Support swapping RX/TX pins for ESC telemetry

[niklaut]

Solved Problem

For the FMUv6s target, we only had one UART TX pin left, so we need to swap the pins internally to receive ESC telemetry data.

Solution

Adds a -x command line flag to the dshot telemetry command. It is intended to be set in the board config, ie dshot telemetry -d /dev/ttyS5 -x in our case. On targets that don't support pin swapping, this selects the half-duplex single-wire mode on the TX pin, so that it can also receive.

Unfortunately I couldn't quite get this to be backward compatible since the previous implementation did not use a command line argument for the device name. But I updated all calls for it in the repo.

Changelog Entry

For release notes:

Feature Allow swapping of RX/TX pin of DShot Telemetry.

Alternatives

We could also add a parameter for the inversion? However, I kinda doubt that a generic serial port will have their pins inverted. This setting is only useful for boards with hardwired ESC telemetry pins.

Test coverage

• Tested in hardware on Skynode S.
• Tested in hardware on Skynode X.
• Upgraded all dshot telemetry /dev/ttyX to dshot telemetry -d /dev/ttyX

[github-actions]

🔎 FLASH Analysis

px4_fmu-v5x [Total VM Diff: 176 byte (0.01 %)]

    FILE SIZE        VM SIZE    
--------------  -------------- 
+0.0%    +176  +0.0%    +176    .text
  +0.0%     +68  +0.0%     +68    [section .text]
   +61%     +56   +61%     +56    DShotTelemetry::init()
  +5.3%     +20  +5.3%     +20    DShot::custom_command()
  +3.1%     +16  +3.1%     +16    DShot::print_usage()
  [NEW]     +12  [NEW]     +12    CSWTCH.171
   +11%     +12   +11%     +12    DShot::init_telemetry()
  +2.0%      +8  +2.0%      +8    DShot::Run()
 -30.8%      -4 -30.8%      -4    g_nullstring
  [DEL]     -12  [DEL]     -12    CSWTCH.170
+0.0%     +77  [ = ]       0    .debug_abbrev
+0.0%      +4  [ = ]       0    .debug_frame
+0.0%    +359  [ = ]       0    .debug_info
+0.0%    +116  [ = ]       0    .debug_line
   +17%      +1  [ = ]       0    [Unmapped]
  +0.0%    +115  [ = ]       0    [section .debug_line]
+0.0%     +75  [ = ]       0    .debug_loclists
+0.0%      +1  [ = ]       0    .debug_rnglists
  [NEW]      +2  [ = ]       0    [Unmapped]
  -0.0%      -1  [ = ]       0    [section .debug_rnglists]
+0.0%     +56  [ = ]       0    .debug_str
+0.4%      +1  [ = ]       0    .shstrtab
+0.0%     +35  [ = ]       0    .strtab
  [DEL]     -11  [ = ]       0    CSWTCH.170
  [NEW]     +11  [ = ]       0    CSWTCH.171
  [NEW]     +33  [ = ]       0    DShot::_telemetry_swap_rxtx
  +3.3%      +1  [ = ]       0    DShot::init_telemetry()
  +3.4%      +1  [ = ]       0    DShotTelemetry::init()
+0.0%     +32  [ = ]       0    .symtab
  [DEL]     -32  [ = ]       0    CSWTCH.170
  [NEW]     +32  [ = ]       0    CSWTCH.171
  [NEW]     +32  [ = ]       0    DShot::_telemetry_swap_rxtx
   +33%     +16  [ = ]       0    __nxsem_tickwait_uninterruptible_veneer
 -33.3%     -16  [ = ]       0    __stm32_dmastop_veneer
 -25.0%     -16  [ = ]       0    __stm32_recvfifo_veneer
   +50%     +16  [ = ]       0    __uart_recvchars_veneer
-2.0%    -176  [ = ]       0    [Unmapped]
+0.0%    +756  +0.0%    +176    TOTAL

px4_fmu-v6x [Total VM Diff: 184 byte (0.01 %)]

    FILE SIZE        VM SIZE    
--------------  -------------- 
+0.0%    +184  +0.0%    +184    .text
  +0.0%     +64  +0.0%     +64    [section .text]
   +61%     +56   +61%     +56    DShotTelemetry::init()
  +5.3%     +20  +5.3%     +20    DShot::custom_command()
  +3.1%     +16  +3.1%     +16    DShot::print_usage()
  [NEW]     +12  [NEW]     +12    CSWTCH.171
   +11%     +12   +11%     +12    DShot::init_telemetry()
  +2.0%      +8  +2.0%      +8    DShot::Run()
  +0.0%      +8  +0.0%      +8    g_cromfs_image
  [DEL]     -12  [DEL]     -12    CSWTCH.170
+0.0%     +77  [ = ]       0    .debug_abbrev
+0.0%      +4  [ = ]       0    .debug_frame
+0.0%    +359  [ = ]       0    .debug_info
+0.0%    +116  [ = ]       0    .debug_line
  [NEW]      +1  [ = ]       0    [Unmapped]
  +0.0%    +115  [ = ]       0    [section .debug_line]
+0.0%     +75  [ = ]       0    .debug_loclists
-0.0%      -3  [ = ]       0    .debug_rnglists
  [DEL]      -2  [ = ]       0    [Unmapped]
  -0.0%      -1  [ = ]       0    [section .debug_rnglists]
+0.0%     +56  [ = ]       0    .debug_str
+0.4%      +1  [ = ]       0    .shstrtab
+0.0%     +35  [ = ]       0    .strtab
  [DEL]     -11  [ = ]       0    CSWTCH.170
  [NEW]     +11  [ = ]       0    CSWTCH.171
  [NEW]     +33  [ = ]       0    DShot::_telemetry_swap_rxtx
  +3.3%      +1  [ = ]       0    DShot::init_telemetry()
  +3.4%      +1  [ = ]       0    DShotTelemetry::init()
+0.0%     +32  [ = ]       0    .symtab
  [DEL]     -32  [ = ]       0    CSWTCH.170
  [NEW]     +32  [ = ]       0    CSWTCH.171
  [NEW]     +32  [ = ]       0    DShot::_telemetry_swap_rxtx
-2.6%    -184  [ = ]       0    [Unmapped]
+0.0%    +752  +0.0%    +184    TOTAL

Updated: 2025-05-27T08:32:56

[PetervdPerk-NXP]

On the IMXRT we don't support TIOCSSWAP unfortunely, however when using TIOCSSINGLEWIRE the TX pin becomes half-duplex but also has the capability to RX as well.

I do happen to have a board that needs this feature since you're doing the PR, would be able to add TIOCSSINGLEWIRE as an feature as well?

[niklaut]

would be able to add TIOCSSINGLEWIRE as an feature as well?

Sure, absolutely. Maybe -s is a bit too confusing to use for both Swap and Singlewire though. Maybe -x for swap (x = "cross over") and -s for single wire?

[PetervdPerk-NXP]

would be able to add TIOCSSINGLEWIRE as an feature as well?

Sure, absolutely. Maybe -s is a bit too confusing to use for both Swap and Singlewire though. Maybe -x for swap (x = "cross over") and -s for single wire?

Good point, i think using -s and -x makes the most sense yeah.

[niklaut]

I've tested both swapping and single wire on the STM32H7 and they both work for my use-case.
I've implemented -x as pin swap or single wire mode. The end result is the same, and we don't expect people to use the TX pin of the same UART for anything else.

[PetervdPerk-NXP]

LGTM, don't have the hardware right now, but i'll test it when I have.

[alexcekay]

Nice 👍