Fix/improve 'lf cotag demod' auto threshold estimation
and demodulation of COTAG traces with high value spikes and
COTAG traces with very low amplitude values.
Using "data load" on very large trace files (e.g. dumps with 700k samples,
such as COTAG dumps) was causing crashes on some platforms due to stack overflow
since VLA temporary buffers were used.
Replace VLAs in computeSignalProperties() and removeSignalOffset() with
heap-allocated buffers instead.
- "lf cotag reader": now behaves mostly the same as "lf read",
except it is dedicated for COTAG.
- It makes sure to send COTAG start sequence and to use
real time sampling mode to get enough samples.
- "lf cotag demod": Working client-side COTAG demodulation implementation
(high-low raw demod + Manchester demodulation from high/low demod).
- Now uses samples directly from g_GraphBuffer
- Supports 3 types of COTAG preamble (1 active, 2 passive), which
can start anywhere in the samples.
- Prints full 128-bits of COTAG tag data in 3 formats: raw,
groupped by nibbles, and in hex, and identifies and
prints out the card number.
Signed-off-by: andrej@towelbyte.net
towelbyte