- Remove wall-of-text comment block (moved to PR description)
- Simplify output: "Genuine" vs "Weak" per Iceman's feedback
- Remove explanatory INFO line
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The existing mfulc_fingerprint() detects known counterfeit families
(Lab401 DW, Feiju FJ8010, USCUID-UL v1/v2, Giantec GT23SC4489,
Hospitality) by their deterministic command and CRC tells, returning
early on a match. This commit adds a generic weak-PRNG catch-all for
clones that evade those fixed profiles.
mfulc_prng_test() runs on the genuine-looking fallthrough, right before
the "likely MF0ICU2" verdict. It collects RndB nonces via repeated 1A00
auth starts and flags a duplicate:
- Genuine NXP draws RndB from a hardware TRNG (2^64 space), so it
never repeats across a realistic batch.
- Giantec/USCUID use a 16-bit LFSR whose wakeup-timing seed
concentrates the observed pool to roughly 2^6.5, so a duplicate
appears within 12 authentications.
A single collision is proof of a weak PRNG. This backstop catches
weak-RNG clones, including future variants the profile table does not
know. MFULC_PRNG_SAMPLES (40) and MFULC_PRNG_CLEAR (39) are tunable.
Tested on genuine NXP MF0ULC (UID 04 1C BE 92 E7 7A 80): 40 unique
nonces, no collision, verdict "consistent with a genuine TRNG".
Ref: "BREAKMEIFYOUCAN!", eprint.iacr.org/2026/100, Section 11.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012e7xUDfDUXJQU7SWV3ULS1
The sim sent ATQA 02 00 for 4, 7 and 10 byte UIDs, because the ATQA
override replaces the size bits the firmware sets. The PN533 of an
SCL3711 stopped after cascade level 1 and never found a 7 or 10 byte
UID. The ATQA now codes the size of the UID the device simulates, 4
bytes from emulator memory when the UID is all zero, and the start
message says when that happens.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
clone took negative FC and CN, and MS values above 5 and 22 bits,
without a word: MS FC 40 wrote the card of FC 8. sim let ids of 32
bits or more change the fixed bits, so --id 2147483648 simulated id 0,
and printed the shifted frame as Internal Id. The sim help example
used --cn, the option is --id, and the clone hint named lf keri read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
clone and sim accepted FC up to 999 and CN up to 99999, but the frame
holds 8 bits of FC and 16 of CN, so FC 300 wrote the card of FC 44.
The clone hint also named lf paradox read, which does not exist.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
HPP32 wrote its card number 10 bits past the 32-bit frame, it now uses
bits 13-30 as cardinfo shows. H10320 wrote its first digit over the
leading one, so decode refused most numbers, and two of its parity
masks had typos. S12906b needed ParityValid, which no caller sets.
HGen37 decode checked bit 28 instead of 26. The limits of nine formats
did not match the fields Pack writes. wiegand list printed the usage
of wiegand info.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A NULL TLV has no length, so a run of them swallowed the TLVs after
it. Lock and Memory Control sizes of 0 mean 256, the nibbles of their
third byte are powers of two and the Memory Control size counts bytes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Enter leaves a simulation running on the device, by design. The next
hf thinfilm sim stopped it, took the PM3_EOPABORTED reply left behind
for its own and printed "Done!" at once. A ping round trip before
starting drops that reply.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Resolve CHANGELOG.md: keep both the EM4x50 entries and the flash and
CI entries from #3748 and #3750.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The values went through uint8_t or & 0xFF, so --max 300 ran with 44
and --samples 1000 with 232. The --fs clocks stay 8-bit in the FSK
demod and are now refused above 255.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Since cd5906674 the flash animation runs whatever stdout is, and a
redirected flash of fullimage.elf wrote a 7 MB log.
flash_pm3() replaced a flash error with the result of
flash_stop_flashing(), always PM3_SUCCESS, so a failed flash ended with
"All done". main() exited 0 after --flash, --dumpmem and
--reboot-to-bootloader whatever happened. dumpmem printed its negative
error code with %u.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The gen2 branch refused `-u` with "Not supported on this card". These
cards do take a UID write: a five command knock, the new UID one byte
per command (always seven commands, zero padded), then a HLTA which
commits it. The card answers the halt instead of staying silent, and
without it everything before is discarded when the field drops - which
is why the sequence is absent from the vendor app, whose HLTA is emitted
by the Android NFC stack on teardown.
The new UID must match the card's current UID length, the parts are
physically 4b or 7b. Outcome is decided by re-selecting and comparing
the UID, not by the status byte.
Verified on a 4 byte card. The 7 byte path is inferred from the sequence
always carrying seven UID bytes and is untested.
A refused -r / -u now returns an error instead of PM3_SUCCESS.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Every bidirectional EM4x50 command (rdbl, info, dump, lf search) failed
on a PM3 Easy with a 500 uH antenna, while `lf em 4x50 reader` worked.
A raw ADC capture of the exchange showed RM being sent into the tag data
right after the double listen window: find_em4x50_tag() consumes the
first window, the second one is then taken as the first, and
invalid_bit() cannot reject data because its single 3/4-bit sample of
the AC-coupled, saturating signal always sits between g_Low and g_High.
