Telltale · runs on QNX 8.0

The car stopped.The speedometerdidn't.Telltale noticed.

QNX Hackathon 2026 · Automotive & Mobility

Monitor off · redrawn

48frozen

Telltale armed

0verified
Speed over time on a real stop: the true speed falls from 49.01 to 0 km/h; with the monitor off the glass stays at 48; with Telltale armed the glass is caught 20.3 ms after the last true frame and the true value is back on the glass 53.2 ms after it.0204060km/hcaught 20.3 ms · truth back on glass 53.2 msmonitor off: the glass stays 48truth: 0 km/h
Telltalereal comma2k19 stop · timings: incident TT-FRZ-0001 · left glass redrawn · QEMU TCG, not hardware

Two numbers · QEMU (TCG), not hardware

Milliseconds, not half a minute.

53.2 ms

last true frame to truth on the glass, one incident (regression runs: 34.5–54.4)

29.9 s

of 30, wrong with the monitor off (median of 40 runs)

Telltaleincidents.csv · bench-ablation.csv · QEMU TCG

Clusters fail. Drivers can't tell.

23V-298…may fail to display any information…Kia, 2023 Sportage +
24V-802…may not illuminate at startup or while driving.Ford, 2024 Mustang
26V-372…previously repaired incorrectly…Ford and Lincoln, 2019

blank

48

frozen: looks fine

Those are blank. A frozen cluster looks fine.

Telltaleapi.nhtsa.gov/recalls · campaigns 23V298, 24V802, 26V372 (saved in evidence/nhtsa) · glasses drawn

The idea

Read the glass back. Put the truth on it.

  1. 1digits decoded from the composed display
  2. 2lamps checked against the vehicle state
  3. 3a safe layer drawn by Telltale
Telltale — read-back of the QNX cluster during the hold: true digits 42 under an amber banner, DISPLAY FAULT CAUGHT, VALUE_MISMATCH, 19 ms. 123
Telltalescreen_read_display capture · run t-debug · +466 ms after the catch · engineering view

How it works

Two processes, one screen. Only one can lie.

vehfeed sets the vehicle state in vehicled, the /dev/vehicle resource manager. The cluster renderer reads it and posts its window to QNX Screen. telltale reads the vehicle state and reads the composed display back from Screen; on a lie it posts a safe layer on top, kills the renderer, which slm or a shell loop respawns, and verifies the new renderer before hand-back. vehfeedreal comma2k19 speed vehicledresource manager /dev/vehicle clusterrenderer_tdraw prio 20 QNX Screencomposes what the driver sees telltalemonitor_t · SCHED_FIFO 60read back → decode → compare→ safe layer, kill, verify devctl SET ionotify pulse + devctl GET devctl GET_SINCE, every check screen_post_window screen_read_display safe layer, z-top SIGKILL if hung, wrong or blank → respawned by slm or a shell loop
Telltalered = the path that can lie · green = the path that checks it

One incident, from the target's own log

Caught, covered, repaired: 319.7 ms.

telltale.log · TT-FRZ-0001 · make test-target at 985ab29QEMU TCG
DETECT TT-FRZ-0001 TICK_STALL shown=48 truth=46 lamps_shown=0x01 lamps_truth=0x00 caught_ms=20.3
SAFE verified on glass TT-FRZ-0001 (33.0 ms after detect, composed check)
SIGKILL renderer pid 34119714 (hung)
RESTARTING -> NORMAL
20.3caught58.7SIGKILL
unverifiedsafe layer: true speedhanded back
0last true frame53.2truth on glass319.7verified hand-back

ms from the last true frame · bar to scale

Telltaleincidents.csv · timestamps from the monitor

Remove QNX and it breaks.

Screen read-back

Every 20 ms, against the cluster's own glyph atlas.

screen_read_displayTelltale — crop of a real read-back: the digits 42 on the glass during a hold.
  1. Resource manager/dev/vehicle
  2. Message passingdevctl · ionotify
  3. secpol22/22 checks
  4. PrioritiesSCHED_FIFO 60
Telltaleon-target suite 163 passed / 0 failed at 985ab29 · secpol 13 denials of 22 checks

The bench · QEMU TCG

260 faults. 0 missed. 0 false alarms.

Every detection, one dot

Detection time of every injected fault, by class, with each class p95: BLANK 24.8, CRASH 31.1, MISDRAW 55.0, FREEZE 129.8, STARVE 175.8, STALE 271.2 ms; 0 missed.050100150200250300ms from fault to detection · QEMU TCG100 ms stretch197.3 ms boundBLANKn = 50p95 24.8CRASHn = 50p95 31.1MISDRAWn = 10p95 55.0FREEZEn = 50p95 129.8STARVEn = 50p95 175.8STALEn = 50p95 271.2

Monitor off: the lie stays

Monitor off: seconds the glass was wrong in each of the 40 runs of a 30-second window; median 29.9 s, every run still wrong when the window closed.0102030 s40 runs, monitor disarmed, sortedseconds wrong of a 30 s window · QEMU TCG
0false alarms in 91,029 checks over 30 minutes of a real drive, monitor armed
TelltaleSTALE missed its 200 ms target: p95 271.2 ms · bench-*.csv

Honest by design

Not run on hardware yet.

Limits
  • Every timing is QEMU (TCG), not hardware
  • Nothing restarts the monitor if it dies
  • The gear indicator is drawn, not verified
Next
  • Time it on real hardware
  • Supervise the monitor
  • Verify the gear indicator
Telltalelimits from README · Honest limits

The ask

Judge it on the glass.

Read the code, the evidence and the codelab. Every bench figure regenerates from the committed CSVs.

telltale.edycu.dev

TelltaleApache-2.0 · suite and incident at 985ab29 · bench builds in each CSV's git_sha column