The system tracing that is called via the performance monitor optionally reads, parses and hooks kernel messages.
It translates the kernel tracing level into known fixed tracing level types, e.g. KERN_ERR, KERN_WARNING.
However if the tracing level was higher than any of the known levels, it would not pass that level onto the hook.
As a consequence, in Thor for example, which captures high importance kernel messages and forwards them to the workunit, all low importance messages were captures and forwarded to the workunit.