48ee413c9b
v1.2.0: Add 4 new protocols, fix 5 existing, port Kia V6 encoder
...
New protocols (from ProtoPirate):
- Mazda V0: 433MHz, pair-based decoding, XOR deobfuscation
- Mitsubishi V0: 868MHz, PWM, bit negation + XOR unscrambling
- Porsche Touareg: 433/868MHz, sync preamble, 24-bit rotate cipher
- Fiat V1 (Magneti Marelli BSI): 433MHz, auto-detected timing variants
Protocol fixes aligned with ProtoPirate reference:
- Kia V1: Fix Manchester level mapping (inverted convention), off-by-one
bit count, CRC4 simplified to 7 bytes + offset 1
- Kia V2: Fix CRC byte layout (was double-counting nibbles), off-by-one
bit count, preamble short-HIGH handling
- Fiat V0: Fix endbyte transform (remove (<<1)|1), standard Manchester
encoder (was differential), 7 btn bits (was 6), gap path fallback
- PSA: Fix modified TEA (XTEA-like dynamic key selection), XOR decrypt
byte mapping, critical key2_low construction bug, encoder polarity and
preamble, add key1_high nibble validation, dual preamble patterns
- Ford V0: Add calculate_checksum (sums all serial+count bytes) for encoder
Kia V6 encoder ported:
- Forward AES-128 (SubBytes, ShiftRows, MixColumns, encrypt)
- encrypt_payload: build plaintext, AES encrypt, pack into 3 parts
- Two-pass Manchester upload (640 + 38 preamble pairs)
- fx_field extraction stored in DecodedSignal.extra for encode roundtrip
Updated README, protocol docs, and version bump to 1.2.0.
2026-03-22 12:05:01 -04:00
84b75ee6b5
Fix ARM compilation: vendor libhackrf with c_char portability fix
...
The upstream libhackrf v0.1.1 crate fails to compile on ARM because
c_char is u8 on ARM (not i8 like on x86). The transmute at lib.rs:102
explicitly casts &[i8] -> &[u8], but on ARM the source is already &[u8],
causing a type mismatch.
Vendored the crate at vendor/libhackrf/ with the transmute replaced by
slice::from_raw_parts with a pointer cast, which works on both
architectures. Also cleaned up transmute calls in transfer.rs.
Fixes #1
2026-03-22 11:14:41 -04:00
KaraZajac