Upon updating to latest low-can, it was discovered that steering wheel LIN messages did not generate the expected signals anymore.  An investigation using git bisect revealed that commit 59bffa4 in low-can master branch ("Update function rx_filter_can for multi frame prevision.") was where the behavior broke.  Code inspection and some instrumentation revealed that the RX filter bitmasks are incorrectly generated, with the culprit being the change to using the encoder_t::encode_data function.  The steering wheel signal definitions made this visible since they are all 1-bit fields in the message.  There's a simple fix to encode_data that results in correct bitmasks that I have in hand and will track with this JIRA.