2026-07-22 22:53:28 +02:00
#!/bin/sh
# SPDX-License-Identifier: GPL-3.0-only
# SPDX-FileCopyrightText: Copyright (C) 2026 Catcrafts®
2026-09-01 18:10:04 +02:00
# ims-pdn-up.sh — boot bring-up for the `ims` PDN (journal/ims.md s45).
2026-07-22 22:53:28 +02:00
#
2026-09-01 18:10:04 +02:00
# Finds-or-creates the ims bearer via ModemManager, connects it, and
2026-07-22 22:53:28 +02:00
# configures the muxed netdev with the MM-assigned address (the s34 recipe,
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
# automated), reads the P-CSCF list the network put in the PDN's PCO from the
# modem (imsd-pcscf) unless PCSCF= names one, and gives the configured or
# the first discovered address (the daemon's fresh target) a host route —
# never a default route (F15 below). Idempotent — safe to run when the PDN
# is already up. Runs as
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
# ExecStartPre of imsd.service, so imsd starts once the PDN exists — or once
# this script has written IMSD_IDLE= and exited 0 with nothing to do, when
# imsd logs the idle and exits cleanly (idle_exit below); a nonzero exit
# fails the unit and systemd retries per Restart/RestartSec.
2026-09-01 18:10:04 +02:00
#
# Configuration, via the unit's EnvironmentFile /etc/imsd.env (also read
# directly so manual runs behave the same):
2026-09-17 13:45:19 +02:00
# IMS_APN ims APN name (default ims)
2026-09-01 18:10:04 +02:00
# IMS_IP_TYPE bearer ip-type (default ipv6)
2026-09-17 13:45:19 +02:00
# IMS_PROFILE_ID 3GPP profile index to connect through (default: looked up
# by APN in the modem's profile list; empty/none = connect by
# APN string). A profile-indexed call is exempt from the
# modem's attach-family forcing that refuses an APN-string
# IPv6 request when the attach PDN was granted IPv4-only
# (journal/ims 2026-09-16/17 bench: Odido, O2 UK, Swisscom).
2026-09-01 18:10:04 +02:00
# IMS_REG_TIMEOUT max seconds to wait for network registration before
# connect attempts start counting anyway (default 300)
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
# PCSCF one P-CSCF address, by hand: it wins over discovery
# (which is then skipped) and gets the host route; unset,
# imsd-pcscf reads the PDN's PCO list from the modem and
# this script hands it to imsd as PCSCF_DISCOVERED= in
# /run/imsd.env, the first of them routed
# IMSD_PCSCF the reader's path (default /usr/libexec/imsd-pcscf) — a
# bench seam, from the environment only (not from the file)
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
# IMSD_ENV the configuration file's path (default /etc/imsd.env) —
# a bench seam. The file is optional (overrides only); its
# existence changes nothing
2026-07-22 22:53:28 +02:00
log( ) { echo " ims-pdn-up: $* " ; }
kv( ) { mmcli " $@ " -K 2>/dev/null; }
2026-09-01 18:10:04 +02:00
# mmcli's multi-line error text as one journal-friendly line
squash( ) { printf '%s' " $1 " | tr '\n' ' ' | sed 's/ */ /g' ; }
modem_state( ) { kv -m " $MODEM " | sed -n 's/^modem\.generic\.state *: *//p' ; }
packet_state( ) { kv -m " $MODEM " | sed -n 's/^modem\.3gpp\.packet-service-state *: *//p' ; }
2026-07-22 22:53:28 +02:00
bearer_paths( ) {
kv -m " $MODEM " | sed -n 's/^modem\.generic\.bearers\.value\[[0-9]*\] *: *//p'
}
2026-10-06 17:54:09 +02:00
# A disabled modem has its radio off (airplane mode): leave it alone. A bearer
# connect on it fails with no-service and leaves ModemManager's state
# `enabled` with the modem still in low power; switching the radio back on
# then no longer brings the modem up (dev phone 2026-10-06).
radio_off( ) { [ " $( modem_state) " = disabled ] ; }
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
# A locked SIM (a PIN not entered yet) is the same kind of state the user
# controls: the modem registers once the PIN is in, so like a disabled radio
# it exits non-zero and the unit's restart tries again.
sim_locked( ) { [ " $( modem_state) " = locked ] ; }
unlock_required( ) { kv -m " $MODEM " | sed -n 's/^modem\.generic\.unlock-required *: *//p' ; }
2026-10-06 17:54:09 +02:00
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
IMSD_ENV = ${ IMSD_ENV :- /etc/imsd.env }
2026-09-01 18:10:04 +02:00
envval( ) { # $1 = key — for manual runs; under systemd the vars are inherited
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
sed -n " s/^ $1 =//p " " $IMSD_ENV " 2>/dev/null | tail -n1 | tr -d '"'
2026-09-01 18:10:04 +02:00
}
IP_TYPE = ${ IMS_IP_TYPE :- $( envval IMS_IP_TYPE) }
IP_TYPE = ${ IP_TYPE :- ipv6 }
2026-09-17 13:45:19 +02:00
IMS_APN = ${ IMS_APN :- $( envval IMS_APN) }
IMS_APN = ${ IMS_APN :- ims }
PROFILE_ID = ${ IMS_PROFILE_ID :- $( envval IMS_PROFILE_ID) }
2026-09-01 18:10:04 +02:00
REG_TIMEOUT = ${ IMS_REG_TIMEOUT :- $( envval IMS_REG_TIMEOUT) }
REG_TIMEOUT = ${ REG_TIMEOUT :- 300 }
2026-07-22 22:53:28 +02:00
# ---- wait for a modem (MM + modem firmware take a while after boot)
n = 0
while :; do
MODEM = $( mmcli -L 2>/dev/null | sed -n 's,.*/Modem/\([0-9]*\).*,\1,p' | head -n1)
[ -n " $MODEM " ] && break
n = $(( n + 1 ))
[ " $n " -ge 60 ] && { log "no modem after 120 s" ; exit 1; }
sleep 2
done
log " modem $MODEM "
