Bug fixes:
- Non-movement commands were not correctly synchronised with movement commands when there were more than 8 non-movement commands interspersed with more than 2 seconds of movement
- If M28/M29 was used in a macro file to then the commands between M28 and M29 were executed as well as being written to the target file
- If you used the jog buttons in DWC before homing the axes, and you didn't use M564 H0 in config.g, then as well as the "insufficient axes homed" error message, a stack underflow error message was produced
- Neither M561 nor G29 S2 adjusted the user Z coordinate if bed compensation was previously active (G29 S2 did if you ran it twice)
- Duet 06/085 only: fixed a buffer overflow in the Netbios responder code
- Duet 2 series: when an under-voltage event occurred, spurious driver status warnings/errors were sometimes reported
- Duet 2 series: when an under- or over-voltage event occurred, the VIN voltage reported was the current voltage, not the voltage when the event was recorded
- When an axis was made visible and later hidden, subsequent move commands sometimes sent step commands incorrectly to the driver(s) associated with that axis. This could cause unwanted movement if the axis was still mapped to a real driver. If it was mapped to a dummy driver, it could still cause step errors to be recorded and/or some movements to be slowed down.
- The longest loop time reported by M122 was distorted by the fact that M122 itself takes a long time to execute, due to the volume of output it produces and the need to synchronise with the Network task

New features/changed behaviour:
- On the Duet 2 Maestro, the 2 optional add-on drivers are now assumed to be TMC2224 with UART interface
- When the Z probe type is set to 9 for BLTouch, the probe output is no longer filtered, for faster response
- If an error occurs while reading or writing the SD card, the operation is retried. The M122 report includes the maximum number of retries that were done before a successful outcome.
- On the Duet WiFi/Ethernet, at startup the firmware does additional retries when checking for the presence of a DueX2 or DueX5 and/or additional I/O expansion board
- When a resurrect.g file is generated, it now includes G92 commands just before it invokes resurrect-prologue.g, to set the assumed head position to the point at which power was lost or the print was paused. This is to better handle printers for which homing Z is not possible when a print is already on the bed. Caution: this doesn't allow for any Z lift or other movement in the power fail script.
- RTOS builds only: added a separate software watchdog to monitor the Heat task
- RTOS builds only: in the M122 report, the software reset data now includes which task was active, and only owned mutexes are listed
- Upgraded compiler to 2018-q2-update