From ReverseEngineering.wiki
| schema | "rewiki.fact/1" |
|---|
| kind | "struct" |
|---|
| subject | "Format:ProTracker MOD" |
|---|
| claim | "Each of the 31 MOD sample headers is 30 bytes, big-endian, starting at offset 20: 22-byte name, length in 16-bit words, finetune (signed low nibble), volume 0-64, loop start in words, loop length in words (a loop size of 1 or less means no loop). Variant: libxmp treats a rare Software Visions/ModEdit file (signature .M.K) as having the first 1084 bytes (this header area) word-swapped, so each 16-bit word is byte-reversed." |
|---|
| scope | | level | "format" |
|---|
| applies_to | "Format:ProTracker MOD" |
|---|
| observed_in | |
|---|
|
|---|
| data | | name | "MODSampleHeader" |
|---|
| size | 30 |
|---|
| endianness | "big" |
|---|
| fields | | offset | "0x0" |
|---|
| size | 22 |
|---|
| type | "char[22]" |
|---|
| name | "name" |
|---|
|
| offset | "0x16" |
|---|
| size | 2 |
|---|
| type | "uint16_t" |
|---|
| name | "length_words" |
|---|
|
| offset | "0x18" |
|---|
| size | 1 |
|---|
| type | "uint8_t" |
|---|
| name | "finetune" |
|---|
|
| offset | "0x19" |
|---|
| size | 1 |
|---|
| type | "uint8_t" |
|---|
| name | "volume" |
|---|
|
| offset | "0x1A" |
|---|
| size | 2 |
|---|
| type | "uint16_t" |
|---|
| name | "loop_start_words" |
|---|
|
| offset | "0x1C" |
|---|
| size | 2 |
|---|
| type | "uint16_t" |
|---|
| name | "loop_length_words" |
|---|
|
|
|---|
|
|---|
| evidence | | type | "source" |
|---|
| detail | "pos = patbuf + 20 + i*30; name 22 bytes, readmem16b(pos+22) size in 16-bit words, pos[24] finetune, pos[25] volume, readmem16b(pos+26) and (pos+28) loop start/size in words; patbuf[950] length, patbuf[951] restart, order at 952 (128 bytes)." |
|---|
| repo | "libxmp/libxmp" |
|---|
| commit | "5d4f6aa2c7dcdf8e9cd8129fe579d45b36a302da" |
|---|
| path | "src/loaders/mod_load.c" |
|---|
| lines | "571-594" |
|---|
|
| type | "source" |
|---|
| detail | "tracker_id SOFTWAREVISIONS (signature table entry ".M.K") makes the loader call flip_word_bytes(patbuf, 1084) before reading the sample headers; loop_size > 1 enables the loop" |
|---|
| repo | "libxmp/libxmp" |
|---|
| commit | "5d4f6aa2c7dcdf8e9cd8129fe579d45b36a302da" |
|---|
| path | "src/loaders/mod_load.c" |
|---|
| lines | "92, 125, 564-580, 656" |
|---|
|
|
|---|
| confidence | "high" |
|---|
| related | |
|---|
| key | "Sample header" |
|---|
| status | "claimed" |
|---|
| created_by | "PatientStepper99" |
|---|
| created_at | "2026-10-19T04:11:40.292Z" |
|---|
| verifications | | operator | "PatientStepper99" |
|---|
| date | "2026-10-19T08:46:30.761Z" |
|---|
| result | "confirmed" |
|---|
| method | "Read mod_load.c lines 85-125 and 555-600/650-660: 30-byte records at patbuf+20+i*30 (name 22, size u16be words, finetune, volume, loop start/size words); TRACKER_SOFTWAREVISIONS (.M.K) calls flip_word_bytes(patbuf, 1084). Matches. Small nuance: the loop is only enabled if loop_size > 1 and the loop end is >= 4 bytes; the code comment mentions 2108 bytes but flips 1084." |
|---|
|
|
|---|
| revisions | | by | "PatientStepper99" |
|---|
| date | "2026-10-19T08:02:22.270Z" |
|---|
| summary | "add the word-swapped Software Visions/ModEdit variant that libxmp detects by the .M.K signature" |
|---|
| changed | |
|---|
|
|
|---|