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
# ---- nothing to do: a clean exit, not a retry loop. imsd reads IMSD_IDLE=
# from /run/imsd.env, logs it and exits 0, which Restart=on-failure leaves
# alone — the unit stays inactive until the next boot, a `systemctl start
# imsd` or a start of imsd-ofonod (its Wants=imsd.service, the Phosh image).
# Installing the package and inserting a SIM is the opt-in; no configuration
# file changes what follows. The dead ends are facts: no SIM (the modem's
# `failed` state with a SIM reason; final until ModemManager re-probes;
# checked here and while waiting for registration below), a profile-indexed
# connect series refused ten times in a row — each a persistent 3GPP cause
# in ModemManager's typed error (not subscribed, unknown APN, barred, not
# authorised in this PLMN, authentication failed, the service option not
# supported, the wrong PDN type), with packet service; counted only while the
# modem reads registered, connecting or connected with packet service not
# detached — any other read, a timeout or a transient cause starts the count
# again — and — the daemon's own, with the same clean exit — a
# bring-up that ends with no P-CSCF (none configured, the PCO names none)
# and no registration of imsd's own to resume. One attempt per boot (or per
# such imsd-ofonod start). What is not a fact exits 1 like a disabled radio
# and the restart retries: ten failures without coverage, a SIM still waiting
# for its PIN, a modem without an ims profile (the APN-string connect is
# tried), an APN-string series (the modem's own wall under an IPv4 attach,
# F6), a P-CSCF reader that did not answer (PCSCF_DISCOVERY=failed, below).
idle_exit( ) { # $1 = key, $2 = log text
log " $2 "
printf 'IMSD_IDLE=%s\n' " $1 " > /run/imsd.env
log " wrote /run/imsd.env: IMSD_IDLE= $1 "
exit 0
}
sim_check( ) {
[ " $( modem_state) " = failed ] || return 0
REASON = $( kv -m " $MODEM " | sed -n 's/^modem\.generic\.state-failed-reason *: *//p' )
case " $REASON " in
sim-missing| esim-without-profiles| sim-error)
idle_exit no-sim " no SIM ( $REASON ): not bringing the ims PDN up — systemctl start imsd once one is in " ; ;
esac
}
sim_check
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
# ---- the ims profile index (WDS profile list), unless configured. The
# lookup itself can fail while ModemManager is still enabling the modem; a
# failed lookup is not "no profile" (the APN-string request it would fall
# back to is the one the modem refuses under an IPv4 attach), so it is
2026-09-24 18:58:59 +02:00
# retried after the registration wait; a list without the APN is final —
# the APN string is used — only once the modem is registered.
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
LOOKUP = pending
if [ -n " $PROFILE_ID " ] ; then
[ " $PROFILE_ID " = none ] && PROFILE_ID =
LOOKUP = done
2026-09-17 13:45:19 +02:00
fi
2026-09-24 18:58:59 +02:00
lookup_profile( ) { # $1 = modem state, $2 = log suffix; sets PROFILE_ID, LOOKUP=done on a final answer
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
[ " $LOOKUP " = done ] && return 0
if OUT = $( qmicli -d qrtr://0 --wds-get-profile-list= 3gpp 2>& 1) ; then
2026-09-20 16:58:43 +02:00
# a list with no profiles at all is the modem still coming up, not
# a modem without an ims profile (dev phone cold boot 2026-09-20
# 16:47: exit 0, no entries, index 2 present a minute later)
if ! printf '%s\n' " $OUT " | grep -q -E '^[[:space:]]*\[[0-9]+\] 3gpp' ; then
2026-09-24 18:58:59 +02:00
log " ims profile lookup returned no profiles $2 "
2026-09-20 16:58:43 +02:00
return 0
fi
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
PROFILE_ID = $( printf '%s\n' " $OUT " | awk -v apn = " $IMS_APN " '
/^[ \t ] *\[ [ 0-9] +\] 3gpp/ { idx = $1 ; gsub( /[ ^0-9] /, "" , idx) }
/APN:/ { a = $0 ; sub( /.*APN: '"' "'/, " ", a); sub(/'" '"' .*/, "" , a) ;
if ( tolower( a) = = tolower( apn) && idx != "" ) { print idx; exit } } ' )
2026-09-24 18:58:59 +02:00
if [ -n " $PROFILE_ID " ] ; then
LOOKUP = done
log " ims profile: index $PROFILE_ID (apn $IMS_APN ) "
return 0
fi
# a list without the APN is final only once the modem is registered:
# while it enables, the list can be partial (dev phone cold boot
# 2026-09-24 18:18: entries but no ims at `enabling`, ims listed when
# read minutes later), and on a walled carrier the APN-string fallback
# that would follow is the request the modem refuses
case " $1 " in
registered| connecting| connected)
LOOKUP = done
log " ims profile: none for apn $IMS_APN , connecting by APN string " ; ;
*) log " ims profile: not in the list yet $2 " ; ;
esac
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
else
2026-09-24 18:58:59 +02:00
log " ims profile lookup failed $2 : $( squash " $OUT " ) "
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
fi
}
2026-09-24 18:58:59 +02:00
STATE = $( modem_state)
lookup_profile " $STATE " " (modem state: $STATE ) "
