Skip to content

omamobile-modem-sit 0.3.42: set the internet bearer up again when the modem drops it - #817

Merged
derekross merged 5 commits into
mainfrom
sitd-data-call-log
Nov 15, 2026
Merged

derekross merged 5 commits into
mainfrom
sitd-data-call-log

Conversation

@derekross

@derekross derekross commented Nov 15, 2026 •

Copy link
Copy Markdown
Owner

The cause of mobile data dying across sleeps with the modem up on the 7 Pro, found by two M6-D runs on 2026-11-15 (SUSPEND-PLAN): the phone's side of the link kept working (uplink consumed, control channel answering), and with this PR's logging the modem was seen to list the internet bearer (CID 3) gone at the very wake where data stopped (12:19:40, wake 10; IMS on CID 2 kept). sitd kept that indication for start-up only and omamobile-sit-data reads the bearer's setup once, so nothing set it up again.

This PR:

  • logs each IND_DATA_CALL_LIST_CHANGED: each bearer's summary (never an address) and whether the internet bearer has its setup addresses and DNS (only the named fields, bytes 5-0x42 and the P-CSCF list), is dormant, inactive or gone; never lets the line stop sitd;
  • reconnects: a drop of the internet bearer while running schedules a setup again; once packet-registered and only if the modem still lists it gone or inactive (checked by CID), it clears the stale record, sets the bearer up with start-up's APN and IP type, and restarts omamobile-sit-data (try-restart) to take the new address. A failed try retries after 15 s, doubling to 5 minutes. No scheduling during start-up or inside its own setup.

Tests: 17 in test/test_data_call_change.py (layout sizes, dormant, gone, inactive, no address in the line, scheduling, start-up's settings, stale record cleared, listed-again, not registered, backoff); test/all passes with the simulator. The phone run of the reconnect is the next M6-D.

🤖 Generated with Claude Code

Phone run (the reference 7 Pro, 2026-11-15 12:58-13:44, M6-D with this head 0eaf3b1 on kernel-2026.11.15, device-cheetah 0.1.59, the modem's inhibitor killed for the boot, on the charger): 85 wakes in 45 minutes, no storage errors; the modem dropped the internet bearer (CID 3) nine times, about every five minutes, and sitd set it up again at the first try each time ("dropped… setting it up again in 3 s" then "set up again"); 78 of 85 data checks 3/3, each failure the moment of a drop and the next wake 3/3 again; awake at the end 3/3. Before this, data died at the first drop and stayed down (two runs, SUSPEND-PLAN M6-D).

…he internet bearer

M6-D (2026-11-15) showed mobile data stop after a sleep with the modem up
while the phone's side of the link still worked: uplink packets taken by
the modem, the control channel answering, no downlink at all. sitd keeps
the modem's IND_DATA_CALL_LIST_CHANGED for start-up and nothing more, and
omamobile-sit-data reads the bearer's setup once, so a bearer the modem
drops or replaces under it would look exactly like that. sitd now logs one
line per change: each listed bearer's summary (never an address) and
whether the internet bearer is still as set up, listed with other
addresses or DNS, inactive, or gone. The next run says which; the
reconnect comes after it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross and others added 2 commits November 15, 2026 12:13
…-CSCF fields; say dormant and differing sizes; never let the line stop sitd

The first build on the phone called a bearer with its setup addresses
"listed with other addresses or DNS" at boot: the setup response and a
list item differ past the fields both name. Review of #817 found the same.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… modem drops it

The second M6-D run (2026-11-15) showed the cause: at the wake where data
stopped the modem listed the internet bearer (CID 3) gone, IMS kept, and
nothing set it up again. sitd now schedules a reconnect when a data-call
change drops the internet bearer while it is running; once packet-
registered, and only if the modem still lists it gone or inactive, it clears
the stale record, sets the bearer up with start-up's APN and IP type, and
restarts omamobile-sit-data to take the new configuration. A failed try
comes back after 15 s, doubling to 5 minutes. Stock's data framework does
the same retry; this was never sleep-specific.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@derekross derekross changed the title omamobile-modem-sit 0.3.42: sitd logs each data-call change against the internet bearer omamobile-modem-sit 0.3.42: set the internet bearer up again when the modem drops it Nov 15, 2026
derekross and others added 2 commits November 15, 2026 12:45
…e route service stopped first and started again only if it ran; a timed-out setup released; no tries in a call or with the radio off; twelve tries)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…, and read the data switch again before each try

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@W4HO-ham

Copy link
Copy Markdown
Contributor

This fits what the 9a showed this morning: after a night of modem-up sleeps on a test kernel with the zumapro guard, the CP was ONLINE, IMS registered and voice and SMS answered, but rmnet0 passed nothing (0 of 5) with the bearer reported up (reported on linux-omarchy-mobile#258). That boot's sitd was 0.3.37, so nothing logged the data call list. Tonight I'll run the 9a in M6-D's shape with this head's package: Wi-Fi off, on the charger, RTC wakes, a ping and a DNS query over rmnet0 5 s after every wake and CPIF's queues before and after (modem-sleep-series.sh in #758), ten sleeps first, under #809's terms, and post the log lines here.

@derekross
derekross merged commit e1571fc into main Nov 15, 2026
2 checks passed
@derekross
derekross deleted the sitd-data-call-log branch November 15, 2026 18:46
derekross added a commit that referenced this pull request Nov 15, 2026
… with the modem up on by default on the 7 Pro again, and Suspend works

What killed mobile data across modem-up sleeps was the network dropping the internet bearer every few minutes with nothing setting it up again; omamobile-modem-sit 0.3.42 (#817) reconnects it, and M6-D with it passed on the reference phone (85 wakes in 45 minutes, nine drops each set up again at the first try, data back by the next wake). device-cheetah 0.1.60 ships the suspend-verified declaration again and conflicts with omamobile-modem-sit before 0.3.42. AGENTS' modem-up entry: on by default on the 7 Pro with those versions, sleeps only through logind there, the attended-test rule wherever the policy keeps the inhibitor, and #809's exception for W4HO-ham's own 9a. DEVICES: the 7 Pro's Suspend cell is works.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants