arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. by dakejahl · Pull Request #19957 · apache/nuttx · GitHub

Summary

Why: imxrt_transmit() picks the lowest free TX mailbox. FlexCAN breaks an arbitration tie between mailboxes holding equal CAN IDs by taking the lowest mailbox number, so refilling a just-drained low mailbox while older frames are still pending in higher ones puts the newer frame on the wire first. imxrt_txdone_work() re-polls after each individual TX completion, which is exactly the condition that triggers it.
What: arch/arm/src/imxrt/imxrt_flexcan.c, TX mailbox selection only.
How: a new imxrt_txmb_next() returns a mailbox above every pending one and wraps back to the bottom only once the ring has drained, so a newer frame can never win arbitration against an older one. imxrt_txringfull() is expressed in terms of it. Ordering then holds for any transfer length; the cost is that the ring stalls at a wrap rather than refilling immediately.
Every frame of a multi-frame transport transfer carries the same CAN ID, so any such protocol is affected. On DroneCAN the receiver sees a broken toggle bit and discards the transfer.
A second commit fixes pre-existing nxstyle violations in the same file, per CONTRIBUTING.md section 2.1. It is whitespace only — git diff -w against its parent is empty, and the resulting binary is byte-identical.

Impact

New feature? NO. Bug fix.
Impact on user? NO — no API, Kconfig or behaviour change to adapt to. Multi-frame CAN traffic that was silently corrupted now arrives intact.
Impact on build? NO.
Impact on hardware? YES — arch/arm/src/imxrt only, any i.MX RT board using FlexCAN via SocketCAN. s32k1xx_flexcan.c and s32k3xx_flexcan.c select a mailbox the same way and appear to share the defect, but I have no S32K hardware and have deliberately left them alone.
Impact on documentation? NO.
Impact on security? NO.
Impact on compatibility? NO.

Testing

I confirm that changes are verified on local setup and works as intended:

Build Host: Linux x86_64, arm-none-eabi-gcc 13.2.1 20231009
Target: ARM, i.MX RT1176 (ARK FMU-v6XRT), FlexCAN1 at 1 Mbit, SocketCAN + DroneCAN

How to reproduce: put an i.MX RT board on a CAN bus with a second node, have the board publish a message large enough to need several frames, and decode CAN-ID plus tail byte per frame with an analyser. Frames of one transfer arrive out of order, so the toggle bit breaks and the receiver drops the transfer. The second node acts as the control — its own multi-frame transfers stay intact, which places the fault on the i.MX RT transmit path rather than the bus.

Measured with an ARK X20 GNSS node (node 124) and a CUAV Babel as the analyser. incomplete counts transfers whose frames never all arrived; togerr counts toggle-bit violations, i.e. reordering.

Testing logs before change:

# BEFORE  17455 frames / 60.0s (291 fps)
 src stream                            frames  xfers  multi incomplete  togerr   loss%
   1 20030 remoteid.BasicID               240     60     60          0       4    6.67
   1 20031 remoteid.Location              360     60     60          2      47   81.67
   1 1081 indication.LightsCommand        600    600      0          0       0    0.00
   1 341 protocol.NodeStatus               60     60      0          0       0    0.00
 124 1063 gnss.Fix2                      6000    600    600          0       0    0.00
 124 1061 gnss.Auxiliary                 1800    600    600          0       0    0.00
bad transfers by source node: {1: 53, 124: 0}

Only the i.MX RT board's multi-frame streams are affected; its single-frame streams and every stream from node 124 — including 600 ten-frame gnss.Fix2 transfers — are clean.

Testing logs after change:

# AFTER  35513 frames / 120.0s (296 fps)
 src stream                            frames  xfers  multi incomplete  togerr   loss%
   1 20030 remoteid.BasicID               480    120    120          0       0    0.00
   1 20031 remoteid.Location              720    120    120          0       0    0.00
   1 20033 remoteid.System                600    120    120          0       0    0.00
   1 1081 indication.LightsCommand       1198   1198      0          0       0    0.00
   1 341 protocol.NodeStatus              120    120      0          0       0    0.00
 124 1063 gnss.Fix2                     12010   1201   1201          0       0    0.00
 124 1061 gnss.Auxiliary                 3603   1201   1201          0       0    0.00
bad transfers by source node: {1: 0, 124: 0}

Driver counters after the change, sampled twice 45 s apart to show nothing is accumulating:

CAN1 status:
	HW errors: 0
	IO errors: 0
	RX frames: 47623   ->   62863
	TX frames: ...
UAVCAN node status:
	Transfer errors:   4   ->   4      (static; incurred during the reflash reboot)
	RX transfers:   8580   ->  11324

This patch alone fixes the 4-frame stream and takes the 6-frame one from 81.7% to 40%. The remainder was two further bugs above this driver, fixed separately in PX4 (a frame discarded rather than retried when the ring is full, and a TX queue bypass). The "after" figures above are with all three applied — this patch is necessary but not on its own sufficient for that particular stack.

