Debugging Support in Hardware

In addition to JTAG (external control)
      the processor usually offers a debug register block

This is a set of registers that the CPU itself can use
Single-step control
If set, returning from interrupt executes one instruction only
Breakpoint registers
Usually the debugger writes to code space
But in the uC world it's not possible, you need hw support
Watchpoint registers (write and/or read)
Doing this in software is awfully slow

The Cortex-M architecture specifies an optional debug unit
And the 11U family does not implement it
So no breakpoints, ever.

Unless we wisely instrument the code
Any ideas about how to do it?