make usart write buffer const
sub out magic numbers with defines
fix edge case handling 255/256 byte frames (including crc)
add sanity checks to avoid buffer overrun on some "should never happen" edge cases
don't wait for rats reply from card before listening to next reader frame
cap fsci to 8 (256 bytes) as that's the most the proxmark3 codebase currently handles
eliminate 1k of ram usage by tweaking how emulation responses are sent
this could happen if TC2 was already 0 when it was started or reset
resulting in the initial reset not happening until TC0 had overflowed for
the first time as the delay to ensure this didn't happen would be missed
when TC2 was already 0
the new behaviour results in TIOA0 being toggled when a software trigger
of TC0 happens which makes TC2 reset immediately without having to wait
for TC0 to overflow
Fixes keyCount_mutex[0] and keyCounts[0] not being properly initialised
Fixes progress reporting not properly clearing the line
Switch malloc(a*b) style calls to calloc(a, b)
This was implemented with a new pair of RPCs CMD_HF_MIFARE_READBL_EX and CMD_HF_MIFARE_WRITEBL_EX
these RPCs support all combinations of read/write commands, wakeup, and auth options so
in time can replace the other MFC read/write commands too reduce armsrc code size
and complexity.
Also added config parsing for the gdm cfg block when reading with hf mf gdmcfg and
explicitly with hf mf gdmparsecfg.
This was broken in commit 17ab86c52 as the forced rounding up of
the size to 4-byte alignment in BigBuf_malloc made the size check
possibly larger than the buffer size as the check was always +3 on
the requested size rather than the rounded size. This was made
worse by BigBuf_max_traceLen not taking into account alignment
either and the alignmentn check in hfsnoop.c checking to 2 byte
alignment instead of 4 byte alignment.
The alignment size check now checks the size after alignment
rounding, and BigBuf_max_traceLen takes into account alignment
losses too reducing the need for BigBuf consumers to have to care
about alignment.
Allow button presses to break
Handle non-zero CID from reader by relaying RATS to the card and improving PPS and WTX handling
More reliably cook ATS values
Ignore packets that look like they're coming from the client (NG packets)
Magic detection no longer stops when a single type is found as cards may support multiple types of magic, so all detected types will be reported now.
GDM/USCUID chips are now detected when GDM magic auth is disabled but magic WUP (40 or 20) is enabled.
Gen2/CUID/DirectWrite is now detected when default keys and ACLs are used by attempting to write to block 0 but aborting before actually completing the write.
reliability. Now relies on the legacy config block for SR detection
and the end-of-SIO detection no longer partially cuts off the SIO
for any dumps I have.
This fixes a regression caused by edd8e36 (PR #2021).
gdm auth has no concept of A or B keys, a partial cleanup in the
previous PR removed the key type from the arm code, but the
corresponding removal was missed in the client side resulting in
mismatched structs for cmd CMD_HF_MIFARE_G4_GDM_WRBL.
Previously reader durations would be reported to be twice as long
in the trace vs reality, and as a result the start time would be
earlier than it should have been as well.
Also fixes a minor sniff trace timing with card dual subcarrier
responses.