Build log:

$ make ark_fmu-v6xrt_default
Memory region         Used Size  Region Size  %age Used
           flash:     2336728 B      3968 KB     57.51%
            sram:      107668 B      1792 KB      5.87%
            itcm:      217120 B       256 KB     82.82%

./tools/checkpatch.sh -c -u -m -g master..HEAD → ✔️ All checks pass.

FYI @PetervdPerk-NXP

…order. Backport of apache/nuttx#19957. imxrt_transmit() handed out the lowest free TX mailbox. FlexCAN breaks an arbitration tie between mailboxes holding equal CAN IDs by taking the lowest mailbox number, so refilling a just-drained low mailbox while older frames are still pending in higher ones puts the newer frame on the wire first. Every frame of a multi-frame transport transfer carries the same CAN ID, so this reorders transfers. On a DroneCAN bus the receiver sees a broken toggle bit and discards the transfer: measured on the wire on an ARK FMU-v6XRT, a 6-frame message failed 82% of the time, while a GNSS node on the same bus was flawless over the same capture. Hand out a mailbox above every pending one instead, and wrap back to the bottom only once the ring has drained. Assisted-by: Claude:claude-opus-5 Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com>

MemBrowse Memory Report

No memory changes detected for:

Correct the switch case label indentation in imxrt_netinitialize(), the brace alignment in imxrt_ioctl(), and the indentation of the ERR005829 workaround statements so imxrt_flexcan.c passes nxstyle. The last of these is only reported since commit c81cc02. These predate this series; CONTRIBUTING.md section 2.1 asks that modified files be brought into compliance even where the contributor did not introduce the problem. Whitespace only, no functional change: `git diff -w` against the parent is empty. Assisted-by: Claude:claude-opus-5 Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com>

imxrt_transmit() handed out the lowest free TX mailbox. FlexCAN breaks an arbitration tie between mailboxes holding equal CAN IDs by taking the lowest mailbox number, so refilling a just-drained low mailbox while older frames are still pending in higher ones puts the newer frame on the wire first. imxrt_txdone_work() re-polls after each individual TX completion, which is exactly the condition that triggers it. Every frame of a multi-frame transport transfer carries the same CAN ID, so this reorders transfers. On a DroneCAN bus the receiver sees a broken toggle bit and discards the transfer: measured on the wire, a 6-frame message from an i.MX RT1176 failed 82% of the time, while a node on the same bus running a different controller was flawless over the same capture. Hand out a mailbox above every pending one instead, and wrap back to the bottom only once the ring has drained. Ordering then holds for any transfer length; the cost is that the ring stalls at a wrap rather than refilling immediately. Assisted-by: Claude:claude-opus-5 Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com>

…order. Backport of apache/nuttx#19957. imxrt_transmit() handed out the lowest free TX mailbox. FlexCAN breaks an arbitration tie between mailboxes holding equal CAN IDs by taking the lowest mailbox number, so refilling a just-drained low mailbox while older frames are still pending in higher ones puts the newer frame on the wire first. Every frame of a multi-frame transport transfer carries the same CAN ID, so this reorders transfers. On a DroneCAN bus the receiver sees a broken toggle bit and discards the transfer: measured on the wire on an ARK FMU-v6XRT, a 6-frame message failed 82% of the time, while a GNSS node on the same bus was flawless over the same capture. Hand out a mailbox above every pending one instead, and wrap back to the bottom only once the ring has drained. Assisted-by: Claude:claude-opus-5 Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com>

Thanks a lot for fixing this long outstanding bug we had.

…order. (#394) * [BACKPORT] arch/arm/src/imxrt: Fix FlexCAN coding style violations. Correct the switch case label indentation in imxrt_netinitialize(), the brace alignment in imxrt_ioctl(), and the indentation of the ERR005829 workaround statements so imxrt_flexcan.c passes nxstyle. Whitespace only, no functional change. Assisted-by: Claude:claude-opus-5 Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com> * [BACKPORT] arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. Backport of apache/nuttx#19957. imxrt_transmit() handed out the lowest free TX mailbox. FlexCAN breaks an arbitration tie between mailboxes holding equal CAN IDs by taking the lowest mailbox number, so refilling a just-drained low mailbox while older frames are still pending in higher ones puts the newer frame on the wire first. Every frame of a multi-frame transport transfer carries the same CAN ID, so this reorders transfers. On a DroneCAN bus the receiver sees a broken toggle bit and discards the transfer: measured on the wire on an ARK FMU-v6XRT, a 6-frame message failed 82% of the time, while a GNSS node on the same bus was flawless over the same capture. Hand out a mailbox above every pending one instead, and wrap back to the bottom only once the ring has drained. Assisted-by: Claude:claude-opus-5 Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com> --------- Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com>