2026-09-17 13:45:19 +02:00
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
# ---- find a connected ims bearer; else find-or-create one and connect it.
# APN names are case-insensitive. A DISCONNECTED bearer is reused only when
# it is the request this run would make (same profile index, or same APN when
# connecting by APN string): a leftover APN-string bearer from a run whose
# lookup failed is the walled request itself, and reconnecting it fails on
# every restart until the modem is reset (field: O2 UK, 2026-09-20).
has( ) { printf '%s\n' " $INFO " | grep -qi " ^bearer\. $1 *: * $2 \$ " ; }
find_ims_bearer( ) { # $1 = required bearer.status.connected value; $2 = "relaxed" ignores ip-type
2026-07-22 22:53:28 +02:00
for B in $( bearer_paths) ; do
INFO = $( kv -b " $B " ) || continue
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
has status.connected " $1 " || continue
2026-09-17 13:45:19 +02:00
if [ -n " $PROFILE_ID " ] ; then
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
if [ " $1 " = yes ] ; then
has properties.profile-id " $PROFILE_ID " || has properties.apn " $IMS_APN " || continue
else
has properties.profile-id " $PROFILE_ID " || continue
fi
2026-09-17 13:45:19 +02:00
else
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
has properties.apn " $IMS_APN " || continue
2026-09-17 13:45:19 +02:00
fi
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
[ " $2 " = relaxed ] || has properties.ip-type " $IP_TYPE " || continue
2026-07-22 22:53:28 +02:00
echo " $B "
return 0
done
return 1
}
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
# leftover disconnected APN-string ims bearers when this run connects by
# profile index: never reused (above), deleted so a later run without an
# index cannot pick one up either
delete_stale_bearers( ) {
[ -n " $PROFILE_ID " ] || return 0
for B in $( bearer_paths) ; do
INFO = $( kv -b " $B " ) || continue
has status.connected no || continue
has properties.apn " $IMS_APN " || continue
has properties.profile-id " $PROFILE_ID " && continue
OUT = $( mmcli -m " $MODEM " --delete-bearer= " $B " 2>& 1) &&
log " deleted stale bearer $B (apn $IMS_APN , no profile index) " ||
log " could not delete stale bearer $B : $( squash " $OUT " ) "
done
}
2026-07-22 22:53:28 +02:00
BEARER = $( find_ims_bearer yes)
2026-09-01 18:10:04 +02:00
# ---- gate the connect attempts on network registration: the 10x10 s window
# below is shorter than some carriers' post-boot attach (measured ~2 min), so
# without this every attempt can fail on no-service and the retries end
# before the network is even attached
if [ -z " $BEARER " ] ; then
waited = 0 last =
while :; do
STATE = $( modem_state)
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
case " $STATE " in registered| connecting| connected) break ; ; failed) sim_check ; ; esac
2026-09-01 18:10:04 +02:00
[ " $STATE " != " $last " ] && log " waiting for registration (state: ${ STATE :- unknown } ) "
last = $STATE
if [ " $waited " -ge " $REG_TIMEOUT " ] ; then
2026-10-06 17:54:09 +02:00
if [ " $STATE " = disabled ] ; then
log " modem still disabled after $REG_TIMEOUT s (radio off): not connecting "
exit 1
fi
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
if [ " $STATE " = locked ] ; then
log " SIM locked ( $( unlock_required) ) after $REG_TIMEOUT s: not connecting until the PIN is entered "
exit 1
fi
2026-09-01 18:10:04 +02:00
log " not registered after $REG_TIMEOUT s — attempting anyway "
break
fi
sleep 5
waited = $(( waited + 5 ))
done
log " modem state: $( modem_state) , packet service: $( packet_state) "
2026-09-24 18:58:59 +02:00
# a lookup with no final answer yet: it failed, or its list lacked the
# APN while the modem was still coming up
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
n = 0
while [ " $LOOKUP " != done ] && [ " $n " -lt 3 ] ; do
n = $(( n + 1 ))
sleep 5
2026-09-24 18:58:59 +02:00
STATE = $( modem_state)
lookup_profile " $STATE " " (retry $n , modem state: $STATE ) "
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
done
if [ " $LOOKUP " != done ] ; then
PROFILE_ID =
log "ims profile: lookup keeps failing, connecting by APN string"
fi
BEARER = $( find_ims_bearer yes) # the index may now match a connected profile-indexed PDN
[ -n " $BEARER " ] || delete_stale_bearers
2026-09-01 18:10:04 +02:00
fi
2026-07-22 22:53:28 +02:00
n = 0
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
REFUSED = 0 # connects refused in a row, each a persistent 3GPP cause with the modem registered/connecting/connected and packet service not detached (stricter than the daemon's PacketServiceUp)
2026-07-22 22:53:28 +02:00
while [ -z " $BEARER " ] ; do
n = $(( n + 1 ))
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
if [ " $n " -gt 10 ] ; then
# only a profile-indexed series the network refused says anything
# about it: an APN-string series (a lookup that kept failing, a late
# registration) meets the modem's own wall under an IPv4 attach (F6),
# which the restart's fresh lookup gets past, and a failure without
# packet service is a coverage gap, not a refusal — counted only while
# the modem reads registered, connecting or connected with packet
# service not `detached` (a CS-registered modem still attaching) and
# only for a persistent 3GPP cause in ModemManager's typed error; any
# other read, `disconnecting`, `unknown` or an unanswered mmcli among
# them, and a timeout or a transient cause, reset it — stricter than
# the daemon's PacketServiceUp, which counts an unknown state as
# service — so the
# count is of refusals in a row, each with packet service at the time,
# never one state read after the series: a series that failed without
# coverage and found the network at its end exits 1 like any other,
# and the restart retries on the attached modem
[ -z " $PROFILE_ID " ] || [ " $REFUSED " -lt 10 ] || idle_exit no-ims-pdn " bearer connect refused 10 times in a row by profile index with packet service and a persistent cause (last: $( squash " $OUT " ) ): no ims PDN for this SIM here — systemctl start imsd to try again "
log " bearer connect failed after 10 attempts (modem state: $( modem_state) , last: $( squash " $OUT " ) ) " ; exit 1
fi
2026-10-06 17:54:09 +02:00
