Linux + iPhone Continuity / iMessage / SMS
fix: single adapter machines can have a different dev index
Zack Bartel2fd85ba46ad6
diff --git a/docs/BLUETOOTH.md b/docs/BLUETOOTH.mdindex fdcfb30..d6bac0d 100644--- a/docs/BLUETOOTH.md+++ b/docs/BLUETOOTH.md@@ -856,6 +856,7 @@ checks the daemon does not make.|---|---|---|| `systemctl enable --now tether-btclass@hci0` hangs in `activating (start)` forever | `btmgmt` epolls its stdin before running the command, and epoll rejects the `/dev/null` systemd hands it, so it waits having done nothing | Update the unit -- it pipes into `btmgmt` now. On an older build, `sudo btmgmt class 4 8` by hand sets the class until `bluetoothd` restarts, see 2026-09-01 below || `systemctl enable --now tether-btclass@hci0` says `Unit [email protected] does not exist` | A portable build (AppImage, Flatpak) installed no unit -- only the distro packages do | `tether --bt-setup` and run the command it prints, which writes the unit first, see 2026-09-05 below |+| `tether-btclass@hci0` fails with `Invalid Index`, having worked for months | The controller re-enumerated -- a failed firmware handshake resets it over USB and it comes back as `hci1` -- so the instance name points at an adapter that no longer exists | Update the unit -- it resolves the adapter itself now and the instance name is only a hint, see 2026-09-11 below. By hand, `ls /sys/class/bluetooth` names the live adapter || Messages and contacts worked, then stopped, and the error mentions a service record | `bluetoothd` restarted and reset the Class of Device | `sudo systemctl enable --now tether-btclass@hci0`, then re-pair if the phone dropped the bond || The phone never offers notifications / Sync Contacts | The class is wrong, or the ANCS advertisement is not running | Check for `class=ok` in `tether --bt-status`, can take minutes || MAP or PBAP reports `forbidden` | The matching toggle on the phone is off | Turn it on. This is not a pairing failure |@@ -3232,3 +3233,35 @@ Not yet verified on hardware: the full sequence on Pro 3, iPhone media rather ththe owner bit on Pro 2, and whether an `off` profile persists if tetherd dies while the budsare away. WirePlumber restores a saved `off` profile unless `session.dont-restore-off-profile`is set.++### 2026-09-11 - The Class of Device unit was pinned to an index the adapter had left++`tether-btclass@hci0` started failing with `Invalid Index` on a machine where it had worked+for months. Nothing in Tether had changed. The kernel log has the whole story:++```+Bluetooth: hci0: Execution of wmt command timed out+Bluetooth: hci0: Failed to send wmt func ctrl (-110)+usb 3-5: reset high-speed USB device number 4 using xhci_hcd+Bluetooth: hci1: Device setup in 1693823 usecs+```++The MediaTek controller's firmware handshake timed out, the USB device was reset, and the+controller came back under the next free index. `hci0` was gone; `/sys/class/bluetooth` held+only `hci1`. A unit instantiated on the index cannot survive that, and the failure reads like+a missing adapter rather than a renamed one.++The instance name is now a hint rather than the target. Each attempt uses `%i` if+`/sys/class/bluetooth/%i` exists and otherwise takes the first adapter present, so an already+enabled `tether-btclass@hci0` keeps working under any index and every documented command+stays as written. Resolving inside the loop rather than once also covers a re-enumeration+that lands mid-loop. The retry window went from 10s to 30s: the reset above put the new+adapter on the bus 13s after boot, which the old window missed even with a correct name.++`$$` in `ExecStart` is a literal `$` -- systemd expands `$a` as one of its own variables and+would hand the shell an empty string.++Not covered: a controller that re-enumerates at runtime with no `bluetoothd` restart. Nothing+re-runs the unit there, and `bluetoothd` applies its own default class to the new adapter, so+the class is wrong until `bluetooth.service` restarts. A udev rule tagging adapter `add` with+`SYSTEMD_WANTS` would close that, and would also remove the manual enable step.diff --git a/nix/module.nix b/nix/module.nixindex c99d27a..68d2792 100644--- a/nix/module.nix+++ b/nix/module.nix@@ -17,12 +17,19 @@ letType = "oneshot";RemainAfterExit = true;# btmgmt epolls its stdin before it runs the command,+ # ${adapter} is a hint: a controller that re-enumerates comes back under+ # the next free index, so resolve the adapter on every attempt.ExecStart = pkgs.writeShellScript "tether-btclass-${adapter}" ''- for attempt in 1 2 3 4 5 6 7 8 9 10; do- echo | ${pkgs.bluez}/bin/btmgmt --index ${adapter} class 4 8 >/dev/null 2>&1- if echo | ${pkgs.bluez}/bin/btmgmt --index ${adapter} info 2>/dev/null \- | ${pkgs.gnugrep}/bin/grep -q "class 0x..0408"; then- exit 0+ for attempt in $(${pkgs.coreutils}/bin/seq 30); do+ hci=${adapter}+ [ -e /sys/class/bluetooth/$hci ] \+ || hci=$(${pkgs.coreutils}/bin/ls /sys/class/bluetooth 2>/dev/null | ${pkgs.coreutils}/bin/head -n1)+ if [ -n "$hci" ]; then+ echo | ${pkgs.bluez}/bin/btmgmt --index $hci class 4 8 >/dev/null 2>&1+ if echo | ${pkgs.bluez}/bin/btmgmt --index $hci info 2>/dev/null \+ | ${pkgs.gnugrep}/bin/grep -q "class 0x..0408"; then+ exit 0+ fifi${pkgs.coreutils}/bin/sleep 1donediff --git a/packaging/systemd/[email protected] b/packaging/systemd/[email protected]index 96a8e99..c7f2f2d 100644--- a/packaging/systemd/[email protected]+++ b/packaging/systemd/[email protected]@@ -3,8 +3,9 @@# bluetoothd resets the class to its own default every time it# starts, and when it does, MAP and PBAP stop working with an OBEX error.#-# %i is the adapter, e.g. [email protected].-Description=Set Bluetooth Class of Device to A/V Hands-Free on %i for Tether+# %i is a hint, not a requirement. Contollers can fail handshake and comes back+# under the next free index, so the instance name is wrong while the machine still has exactly one adapter.+Description=Set Bluetooth Class of Device to A/V Hands-Free for TetherDocumentation=https://github.com/zackb/tether/blob/main/docs/BLUETOOTH.mdAfter=bluetooth.servicePartOf=bluetooth.service@@ -13,9 +14,13 @@ PartOf=bluetooth.serviceType=oneshotRemainAfterExit=yes# btmgmt epolls its stdin before it runs the command, rejects /dev/null with EPERM.-ExecStart=/bin/sh -c 'for i in 1 2 3 4 5 6 7 8 9 10; do \- echo | btmgmt --index %i class 4 8 >/dev/null 2>&1; \- echo | btmgmt --index %i info 2>/dev/null | grep -q "class 0x..0408" && exit 0; \+ExecStart=/bin/sh -c 'for t in $$(seq 30); do \+ a=%i; \+ [ -e /sys/class/bluetooth/$$a ] || a=$$(ls /sys/class/bluetooth 2>/dev/null | head -n1); \+ if [ -n "$$a" ]; then \+ echo | btmgmt --index $$a class 4 8 >/dev/null 2>&1; \+ echo | btmgmt --index $$a info 2>/dev/null | grep -q "class 0x..0408" && exit 0; \+ fi; \sleep 1; \done; exit 1'TimeoutStartSec=60