Repository navigation
controls: per-door unlatch (VCSEC_requests 0x1F9) + forward door status - #89
Merged
Merged
Conversation
…tatus VC_CMD_DOOR_UNLATCH (0x50, value 0-3 = FL/FR/RL/RR), the CAN equivalent of the Tesla app's Unlatch Door. Captured from 12 app-triggered unlatches on a 2023 Model Y: each is VCSEC_requests with that door's ClosureRequest = OPEN and RequestReason = RKE, held ~800 ms, and the door's latch flips CLOSED->OPENED ~100 ms after the first frame. Injection is phase-locked to the car: can_manager calls a small RX hook, and while an unlatch is in progress each of VCSEC's own 0x1F9 frames releases a semaphore so ours is sent right after it (its all-zero NONE is overwritten within ~1 ms), with a 20 ms fill in between. Refused only while in R/N/D: a parked car whose drive system has gone to standby no longer reports P, and that is the main case this is for. Multi-finger actions 10-13 map to the four doors. VCLEFT_doorStatus (0x102) and VCRIGHT_doorStatus (0x103) are now forwarded to the app so it can show per-door and trunk latch state. VCSEC_requests in bus_1_tesla_vehicle.dbc was a 1-byte stub; replaced with the 8-byte definition (+ VAL tables) from dzid26/ESP32-DualCAN (CERN-OHL-W) and marked decode so its signals are generated. Bus-1 HW filters 21 -> 25. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuM9buJMg6fPxajubfe
ehoeve786
changed the title
- Refuse unlatch above 0.5 kph as well as in R/N/D (DI_speed now decoded). - Build the 0x1F9 frame from scratch instead of the live bytes so other VCSEC requests are never replayed. - Credit dzid26/ESP32-DualCAN in the DBC; sender VEH like the rest. - Comments: 800 ms hold, drop fork PR reference. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuNGK1Ev4AkBBNFW1CT
ehoeve786
added a commit
to ehoeve786/dashpilot
that referenced
this pull request
Nov 3, 2026
The firmware (softwiredtech/dashkit-firmware#89) refuses the unlatch in R/N/D or above 0.5 kph, not "outside Park": a parked car in drive-system standby no longer reports P. The first commit of this PR pairs with that upstream PR, not with fork PR #9 as its message says. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuNGK1Ev4AkBBNFW1CT
Restore the can_frame_live path that was tested on the car; the zero-frame variant from the previous commit is untested. Speed gate stays. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuNGK1Ev4AkBBNFW1CT
Speed-gated live-frame build tested on the car (2023 Model Y, all four doors). Build from scratch now so other VCSEC requests present on the live frame are never replayed for the 800 ms hold. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuNGK1Ev4AkBBNFW1CT
This reverts commit 14cfc22.
Reinstates 14cfc22 after on-car testing: all four doors unlatch with the frame built from scratch, so other VCSEC requests on the live frame are never replayed during the 800 ms hold. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuNGK1Ev4AkBBNFW1CT
Contributor
|
@ehoeve786 nice work, one safety thing: The |
Review feedback: an unread DI_speed defaulted to 0 and an unread DI_gear (-1) passed the R/N/D check, so right after boot, before any DI frame, nothing blocked the unlatch. Both reads must now succeed, as in sport_kickdown.c. A parked car whose drive system has gone to standby still works: the value cache keeps the last P frame. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuL5dzb3mcxzhffpXu8
… asleep With no DI_systemStatus/DI_speed frame since boot (DashKit restarted while the car sits in Accessory and the drive inverter is in standby), allow the unlatch only if both EPB calipers report PARK; otherwise refuse. With DI data the R/N/D + speed gate is unchanged. Adds EPBL/EPBR_internalStatus (0x288/0x228) to the bus-1 decode set (filters 25 -> 27 of 32). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CiTeuL5dzb3mcxzhffpXu8
Contributor
Author
|
Fixed in 29e2797 and 43ba49d: door_unlatch() now fails closed. With DI data it keeps the R/N/D + speed gate, and both reads must succeed. With no DI frame since boot (DashKit restarted while the drive inverter is in standby), it unlatches only if both EPB calipers report PARK (EPBL/EPBR_internalStatus, added to the bus-1 decode set, 27/32 filters). |
ehoeve786
added a commit
to ehoeve786/dashpilot
that referenced
this pull request
Nov 8, 2026