radio_off && { log "modem disabled (radio off): not connecting" ; exit 1; }
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
sim_locked && { log " SIM locked ( $( unlock_required) ): not connecting until the PIN is entered " ; exit 1; }
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
B = $( find_ims_bearer no) # reuse a stale disconnected ims bearer of this run's shape
2026-07-22 22:53:28 +02:00
if [ -z " $B " ] ; then
2026-09-17 13:45:19 +02:00
if [ -n " $PROFILE_ID " ] ; then
SPEC = " profile-id= $PROFILE_ID ,ip-type= $IP_TYPE "
else
SPEC = " apn= $IMS_APN ,ip-type= $IP_TYPE "
fi
OUT = $( mmcli -m " $MODEM " --create-bearer= " $SPEC " 2>& 1)
2026-09-01 18:10:04 +02:00
B = $( printf '%s' " $OUT " | sed -n 's,.*\(/org/freedesktop/ModemManager1/Bearer/[0-9]*\).*,\1,p' )
if [ -n " $B " ] ; then
2026-09-17 13:45:19 +02:00
log " created bearer $B ( $SPEC ) "
2026-09-01 18:10:04 +02:00
else
log " create-bearer attempt $n failed: $( squash " $OUT " ) ; retrying in 10 s "
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
REFUSED = 0 # not the network's answer to a connect
2026-09-01 18:10:04 +02:00
sleep 10
continue
fi
2026-07-22 22:53:28 +02:00
fi
2026-09-01 18:10:04 +02:00
if OUT = $( mmcli -b " $B " --connect 2>& 1) ; then
2026-07-22 22:53:28 +02:00
BEARER = $B
else
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
ST = $( modem_state) ; PS = $( packet_state)
log " connect attempt $n failed (state: ${ ST :- unknown } , packet service: ${ PS :- unknown } ): $( squash " $OUT " ) ; retrying in 10 s "
# a refusal is the network's answer, as ModemManager types it: a
# persistent 3GPP cause — not subscribed, unknown APN, barred, not
# authorised in this PLMN, authentication failed, the service option
# not supported, the wrong PDN type for this attach — with the modem
# attached. A timeout (D-Bus, esm-proc-timeout), a transient cause
# (insufficient resources, out of order, a network failure) or any
# other error starts the count again, as a read without packet service
# does: an outage must not idle VoLTE until the next boot
PERSISTENT =
case " $OUT " in
*MobileEquipment.ServiceOptionNotSubscribed:*| *MobileEquipment.MissingOrUnknownApn:*| *MobileEquipment.OperatorDeterminedBarring:*| *MobileEquipment.ServiceOptionNotAuthorizedInPlmn:*| *MobileEquipment.UserAuthenticationFailed:*| *MobileEquipment.ServiceOptionNotSupported:*| *MobileEquipment.Ipv4OnlyAllowed:*| *MobileEquipment.Ipv6OnlyAllowed:*) PERSISTENT = 1 ; ;
esac
case " $ST " in
registered| connecting| connected) if [ " $PS " = detached ] || [ -z " $PERSISTENT " ] ; then REFUSED = 0; else REFUSED = $(( REFUSED + 1 )) ; fi ; ;
*) REFUSED = 0 ; ;
esac
ims-pdn-up: retry the profile lookup, never reuse a stale APN-string bearer
The ims profile lookup ran the moment ModemManager listed the modem, which
on a cold boot can be while it is still enabling; qmicli's error was
discarded, so a failed lookup read as "no profile" and the script fell back
to the APN-string request, which is the one the modem refuses under an
IPv4 attach. Every refused attempt left a disconnected apn=ims bearer
object behind, and the next restart's "reuse a stale disconnected bearer"
step matched it by APN and reconnected it: ten refusals again, no way out
but a reboot that happens to win the race (O2 UK field report; the same
lost race seen on a KPN unit, where the fallback merely connects).
Log qmicli's error and retry the lookup after the registration wait, fall
back to the APN string only when the list has no such APN, reuse a
disconnected bearer only when it has this run's shape and delete leftover
APN-string ims bearers when connecting by profile index, match the APN
case-insensitively, and on call-already-present adopt the connected bearer
instead of asking the modem again.
Ten field shapes replayed against a mock ModemManager (shipped script
5/10, this one 10/10); on the FP6 a warm restart adopts the live PDN and
a cold boot connects by profile index and registers.
2026-09-20 16:07:46 +02:00
case " $OUT " in *call-already-present*)
# the PDN is up on a bearer this run did not recognise (ip-type or
# APN spelling); use it rather than ask the modem for a second one
if B = $( find_ims_bearer yes relaxed) ; then
log " adopting the connected bearer $B "
BEARER = $B
continue
fi ; ;
esac
2026-07-22 22:53:28 +02:00
sleep 10
fi
done
log " connected: $BEARER "
# ---- configure the muxed netdev with the MM-assigned address
INFO = $( kv -b " $BEARER " )
IFACE = $( echo " $INFO " | sed -n 's/^bearer\.status\.interface *: *//p' )
ADDR = $( echo " $INFO " | sed -n 's/^bearer\.ipv6-config\.address *: *//p' )
PREFIX = $( echo " $INFO " | sed -n 's/^bearer\.ipv6-config\.prefix *: *//p' )
if [ -z " $IFACE " ] || [ -z " $ADDR " ] ; then
log "bearer up but no interface/address in mmcli output"
exit 1
fi
ims-pdn-up: never let the ims PDN be the IPv6 default route
The PGW's router advertisement on the ims netdev offers a default
router, and on IPv4-only wifi that is the phone's only IPv6 default
route. Every AAAA-first client (apk, browsers) then sends its SYN into
the IMS walled garden, which drops it: apk sat in SYN-SENT for nine
minutes on the dev phone, and a field user's remedy was to stop imsd
before package work.
Set accept_ra_defrtr=0 on the muxed netdev before link-up (the router
solicitation goes out once the link-local address passes DAD), flush a
default route an earlier RA installed (an imsd restart), and give the
P-CSCF a host route over the netdev instead. imsd talks to the P-CSCF
only and already installs a host route per media peer; the RA's prefix
and MTU still apply. The host route makes the P-CSCF reachable before
the RA has been processed, so the PDN MTU ModemManager reports is set
on the link as well, the way NetworkManager does for its bearers. A
failed host route, or a P-CSCF that is not an IPv6 address (the PDN is
brought up IPv6-only), fails the unit with a clear message.
Dev phone (KPN, 0.3.8-r0): restart path drops the RA default and
registers over the host route, apk update in 6 s where it hung; MO and
MT calls with audio both ways; cold boot never installs the default
route, the link carries the 1280 MTU before the host route exists,
fresh registration in 1 s.
