attach detction of the chip and library.
    This fix ignores the false detection for a while after device attach.
    The deivce with the glitch is a keyboard of Topre REALFORCE RGB AEAX01.

diff --git a/Usb.cpp b/Usb.cpp
index 7843d49..8c3c54c 100644
--- a/Usb.cpp
+++ b/Usb.cpp
@@ -483,6 +483,9 @@ void USB::Task(void) //USB state machine
                         if((usb_task_state & USB_STATE_MASK) == USB_STATE_DETACHED) {
                                 delay = (uint32_t)millis() + USB_SETTLE_DELAY;
                                 usb_task_state = USB_ATTACHED_SUBSTATE_SETTLE;
+
+                                // disable bus status change interrupt until settled
+                                regWr(rHIEN, (regRd(rHIEN) & ~bmCONDETIE));
                         }
                         break;
         }// switch( tmpdata
@@ -510,6 +513,9 @@ void USB::Task(void) //USB state machine
                                 usb_task_state = USB_ATTACHED_SUBSTATE_RESET_DEVICE;
                         else break; // don't fall through
                 case USB_ATTACHED_SUBSTATE_RESET_DEVICE:
+                        // re-enable bus status change interrupt
+                        regWr(rHIEN, (regRd(rHIEN) | bmCONDETIE));
+
                         regWr(rHCTL, bmBUSRST); //issue bus reset
                         usb_task_state = USB_ATTACHED_SUBSTATE_WAIT_RESET_COMPLETE;
                         break;

https://geekhack.org/index.php?topic=69169.msg2635350#msg2635350

Bottom line:
There is/was confusion at workaround and solution for some reason.
In fact tmk/tmk_keyboard@3eae475f8b98 fixed the problem. 2021-07-23

commit 3eae475f8b98152b466f8f3763a74f0ef2330d9d
Author: tmk <hasu@tmk-kbd.com>
Date:   Thu May 11 00:16:21 2017 +0900

    usb_usb: Remove startup wait
    
    Some keyboards boot up quickly and cannot be enumerated
    when powering up converter and keyboard at same time

diff --git a/converter/usb_usb/main.cpp b/converter/usb_usb/main.cpp
index 5b542a75..2d9f5602 100644
--- a/converter/usb_usb/main.cpp
+++ b/converter/usb_usb/main.cpp
@@ -75,11 +75,13 @@ int main(void)
      */
     sei();
 
+/* Some keyboards bootup quickly and cannot be initialized with this startup wait.
     // wait for startup of sendchar routine
     while (USB_DeviceState != DEVICE_STATE_Configured) ;
     if (debug_enable) {
         _delay_ms(1000);
     }
+*/
 
     debug("init: done\n");

[tmk]

There is no problem in USB_Host_Shield_2.0 as for this.

The fix 227f3aa and its branch should be removed from repo.

https://github.com/tmk/USB_Host_Shield_2.0/commits/upstream_fix_bus_detection

[tmk]

Closed by f0ae8a1