GCS connects only once to powered FC

[mluessi]

• Seen in version (release version or githash): d93a8de
• On flight controller (if applicable): Brain, BrainRE1
• OS: Linux

Issue details:

• Power FC from external 5V
• Connect USB, GCS connects to FC normally
• Disconnect USB
• Reconnect USB, GCS does not establish connection

[DTFUHF]

can repro on Seppuku, Windows, and artifice preview

[DTFUHF]

Also Linux

Changing computers or restarting GCS doesn't fix the problem, only removing and reapplying power to the FC.

[tracernz]

I managed to software reboot Seppuku a few times without issue. As soon as I disconnected the USB cable and then reconnected it... it connects and attempts to send session object (GCSTelemetryStats) but looks like nothing is coming back. Flight side issue by the looks? Perhaps uncovered by #1506? But I don't see how...

[tracernz]

This looks like #1384...

[tracernz]

It's a USB issue. I see it over both HID and VCP, but not over TCP or UART.

[tracernz]

Quanton is the only F4 board with VBUS connected to the correct pin (PA9) for hardware (USB peripheral) VBUS detection and also the only one that doesn't suffer this bug. ;)

[tracernz]

Didn't have Brain/Re1 schematics but looking at the hardware defs they should be both PA9 too? They are configured as no pull though. I assume they have external pull-downs?

[tracernz]

Can't repro success with Quanton now... 👎 When it's in the state where board isn't responding... Wireshark shows me that the USB descriptors can still be read, but the HID interrupts receive a report with status -2 (on Linux, Wireshark reports as ENOENT), whereas normally the response status is 0 (ok).

[mluessi]

RE1 doesn't have an external pull-down on PA9. I can test if enabling the internal one makes a difference.