When I have needed flowcharts for certifications processes, I have made them manually.

As previously noted, a full decode of the source would pick up details not intended for a flowchart.

Think just about things like:

if (trace_spi) emit("SPI: read %u written %u,spi_bytes_read,spi_bytes_written);

That is one extra condition, so an automatic tool that doesn't make use of special tags in the source code would need one extra block for the printout, together with a conditional block for checking if trace is enabled.

Most of the time, the flowcharts are not intended to be read by a programmer, so they should be at a different conceptual level from the code.

During the certification process, there may be a special interest in specific boxes in the flowchart, which is then covered as text or a broken-out flowchart. The flowchart gives a principal overview of the decision tree, while other documentation might be needed to document _why_ specific decisions are needed or specific time constants etc.

It is so easy to produce unreadable flowcharts because of too detailed information.