Four momentary controls on Android and iOS: Unlatch Driver / Passenger / Rear Left / Rear Right Door. Each sends CMD_DOOR_UNLATCH (0x50) with the door index 0-3; gesture values 10-13 let them be bound to multi-finger taps like the other controls. Strings in en/de/hu on both platforms. The firmware (softwiredtech/dashkit-firmware#89, merged) replays the VCSEC_requests 0x1F9 closure request the Tesla app uses and refuses it in R/N/D or while moving; older firmware rejects the opcode as unknown. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
wpmed92
pushed a commit
to softwiredtech/dashpilot
that referenced
this pull request
Nov 8, 2026
… dashboards (#273) * controls: per-door unlatch (DashKit opcode 0x50, gestures 10-13) Four momentary controls on Android and iOS: Unlatch Driver / Passenger / Rear Left / Rear Right Door. Each sends CMD_DOOR_UNLATCH (0x50) with the door index 0-3; gesture values 10-13 let them be bound to multi-finger taps like the other controls. Strings in en/de/hu on both platforms. The firmware (softwiredtech/dashkit-firmware#89, merged) replays the VCSEC_requests 0x1F9 closure request the Tesla app uses and refuses it in R/N/D or while moving; older firmware rejects the opcode as unknown. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * door status: per-door, trunk and liftgate state in CarState for dashboards The firmware forwards VCLEFT_doorStatus 0x102, VCRIGHT_doorStatus 0x103 and VCLEFT_liftgateStatus 0x142. The shared bridge decodes them into six CarState fields, exposed to web and Rive dash apps on both platforms: - doorOpenFL/FR/RL/RR, trunkOpen: 1 while the latch reports OPENED, OPENING, AJAR or CLOSING; 0 for CLOSED, SNA and the fault states. - trunkPosition: powered liftgate angle in degrees from mux 1, kept across mux-0 frames, clamped at 0 (closed reads slightly negative). bridge: DOOR_OFFSET = 47, FIELD_COUNT = 53; BridgeCarState carries the fields for iOS. Only TeslaDashKitMapper is wired (new_gen_juniper.dbc has no definitions for these messages yet). Android: CarState, CanFrameDecoder, Rive dashValues, WebView getters. iOS: CarState (bridge + Codable), Rive dashValues, WebView JSON. dash-apps: README table, web-compose signals (five warning booleans and the trunk angle), host getter map, en/hu/de labels. Host test tesla_closure_mapper_test covers latch mapping, liftgate mux handling and the array layout. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * door status: frunk latch as frunkOpen VCFRONT_status mux 0 (already forwarded for anyDoorOpen) carries the frunk latch status. Decode it with the same rule as the doors into frunkOpen (0/1; the frunk has no position signal): CarState slot DOOR_OFFSET + 6, FIELD_COUNT = 54, carried through the iOS bridge, both apps, Rive dashValues, the WebView getter / JSON, the web-compose signal catalog and the README. Host test covers mux 0 only and the array slot. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
VC_CMD_DOOR_UNLATCH(0x50; value 0–3 = front left / front right / rear left / rear right), the CAN equivalent of the Tesla app's Unlatch Door, multi-finger actions 10–13 for the four doors, and forwards per-door/trunk latch state (VCLEFT_doorStatus0x102,VCRIGHT_doorStatus0x103) and powered-trunk state/position (VCLEFT_liftgateStatus0x142) to the app. Single commit, no new files.Derived from 12 app-triggered unlatches captured on the car (3 per door): each is
VCSEC_requests0x1F9 with that door'sClosureRequest = 3 (OPEN)andRequestReason = 1 (RKE)held ~800 ms; the latch flips CLOSED→OPENED ~100 ms after the first frame.60 00 10 00 00 00 00 0000 03 80 00 00 00 00 0000 18 00 04 00 00 00 0000 C0 00 20 00 00 00 00Injection: RMW on the live 0x1F9 frame, phase-locked to the car.
can_managercallsvehicle_control_on_frame(); while an unlatch is active each of VCSEC's own 0x1F9 frames gives a semaphore and ours is sent immediately after (its NONE is overwritten within ~1 ms), with a 20 ms fill between. Tested on a 2023 Model Y as the most consistent variant; a plain 20 ms burst also worked but less reliably.Gate: refused only in R/N/D. A parked car whose drive system has gone to standby (PRND gone from the screen) stops reporting P; a "must be P" gate blocked exactly that case during testing.
DBC:
VCSEC_requestswas a 1-byte stub; replaced by the 8-byte definition + VAL tables from dzid26/ESP32-DualCAN (CERN-OHL-W) and markeddecode. Bus-1 HW filters 21 → 25.dbc_generated.{c,h}regenerated withtools/gen_dbc.py(output matches) and the host DBC test (tools/test/test_dbc.c) passes.Pairs with softwiredtech/dashpilot#272, which adds the four controls on Android and iOS and decodes the forwarded door/trunk/liftgate state into CarState for dashboards.
🤖 Generated with Claude Code
https://claude.ai/code/session_01CiTeuLpxE64AAygE8U6A5F