pl2303 built out of tree for the IMU's PL2303GC adapter; BOM and rebuild procedure

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

docs/BOM.md

@@ -181,8 +181,10 @@ Header facts (carrier board spec SP-11324-001 v1.3, Table 3-3 and 5-3; raw copy
 - Do not zero the accelerometer at startup the way smart-serow's `imu_calibrate()` does: that removes gravity, which tilt and grade need.
 - The serial device belongs to `dialout` (the header's `/dev/ttyTHS1` and USB adapters alike); a service gets it with
   `SupplementaryGroups=dialout`, not by adding users to the group.
-- **USB-UART adapter:** JetPack's kernel has drivers for CP210x and FTDI only (no PL2303, no CH341). In use: a CP2102 (`10c4:ea60`,
+- **USB-UART adapter: a PL2303GC (`067b:23a3`)**, the one the IMU enclosure is designed around (2026-10-14). JetPack's kernel has
-  generic serial `0001`, so it can only be told apart by port). A PL2303GC (`067b:23a3`) enumerates but gets no driver.
+  drivers for CP210x and FTDI only, so `pl2303` is built out of tree in `~/infra/pl2303/` ([report](impl/reports/PL2303-DRIVER.md));
+  rebuild it after a kernel update. A CP2102 (`10c4:ea60`, generic serial `0001`) was used for the first test and works with no
+  build. Neither has a unique serial: the `imu` service finds the WT61 by trying each adapter, so they are interchangeable.
 - The temperature field is the chip's own temperature, not the air's: `raw / 340 + 36.53` °C per the datasheet (smart-serow's `IMU.md`
   says 36.25).
 - **The gyro zeroes itself at power-up**, and yaw resets to 0. The module must be still while it powers up; it is, if it powers with the

docs/impl/reports/PL2303-DRIVER.md

@@ -0,0 +1,41 @@
+---
+type: report
+date: 2026-10-14
+---
+# PL2303 USB-serial driver, built out of tree
+
+The WT61 plus (the IMU) sits behind a USB-UART adapter, and its enclosure is designed
+around a **PL2303GC** (`067b:23a3`). JetPack's kernel ships `CONFIG_USB_SERIAL_PL2303`
+unset: of the plain USB-UART drivers it has only `cp210x` and `ftdi_sio`. The adapter
+enumerated but got no `ttyUSB`. Built the way `qmi_wwan` is ([SIM7600-SETUP](SIM7600-SETUP.md#rebuilding-qmi_wwan-after-a-kernel-update)).
+
+## What is installed
+
+```
+~/infra/pl2303/
+├── Makefile    # obj-m += pl2303.o ; `make install` → /lib/modules/$(uname -r)/misc/ + depmod
+├── pl2303.c    # linux-stable v5.15.148 drivers/usb/serial/pl2303.c
+└── pl2303.h    # linux-stable v5.15.148 drivers/usb/serial/pl2303.h (has PL2303_PRODUCT_ID_GC 0x23a3)
+```
+
+`/lib/modules/5.15.148-tegra/misc/pl2303.ko`, vermagic `5.15.148-tegra SMP preempt mod_unload modversions aarch64`,
+depends on the in-tree `usbserial`. After `depmod` it loads through the USB modalias when the adapter is plugged
+in: no `/etc/modules` entry. Built as the user against `/lib/modules/$(uname -r)/build`; only `make install`
+needs sudo.
+
+The adapter's serial number is not unique, and nothing names it: svc-core's `imu` service finds the WT61 by
+trying every USB-UART adapter (`cp210x`, `pl2303`, `ftdi_sio`, `ch341`), never the modem's `option` ports.
+
+## Rebuilding after a kernel update
+
+Like `pcan` and `qmi_wwan`, it does not survive a kernel update. Sign: the adapter is in `lsusb` but there is
+no new `ttyUSB`, and `yukart/system/imu` says `no WT61 on []`.
+
+```
+cd ~/infra/pl2303
+# if the kernel's upstream base version changed, refresh the sources:
+#   V=v5.15.xxx; for f in pl2303.c pl2303.h; do
+#     curl -fL -o $f "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/plain/drivers/usb/serial/$f?h=$V"; done
+make clean && make && sudo make install
+sudo modprobe pl2303
+```