restore_bufferS32() returns the length it put back and every graph-buffer
caller dropped it, so g_GraphTraceLen kept whatever value the code between
save and restore had left.
In try_detect_modulation() that is the 160 sample psk antenna-settle trim:
`data modulation` gave the buffer's samples back but left it 160 short, and
a following demod lost round(160 / clk) bits off the end. check_chiptype()
loses the same way through the acquisitions its detections make, on the
`lf search` path, and for g_DemodBuffer too. demodTI() loses convLen + 16
on its error path.
save_graphbuffer() / restore_graphbuffer() pair the length and grid offset
with the contents so no call site has to remember them. The visa2000 and
zx8211 sites do not change the length, so converting them is a no-op today
and keeps the next one from being a bug.
Regression test in tools/pm3_tests.sh: `data modulation` on a saved psk1
trace must leave all 20000 samples.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
GetT55xxBlockData extracted every block at config.offset, a bit index into a
demod buffer cached by the last detect - a different capture. Demodulators do
not all start on the same bit, so a later read could return a rotation of the
block with nothing marking it as wrong: psk1 by four or five bits, fsk2a and
nrz by one, manchester and biphase never.
Record the boundary where it is stable instead. Every demodulator reports the
graph sample its first bit sits on, and turn_read_lf_on(137 * 8) makes the
firmware open each read acquisition at the same point, so the word boundary
lands on the same sample every time. Convert back per demodulation.
Block 0 re-anchors from its known value, and a write is now verified by the
value being anywhere in the repeating stream rather than at one cached offset -
which also fixes correct writes reporting as validation failures. A block 0
verify re-detects on a mismatch, not only on a decode failure: the pre-write
config can decode the post-write signal into garbage rather than failing.
Also fixes a -1 password sentinel truncated to 32 bits by a ternary, which made
that detect report a password that was never supplied.
Measured on a T5577 across psk1, fsk2a and direct/nrz: three consecutive dumps
agree field for field, against a baseline where psk1 corrupted 22 of 24.
Co-Authored-By: Claude Opus 5 (1M context)
Send a synchronous zero-length packet after CDC responses whose length is
an exact multiple of the endpoint packet size, preventing oversized host
reads from remaining pending.
# Updates
1. Updated `MEGACASH` vendor name from `Bank Mega` to `PT Bank Mega Tbk` *which is the vendor's full name*.
-r&y.
Signed-off-by: ry4000 <154689120+ry4000@users.noreply.github.com>
# Updates
1. Updated `Fermax` AID to reflect the parent company's name/country and DESFire PACS product.
Signed-off-by: ry4000 <154689120+ry4000@users.noreply.github.com>
# Updates
1. Changed `type` from `e-amusement` to `arcade` for:
- Bandai Namco Passport
- Konami e-amusement pass
- Sega Aime
- Taito NESiCA
2. Updated `description` to include the full name and `(AIC)` annotation to note that this information only applies to the Amusement IC Card variants as opposed to MFC/ICODE SLI/MFC/MFU, respectively.
-r&y.
Signed-off-by: ry4000 <154689120+ry4000@users.noreply.github.com>