- detect the second listen window by the absence of a rising edge over
its 128-cycle low phase (Manchester data has one per bit)
- `lf em 4x50 config` (CMD_LF_EM4X50_CONFIG): threshold, pulse
tolerance, "0" bit gap start/end and RM delay, changeable at runtime
- defaults moved to the centres of the working ranges measured with the
new lf_em4x50_tune.lua sweep: tol 10 (5-15), gap end 38 (28-50),
rm 132 (104-164); the old gap end 32 was 4 cycles from failing
- rdbl/info retry the login + selective read for up to 1 s
- at debug level extended the whole exchange is captured into BigBuf,
fetch it with `data samples`
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Large GET DATA returns 61 xx. Concatenate clear 00 C0 bodies before
unwrap so 255-byte AES and 2K3DES objects print instead of exit 246.
Co-Authored-By: Cursor Grok 4.6 <noreply@cursor.com>
hf seos pacs on 2K3DES cards exited 236 with no message after mutual
auth. Print ExchangeAPDU14a failures, check mbedtls DES-CMAC without
mutating the caller key, decode SELECT 85 at its real length, and emit
short-form 85 when the cryptogram is under 128 bytes.
Co-Authored-By: Cursor Grok 4.6 <noreply@cursor.com>
Bootloader memory-dump cleanup sends CMD_PING to wake the communication
thread after clearing its run flag. The thread can be waiting for another
transmit after the dump ACK, so it sends this command before exiting.
The non-BWM bootrom does not handle CMD_PING: its default dispatch enters
Fatal(), leaving the device unresponsive even though the dump succeeded.
Use CMD_DEVICE_INFO for the wakeup instead. It is already supported by the
bootloader and used to identify it before the dump. Keep the wakeup so
closing the client does not leave the communication thread waiting.
Reproduced on PM5: the dump completed, but the following reset and device
probe failed. With only the rebuilt client changed, the same test script
passed five OS-to-bootloader, memory-dump and reset-to-OS cycles, with a
successful ping after each. No firmware was flashed for this comparison.
Built the client with AppleClang through Make, CMake and experimental_lib;
client/check passed. GCC 16 compiled the changed translation unit with
-Wall -Werror. Existing macOS linker warnings remain. The full GCC host
build was not repeated because of the previously observed Apple SDK
alignment error. Linux and Windows were not tested; this change replaces
an existing protocol command without changing platform-specific code.
Problem
The native hf mfu countertear used a device-side
CMD_HF_MFU_COUNTER_TEAROFF path with a static increment payload,
decided once per run. The pyscript mfuev1_counter_reset.py implements
the same attack with auto-calibration, a 3-phase attack loop,
per-attempt BD delay tuning (hunting) and stuck-value handling,
which the native command lacked.
Change
- rewrite the client command around the generic tear-off hook
(hw tearoff) and raw INCREASE_CNT frames, mirroring the pyscript
step by step: preflight checks, CHECK_TEARING_EVENT support gate,
targeted repair/normalize, BD sweep calibration (working delay
accepted only on 0xBD), and the DECREMENT -> PREPARE (0x00FFFF)
-> RESET phases with the same increment payloads (000100 for
decrement steps, 010000 for reset/repair)
- same delay tuning on counter movement: +1 when unchanged, -1 when
a jump exceeds 10, delay locked after a successful decrement
- remove CMD_HF_MFU_COUNTER_TEAROFF usage from the client; device
side stays untouched
- add Lua port client/luascripts/mfuev1_counter_reset.lua with the
same behaviour, using core.console capture and Enter-to-stop
- regenerate doc/commands.md and doc/commands.json
Behaviour change
Same CLI shape as the pyscript (-c/-f/--DelayBD/--Delay00, legacy
-i/-l/-s/-x options removed). Console output now matches the pyscript.
Testing
gcc warningless for client, armsrc and bootrom (RDV4). make check
green. doc regenerated with GD enabled. Verified 1:1 against the
pyscript and the new Lua port on hardware (EV1 card). Not tested
here: clang, Windows/ProxSpace and macOS. No build files are touched,
so the usual client-side review applies.
Co-Authored-By: opencode agent (GLM-5.3-Flash-Free) <noreply@opencode.ai>
Validate counter choices 0-2, treat only 0x00/0xBD as valid
CHECK_TEARING_EVENT answers, and detect NACK replies from the raw
payload, so retry logic does not mistake a NACK for a real answer.
Building the client with -march=x86-64-v3 (the CentOS/RHEL 10
baseline) fails with gcc 14:
src/emv/emvcore.c:211:24: error: writing 32 bytes into a region of
size 22 [-Werror=stringop-overflow=]
gcc vectorizes the dash fill loop in aidlist_rule() into 32-byte
stores and loses the pos + 2 < dstlen bound. The bound was correct.
Compute the dash count with the same limit and fill it with memset,
so there is no vectorized loop store to mis-size. Output is unchanged.
Reported in #3671.
Co-Authored-By: Grok Bot <noreply@users.noreply.github.com>