2026-09-24 04:12:39 +02:00
# ---- the ims PDN must never be the phone's IPv6 default route (F15). The
# PGW's router advertisement offers a default router; on IPv4-only wifi
# that is the phone's only v6 default, so every AAAA-first client (apk,
# browsers) sends its SYN into the IMS walled garden, which drops it
# silently (dev phone 2026-09-20: a 9-min apk hang). imsd only talks to the
# P-CSCF (host route below) and to media peers (imsd installs those host
# routes per call); the RA's prefix and MTU are still applied (its hop
# limit is not — the kernel default 64 stays). Per-netdev, via /proc, and
# before link-up: the router solicitation goes out once the link-local
# address passes DAD, ~1–
if ! echo 0 > " /proc/sys/net/ipv6/conf/ $IFACE /accept_ra_defrtr " 2>/dev/null; then
log " could not set accept_ra_defrtr=0 on $IFACE (continuing) "
fi
# the PDN's MTU as the modem reports it (1280 here), on the link itself:
# the P-CSCF host route makes the P-CSCF reachable before the RA's MTU
# option has been processed, and a send in that window would otherwise be
# sized by the link's default MTU. NetworkManager does the same for its
# bearers. An RA MTU no larger than the link's is still applied.
MTU = $( echo " $INFO " | sed -n 's/^bearer\.ipv6-config\.mtu *: *//p' )
case " $MTU " in '' | --| *[ !0-9] *) MTU = ; ; esac
if [ -n " $MTU " ] && [ " $MTU " -ge 1280 ] ; then
ip link set " $IFACE " mtu " $MTU " || log " could not set mtu $MTU on $IFACE (continuing) "
fi
2026-07-22 22:53:28 +02:00
ip link set " $IFACE " up || exit 1
ip -6 addr replace " $ADDR / ${ PREFIX :- 64 } " dev " $IFACE " || exit 1
# wait out IPv6 DAD: binding a tentative address gives EADDRNOTAVAIL
# (first boot attempt cost a 120 s systemd retry exactly this way)
n = 0
while ip -6 addr show dev " $IFACE " | grep -q tentative; do
n = $(( n + 1 ))
[ " $n " -ge 10 ] && { log "address still tentative after 10 s" ; break; }
sleep 1
done
ims-pdn-up: never let the ims PDN be the IPv6 default route
The PGW's router advertisement on the ims netdev offers a default
router, and on IPv4-only wifi that is the phone's only IPv6 default
route. Every AAAA-first client (apk, browsers) then sends its SYN into
the IMS walled garden, which drops it: apk sat in SYN-SENT for nine
minutes on the dev phone, and a field user's remedy was to stop imsd
before package work.
Set accept_ra_defrtr=0 on the muxed netdev before link-up (the router
solicitation goes out once the link-local address passes DAD), flush a
default route an earlier RA installed (an imsd restart), and give the
P-CSCF a host route over the netdev instead. imsd talks to the P-CSCF
only and already installs a host route per media peer; the RA's prefix
and MTU still apply. The host route makes the P-CSCF reachable before
the RA has been processed, so the PDN MTU ModemManager reports is set
on the link as well, the way NetworkManager does for its bearers. A
failed host route, or a P-CSCF that is not an IPv6 address (the PDN is
brought up IPv6-only), fails the unit with a clear message.
Dev phone (KPN, 0.3.8-r0): restart path drops the RA default and
registers over the host route, apk update in 6 s where it hung; MO and
MT calls with audio both ways; cold boot never installs the default
route, the link carries the 1280 MTU before the host route exists,
fresh registration in 1 s.
2026-09-24 04:12:39 +02:00
log " $IFACE up, $ADDR / ${ PREFIX :- 64 } , ipv6 mtu $( cat " /proc/sys/net/ipv6/conf/ $IFACE /mtu " 2>/dev/null) "
# a default route an earlier RA already installed (the link was up before
# this run: an imsd restart) — accept_ra_defrtr=0 leaves an existing one in
# place until its lifetime runs out. Any ::/0 via this netdev goes, by the
# policy above.
if [ -n " $( ip -6 route show default dev " $IFACE " 2>/dev/null) " ] ; then
ip -6 route flush default dev " $IFACE " && log " dropped the RA default route on $IFACE "
fi
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
# ---- the P-CSCF. PCSCF= when the admin set one (one IPv6 address; it wins
# and discovery is skipped), else the list the network put in the PDN's PCO,
# read from the modem by imsd-pcscf: a QMI WDS client bound to this netdev's
# mux data port — the binding ModemManager gives its own clients, and only
# such a client is shown the call's settings. The list goes to imsd as
# PCSCF_DISCOVERED= (its own name: this script inherits the previous run's
# /run/imsd.env through the unit, so a stale discovered value must never
# pass for the admin's PCSCF=, and it is never read from the environment
# here). The PDN is brought up IPv6-only, so only IPv6 addresses are usable;
# IPv4 ones and names are logged and dropped. Nothing usable is not this
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
# script's failure, and PCSCF_DISCOVERY= in the hand-off tells the daemon
# which kind it was: `none` — the PCO carried no IPv6 P-CSCF (the reader's
# exit 2), a fact: a registration of imsd's own that can be resumed is
# resumed and stays up, else the daemon idles (its `idle:` exit; after a
# resumable pair's failed attempt the fatal and the restart) — or `failed` —
# the reader did not answer (an error, a timeout, nothing printed), no fact:
# a daemon with nothing to resume exits 1 and the unit's restart runs
# discovery again. `found` and `skipped` (PCSCF= set, an IPv4-only PDN, no
# rmnet mux id) are the other two values.
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
# The P-CSCF is reachable through the ims PDN only: a host route stands in
# for the default — `dev` without a gateway (a raw-IP netdev without
# header_ops, so the kernel resolves no neighbours), the shape imsd uses for
# media peers; of a discovered list the first address — the daemon's fresh
# target — gets one here; imsd routes the P-CSCF it uses itself as well — the
# fresh target before a fresh registration, the resumed SA peer once its SAs
# are confirmed — with the same idempotent replace as a call's media peer, so
# a target this script did not route (a list set in /etc/imsd.env, a route
# lost with the PDN) is reachable.
# Anything else on the IMS network (its DNS servers, a traceroute
# target) needs its own host route. Without the route imsd cannot reach the
# P-CSCF, so a route failure is the unit's failure.
route_pcscf( ) { # $1 = address
OUT = $( ip -6 route replace " $1 /128 " dev " $IFACE " 2>& 1) ||
{ log " host route to P-CSCF $1 failed: $( squash " $OUT " ) " ; exit 1; }
log " host route to P-CSCF $1 dev $IFACE "
}
ims-pdn-up: never let the ims PDN be the IPv6 default route
The PGW's router advertisement on the ims netdev offers a default
router, and on IPv4-only wifi that is the phone's only IPv6 default
route. Every AAAA-first client (apk, browsers) then sends its SYN into
the IMS walled garden, which drops it: apk sat in SYN-SENT for nine
minutes on the dev phone, and a field user's remedy was to stop imsd
before package work.