Brings in PX4/NuttX#394 (backport of apache/nuttx#19957). imxrt FlexCAN handed out the lowest free TX mailbox, so a refilled low mailbox could win arbitration over an older frame of the same CAN ID still queued in a higher one. Multi-frame DroneCAN transfers arrived out of order and were dropped by the receiver. Assisted-by: Claude:claude-opus-5[1m] Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com>

Brings in PX4/NuttX#394 (backport of apache/nuttx#19957). imxrt FlexCAN handed out the lowest free TX mailbox, so a refilled low mailbox could win arbitration over an older frame of the same CAN ID still queued in a higher one. Multi-frame DroneCAN transfers arrived out of order and were dropped by the receiver. Assisted-by: Claude:claude-opus-5[1m] Signed-off-by: Jacob Dahl <dahl.jakejacob@gmail.com> (cherry picked from commit 677d1fa) Assisted-by: GitHub Copilot:gpt-6-astra

dakejahl requested review from michallenc and xiaoxiang781216 as code owners August 24, 2026 17:02

dakejahl mentioned this pull request Aug 24, 2026

[BACKPORT] arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. PX4/NuttX#394

Merged

dakejahl marked this pull request as draft August 24, 2026 17:04

This was referenced Aug 24, 2026

fix(uavcan): stop losing and reordering SocketCAN TX frames PX4/PX4-Autopilot#28363

Merged

[Bug] CAN-based RTK GNSS not working on NXP v6X-RT – inconsistent CAN frames, no RTK Fix PX4/PX4-Autopilot#26868

Closed

github-actions Bot added Arch: arm Issues related to ARM (32-bit) architecture Size: S The size of the change in this PR is small labels Aug 24, 2026

dakejahl force-pushed the fix/imxrt-flexcan-tx-mailbox-ordering branch from c4e8463 to 4f523f1 Compare August 24, 2026 17:20

dakejahl changed the title arch/arm/src/imxrt: flexcan: keep TX mailbox allocation in order arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. Aug 24, 2026

dakejahl added 2 commits August 24, 2026 11:30

dakejahl force-pushed the fix/imxrt-flexcan-tx-mailbox-ordering branch from 4f523f1 to 11d4a6c Compare August 24, 2026 17:31

github-actions Bot added Size: M The size of the change in this PR is medium and removed Size: S The size of the change in this PR is small labels Aug 24, 2026

xiaoxiang781216 approved these changes Aug 24, 2026

View reviewed changes

dakejahl marked this pull request as ready for review August 24, 2026 17:40

PetervdPerk-NXP approved these changes Aug 24, 2026

View reviewed changes

simbit18 approved these changes Aug 25, 2026

View reviewed changes

xiaoxiang781216 merged commit e6696d5 into apache:master Aug 25, 2026
38 checks passed

dakejahl mentioned this pull request Aug 25, 2026

chore(nuttx): bump NuttX for the imxrt FlexCAN TX ordering fix PX4/PX4-Autopilot#28381

Merged

This was referenced Sep 8, 2026

[BACKPORT] fix(uavcan): stop losing and reordering SocketCAN TX frames PX4/PX4-Autopilot#28613

Closed

[BACKPORT] fix(uavcan): stop losing and reordering SocketCAN TX frames (#28363) PX4/PX4-Autopilot#28614

Merged

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. by dakejahl · Pull Request #19957 · apache/nuttx · GitHub

Repository navigation

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. - #19957

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order.#19957
xiaoxiang781216 merged 2 commits into
apache:masterfrom
dakejahl:fix/imxrt-flexcan-tx-mailbox-ordering

dakejahl commented Aug 24, 2026 •
edited
Loading

Uh oh!

Uh oh!

dakejahl commented Aug 24, 2026

Uh oh!

github-actions Bot commented Aug 24, 2026 •
edited
Loading

Uh oh!

Uh oh!

PetervdPerk-NXP left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. by dakejahl · Pull Request #19957 · apache/nuttx · GitHub

Repository navigation

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order. - #19957

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order.#19957xiaoxiang781216 merged 2 commits intoapache:masterapache/nuttx:masterfrom dakejahl:fix/imxrt-flexcan-tx-mailbox-orderingdakejahl/incubator-nuttx:fix/imxrt-flexcan-tx-mailbox-orderingCopy head branch name to clipboard

Conversation

dakejahl commented Aug 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Impact

Testing

Uh oh!

dakejahl commented Aug 24, 2026

Uh oh!

github-actions Bot commented Aug 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

MemBrowse Memory Report

Uh oh!

PetervdPerk-NXP left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

arch/arm/src/imxrt: Keep FlexCAN TX mailbox allocation in order.#19957
xiaoxiang781216 merged 2 commits into
apache:masterfrom
dakejahl:fix/imxrt-flexcan-tx-mailbox-ordering

dakejahl commented Aug 24, 2026 •
edited
Loading

github-actions Bot commented Aug 24, 2026 •
edited
Loading