ADB in the car: the PD hub causes the NCM packet loss; adb push stays slow without it

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DDwfd6Ki8UfG4hK6W5YpJz

docs/ADB-IN-CAR.md

@@ -106,7 +106,20 @@ Tested with the tablet on the Orin, charging through the PD hub:
   165 MB debug APK install failed, and so will `flutter run`'s first install.
   The app's own traffic (small, 5 Hz) is unaffected so far.
 
-Until the link loss is understood, install at the desk over USB, and use the car
+**The PD hub causes the packet loss.** With the tablet on a plain C-to-C cable
-path for `logcat`, `shell` and screenshots. Suspects, untested: the NCM gadget's
+straight to the Orin, ping loss is 0% at every size, even at 100 packets/s, and a
-packet aggregation on the Orin's 5.15 kernel, the PD hub, USB power management on
+raw one-way TCP transfer Orin → tablet runs at **~4.3 MB/s** with 4 retransmits
-either side. The gadget also offers RNDIS, which would be the first comparison.
+in 20 MB.
+
+**adb push is still slow without the hub: ~0.14 MB/s, 175 retransmits in 20 MB**,
+against 4 for the raw transfer. Cause not found. A bulk transfer running beside a
+ping burst loses nothing, so it is not simply traffic in both directions.
+
+Until both are understood, install at the desk over USB, and use the car path for
+`logcat`, `shell` and screenshots. The car still needs the PD hub to charge the
+tablet, so the hub, or its replacement, needs its own test.
+
+A test trap: commands started from `adb shell` on the tablet use Android's
+default network (WiFi), not the USB link. `nc 192.168.55.1 …` from there gets
+no route. Only sockets bound to the Ethernet network, as the app's are, take the
+link.