Set accept_ra_defrtr=0 on the muxed netdev before link-up (the router
solicitation goes out once the link-local address passes DAD), flush a
default route an earlier RA installed (an imsd restart), and give the
P-CSCF a host route over the netdev instead. imsd talks to the P-CSCF
only and already installs a host route per media peer; the RA's prefix
and MTU still apply. The host route makes the P-CSCF reachable before
the RA has been processed, so the PDN MTU ModemManager reports is set
on the link as well, the way NetworkManager does for its bearers. A
failed host route, or a P-CSCF that is not an IPv6 address (the PDN is
brought up IPv6-only), fails the unit with a clear message.
Dev phone (KPN, 0.3.8-r0): restart path drops the RA default and
registers over the host route, apk update in 6 s where it hung; MO and
MT calls with audio both ways; cold boot never installs the default
route, the link carries the 1280 MTU before the host route exists,
fresh registration in 1 s.
2026-09-24 04:12:39 +02:00
PCSCF = ${ PCSCF :- $( envval PCSCF) }
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
DISCOVERED =
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
DISCOVERY = skipped
ims-pdn-up: never let the ims PDN be the IPv6 default route
The PGW's router advertisement on the ims netdev offers a default
router, and on IPv4-only wifi that is the phone's only IPv6 default
route. Every AAAA-first client (apk, browsers) then sends its SYN into
the IMS walled garden, which drops it: apk sat in SYN-SENT for nine
minutes on the dev phone, and a field user's remedy was to stop imsd
before package work.
Set accept_ra_defrtr=0 on the muxed netdev before link-up (the router
solicitation goes out once the link-local address passes DAD), flush a
default route an earlier RA installed (an imsd restart), and give the
P-CSCF a host route over the netdev instead. imsd talks to the P-CSCF
only and already installs a host route per media peer; the RA's prefix
and MTU still apply. The host route makes the P-CSCF reachable before
the RA has been processed, so the PDN MTU ModemManager reports is set
on the link as well, the way NetworkManager does for its bearers. A
failed host route, or a P-CSCF that is not an IPv6 address (the PDN is
brought up IPv6-only), fails the unit with a clear message.
Dev phone (KPN, 0.3.8-r0): restart path drops the RA default and
registers over the host route, apk update in 6 s where it hung; MO and
MT calls with audio both ways; cold boot never installs the default
route, the link carries the 1280 MTU before the host route exists,
fresh registration in 1 s.
2026-09-24 04:12:39 +02:00
case " $PCSCF " in
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
'' )
if [ " $IP_TYPE " = ipv4 ] ; then
log " P-CSCF discovery: not for IMS_IP_TYPE= $IP_TYPE — the bring-up configures only the PDN's IPv6 address "
else
MUX = $( ip -d link show " $IFACE " 2>/dev/null | sed -n 's/.*rmnet mux_id \([0-9]*\).*/\1/p' | head -n1)
if [ -z " $MUX " ] ; then
log " P-CSCF discovery: no rmnet mux id on $IFACE (ip -d link); set PCSCF= "
else
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
DISCOVERY = failed
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
OUT = $( " ${ IMSD_PCSCF :- /usr/libexec/imsd-pcscf } " --mux " $MUX " ${ PROFILE_ID : +--profile " $PROFILE_ID " } --timeout 10 2>& 1) ; rc = $?
V6 = ; SHOWN = ; OTHER = ; FLAGS = ; ERR = ; n = 0
while IFS = read -r line; do
case " $line " in
'' ) ; ;
v4\ *| fqdn\ *) OTHER = " ${ OTHER : + $OTHER } $line " ; ;
flag\ *) FLAGS = " ${ FLAGS : + $FLAGS , } ${ line #flag } " ; ;
*[ !0-9a-f:] *) ERR = " ${ ERR : + $ERR } $line " ; ;
*:*) [ " $n " -ge 8 ] && continue ; n = $(( n + 1 ))
V6 = " ${ V6 : + $V6 } $line " ; DISCOVERED = " ${ DISCOVERED : + $DISCOVERED , } $line " ; SHOWN = " ${ SHOWN : + $SHOWN , } $line " ; ;
*) ERR = " ${ ERR : + $ERR } $line " ; ;
esac
done <<EOF
$OUT
EOF
if [ -n " $V6 " ] ; then
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
DISCOVERY = found
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
log " P-CSCF discovered: $SHOWN ${ FLAGS : + ( $FLAGS ) } "
[ -n " $OTHER " ] && log " P-CSCF discovery also gave $OTHER (not usable on the IPv6-only ims PDN) "
route_pcscf " ${ V6 %% * } " # the first: the daemon's fresh target; imsd routes the P-CSCF it uses itself as well (the fresh target, the resumed peer)
else
case $rc in
0) log " P-CSCF discovery: nothing usable (imsd-pcscf printed no address ${ ERR : + : $ERR } ) " ; ;
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
2) DISCOVERY = none; log " P-CSCF discovery: nothing usable (the PCO carried no IPv6 P-CSCF ${ OTHER : +; it gave $OTHER } ${ FLAGS : +; $FLAGS } ) " ; ;
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
*) log " P-CSCF discovery: nothing usable (imsd-pcscf rc $rc ${ ERR : + : $ERR } ) " ; ;
esac
fi
fi
fi ; ;
*,*) log " PCSCF ' $PCSCF ' is a list; one address here (a list is PCSCF_DISCOVERED, the bring-up's) " ; exit 1 ; ;
ipsec, imsd, ims-pdn-up: the SA reader and the holds key compare in canonical form; LOCAL is canonicalised, PCSCF= accepted in RFC 5952 §4 form only
ParseExistingSa and SaLocals compared PCSCF= and LOCAL= as text against
`ip xfrm` output, which prints RFC 5952 form, so an upper-case or
uncompressed PCSCF= found no SAs and the restart registered fresh — on
KPN into the live-SA refusal; the holds file's pdn key compared the same
way. Both sides of each of those compares now go through
imsd::util::CanonicalAddr, LOCAL is canonicalised once on the way in, and PCSCF= is accepted in
that one spelling only — upper case, leading zeros or an uncompressed run
are refused at bring-up with the spelling to use, a dotted-quad tail (the
form ip prints for an IPv4-mapped address, which CanonicalIp6 does not read)
with the rule alone, a bracketed one by the bring-up script, which routes
the value as typed, with the rule (no brackets) — and a comma-separated
PCSCF= (a list) with its own message instead of reaching the socket layer: one spelling,
instead of a leniency both consumers of the setting would have to share.
Tested: crafter-build test, 7 passed; the upper-case PCSCF= refused with the spelling to use, a bracketed one
refused by the script, and the spelling they name resuming, in builds of
this series on the dev phone (KPN).
2026-10-11 15:55:19 +02:00
*\[ *) log " PCSCF ' $PCSCF ': write the bare address, as ip prints it (no brackets) " ; exit 1 ; ;
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
*:*) log " P-CSCF $PCSCF configured: discovery skipped " ; route_pcscf " $PCSCF " ; ;
ims-pdn-up: never let the ims PDN be the IPv6 default route
The PGW's router advertisement on the ims netdev offers a default
router, and on IPv4-only wifi that is the phone's only IPv6 default
route. Every AAAA-first client (apk, browsers) then sends its SYN into
the IMS walled garden, which drops it: apk sat in SYN-SENT for nine
minutes on the dev phone, and a field user's remedy was to stop imsd
before package work.
Set accept_ra_defrtr=0 on the muxed netdev before link-up (the router
solicitation goes out once the link-local address passes DAD), flush a
default route an earlier RA installed (an imsd restart), and give the
P-CSCF a host route over the netdev instead. imsd talks to the P-CSCF
only and already installs a host route per media peer; the RA's prefix
and MTU still apply. The host route makes the P-CSCF reachable before
the RA has been processed, so the PDN MTU ModemManager reports is set
on the link as well, the way NetworkManager does for its bearers. A
failed host route, or a P-CSCF that is not an IPv6 address (the PDN is
brought up IPv6-only), fails the unit with a clear message.
Dev phone (KPN, 0.3.8-r0): restart path drops the RA default and
registers over the host route, apk update in 6 s where it hung; MO and
MT calls with audio both ways; cold boot never installs the default
route, the link carries the 1280 MTU before the host route exists,
fresh registration in 1 s.
2026-09-24 04:12:39 +02:00
*) log " PCSCF ' $PCSCF ' is not an IPv6 address; the ims PDN is brought up IPv6-only " ; exit 1 ; ;
esac
2026-07-22 22:53:28 +02:00
ims-pdn-up: discover the P-CSCF list once the PDN is up and hand it to imsd
With the netdev configured, the script runs imsd-pcscf against the
netdev's rmnet mux id (and the ims profile index when it knows one) and
hands the IPv6 addresses it prints to imsd as PCSCF_DISCOVERED= in
/run/imsd.env and gives the first — the daemon's fresh target — a host
route (the SA peer of a registration imsd resumes is routed by imsd itself,
as the configured address's arm already left it); IPv4 addresses and names
are logged and dropped (the PDN is IPv6-only), at most eight are handed
over, and nothing usable is logged with its cause and is not the
script's failure: imsd resumes a registration of its own when it can, else
says so and exits, and the restart runs discovery again. PCSCF= still wins and skips discovery; a list in it is refused.
The script does not read the variable from its environment: the unit
hands it the previous run's file, and a stale discovered value must not
pass for the admin's setting. The reader's path is the IMSD_PCSCF seam
the mock bench uses.
Tested: on the mock bench against this commit's script (the twelve 0.3.10
scenarios; this step's discovery arms S13–S17 and S19–S22, with S18 the
pre-existing non-IPv6 refusal; S55, the reader's rc-0 answer without an
address), and in builds of this series on the dev phone (KPN), whose PCO —
two IPv6 addresses, nothing else — reaches the plain found arm; the
configured skip, the bracketed refusal and, through the IMSD_PCSCF seam, the
reader's failure and its exit 2 ran on the phone too; the v4/FQDN line, the
eight-address cap, the IPv4-only and no-mux-id skips, the comma-list refusal
and the rc-0 arm on the mock only. On a tip build of this series (daemon
2ec1a02f62831edb; bench 40, 2026-10-13) a fresh registration reached a target
this script had not routed — the reader stubbed through the IMSD_PCSCF seam to
find none, the address in /etc/imsd.env as PCSCF_DISCOVERED= — over imsd's own
route, and a resume stood with the peer's route removed first.
2026-10-11 16:16:46 +02:00
# ---- hand the connected ims netdev, the PDN's own address and the
# discovered P-CSCF list to imsd (imsd.service reads /run/imsd.env): the
# daemon carries no baked-in interface name and registers on the address
# the network assigned — not on the kernel's temporary SLAAC address from
# the RA, which the netdev lists first and which is deprecated after a day
# and removed after two (ims ledger F18; the SAs and the Contact would go
# with it). PCSCF_DISCOVERED= is written on every run, empty when nothing
packaging, ims-pdn-up, imsd, README: imsd runs without /etc/imsd.env and idles without a SIM
With the P-CSCF coming from the PDN's PCO, the configuration file has no
required line left, so the apk's ConditionPathExists=/etc/imsd.env drop-in
goes: the unit starts on every phone with the package, and VoLTE is on by
default. Installing the package and inserting a SIM is the opt-in; the file
holds overrides only, and its existence changes nothing. By the versioning
rule (AGENTS §11, the first commit of this series) that is what moves the
major number: the release is 1.0.0, and §11 now records it. What used to be the
file's job — keeping a phone nobody configured out of a failing unit — is now
a rule on facts, the same for every phone: a modem in the failed state with
a SIM reason makes the script write IMSD_IDLE=no-sim to /run/imsd.env and
exit 0; a profile-indexed connect series the network refuses ten times in a
row — each a persistent 3GPP cause in ModemManager's typed error (not
subscribed, unknown APN, barred, not authorised in the PLMN, authentication
failed, the service option not supported, the wrong PDN type) with packet
service at the time; counted only while the modem reads registered,
connecting or connected with packet service not detached — stricter than
the daemon's PacketServiceUp; any other read, a timeout or a transient cause
starts the count again — writes IMSD_IDLE=no-ims-pdn; and a bring-up that
ends with no P-CSCF — none configured, none discovered: no IPv6 P-CSCF read
from the PCO (PCSCF_DISCOVERY=none in the hand-off, the reader's exit 2: the
list absent, empty, unreadable or IPv4/FQDN only), discovery skipped (no
rmnet mux id) or no hand-off at all (a manual run) — and no registration of
imsd's own to resume is the daemon's own idle, decided before it takes the bus name (SaPeer and the
holds' old_gone say what can be resumed; a pair the holds wrote off is taken
out of the kernel there) and at the no-target bring-up, where a pair nothing
can resume is written off and taken out of the kernel first. imsd reads the key, logs idle: … and exits 0, which Restart=on-failure
leaves alone, so the unit stays inactive until the next boot, a systemctl
start or a start of imsd-ofonod where that runs (its Wants starts an idle
imsd again). What is not a fact keeps the retry: no coverage, a SIM still
waiting for its PIN, a modem without an ims profile (the APN-string connect
is tried), an APN-string series (the modem's own wall), a resumable pair
whose attempt failed (the fatal), and a P-CSCF reader that did not answer
(PCSCF_DISCOVERY=failed: the daemon exits 1 and the restart runs discovery
again). A phone that never had the file keeps its stopped unit through the
upgrade and starts at its next boot, registering where the rule above lets
it — at the upgrade itself where imsd-ofonod runs, whose restart pulls imsd
in through its Wants (systemd queues the marked unit's try-restart, collapses
it to a restart and starts what the unit Wants in the same transaction:
job_type_collapse, transaction_add_job_and_dependencies, 262~rc3); the
opt-out is systemctl mask --now imsd, since a plain disable is undone by
that Wants. imsd-ofonod.service loses its After=imsd.service: with imsd
starting on every phone, GNOME Calls' backend would otherwise wait behind
the ims PDN bring-up at every boot and at its marked restart at an upgrade,
for the whole bring-up (up to its TimeoutStartSec=900); it starts at once and republishes when imsd's name
appears (OnImsdAppeared).
The configuration file's path is the IMSD_ENV seam the mock bench uses.
AGENTS §6's lint figures are this tree's (109 findings in 16 of 20 files,
2026-10-13; the 72 of 2026-10-06 predate the Qmi files).
Tested: crafter-build test, 8 passed; on the host build the daemon's idle
exit for the script's keys and, with no pair, for every discovery verdict
but failed (none, skipped, no hand-off) — exit 0 — and the exit 1 for
failed; the script's arms on the mock bench; on the dev phone (KPN), in
builds of this series: the daemon's idle at the no-target bring-up with the
dead pair removed and before the bus name for a pair the holds had written
off, the discovery-failed exit 1 and its restart, the start without the file, the pull-in by an imsd-ofonod restart
and the mask holding against it and a reboot, and imsd-ofonod starting at
once with imsd stopped and republishing when imsd's name appeared.
2026-10-11 16:20:07 +02:00
# was discovered or PCSCF= is set; PCSCF_DISCOVERY= says how it came to be
# (found, none, failed, skipped — above).
# The dead end after the PDN is up — nothing to register to (no PCSCF=,
# nothing discovered, no registration of imsd's own in the kernel to resume)
# — is the daemon's to decide (its `idle:` exit before it takes the bus name,
# §7.3): it knows the fresh target, the kernel's SA pair and the holds, and
# PCSCF_DISCOVERY= tells it whether an empty list is the PCO's answer (none:
# the idle) or the reader's failure (failed: exit 1, the restart runs
# discovery again). The hand-off is written on every run.
printf 'DEV=%s\nLOCAL=%s\nPCSCF_DISCOVERED=%s\nPCSCF_DISCOVERY=%s\n' " $IFACE " " $ADDR " " $DISCOVERED " " $DISCOVERY " > /run/imsd.env
log " wrote /run/imsd.env: DEV= $IFACE LOCAL= $ADDR PCSCF_DISCOVERED= $DISCOVERED PCSCF_DISCOVERY= $DISCOVERY "
2026-07-22 22:53:28 +02:00
# ---- open the IMS protected ports in the firewall (rung 5b root cause,
# journal/ims.md s56): pmOS's default nftables INPUT chain is policy-drop and
# explicitly drops all inbound on qmapmux*, which silently killed every
# network-initiated request (reg-event NOTIFY, MT INVITE) after ESP decap —
# the P-CSCF's TCP SYNs to the protected server port never reached the
# listener, so terminating delivery failed and MT calls fell back to CS.
# 45061/45062 = imsd's protected client/server ports (kPortUc/kPortUs).
# Best-effort: never fail the unit over a missing/foreign firewall.
if nft list table inet filter >/dev/null 2>& 1; then
if ! nft list chain inet filter input | grep -q imsd-protected-ports; then
nft insert rule inet filter input iifname "qmapmux*" tcp dport 45061-45062 accept comment '"imsd-protected-ports"' &&
nft insert rule inet filter input iifname "qmapmux*" udp dport 45061-45062 accept comment '"imsd-protected-ports"' &&
log "nftables: opened protected ports 45061-45062 on qmapmux*" ||
log "nftables: rule insert failed (continuing)"
fi
fi
exit 0
