Skip to content

MON 262B (octal) - GetSystemInfo (CPUST)

Returns system information into a caller-supplied 12-word (24-byte) buffer: the system number, the CPU type, the SINTRAN III version, the instruction set, the patch indicator, and the system generation time. The input parameter Number should always be 0. This is an ND-100 monitor call.

Status: byte-verified (dispatch + worker body). The worker is real carved code in 006-S3FS. All addresses/values are octal.

CORRECTED 2026-07-13. The previous version of this folder said the call was misattributed, claiming GOTAB[262B] = 066262B routed to an F1737 "compiler dispatch descriptor" in 025-S3IRPIT, and that the F1737 -> CPUST hop was unproven. That was an artefact of the wrong dispatch model. MON calls are not dispatched through that GOTAB (it is MFELL for 224 of 256 calls); the real worker address comes from MCTAB @ 005620B in segment 044-S3IDPIT. MCTAB[262B] = 063022B = CPUST, byte-proven, and that worker IS carved in 006-S3FS. The old 066262B/F1737 reading came from the disproven "uncarved CALLPROC bridge" story. See ../317B-ExecuteCommand/README.md and SINTRAN/CARVING-HANDOFF.md section 3a.

Dispatch path

flowchart LR
    A["User program<br/>MON 262B"] --> B["ENT14 level-14 entry<br/>026-S3IMPIT :072167B"]
    B --> C["GOTAB[262B] = MFELL<br/>MGOTA=071233B :071515B"]
    C --> D["MFELL level switch<br/>026-S3IMPIT :072114B -> CALLP 032201B"]
    D --> E["MCTAB[262B] = CPUST<br/>MCTAB=005620B :006102B = 063022B"]
    E --> F["CPUST worker<br/>006-S3FS :063022B"]
    class A blue
    class B,C,D,E teal
    class F green
    classDef blue fill:#E3F2FD,stroke:#0D47A1,color:#0D47A1
    classDef teal fill:#E0F7FA,stroke:#00838F,color:#00838F
    classDef green fill:#E8F5E9,stroke:#2E7D32,color:#2E7D32

There is no dashed hop. GOTAB[262B] = MFELL (byte-proven, 74 4c); MFELL is a program-level switch (it writes CALLP = 032201B into the monitor level's P), not a subroutine call. The monitor level then indexes MCTAB and reaches CPUST.

Code location (dispatch path)

Byte offset = (addr - loadbase) in octal words x 2 (decimal). Every offset below was reproduced with dd (see "Verify by hand").

Role Segment Addr (octal) Byte offset Symbol Verdict
GOTAB[262B] slot 026-S3IMPIT.asm 071515B = 072114B 32410 -> MFELL VERIFIED
MCTAB[262B] slot 044-S3IDPIT.asm 006102B = 063022B 2180 -> CPUST VERIFIED
CPUST worker body 006-S3FS.asm 063022B-063301B 29732 CPUST VERIFIED

Verify by hand (from tools/sintran-segment-carver/versions/L-VSX-500/segments/): 

dd if=026-S3IMPIT.bin bs=1 skip=32410 count=2 | od -An -tx1   ->  74 4c   (= 072114B = MFELL)
dd if=044-S3IDPIT.bin bs=1 skip=2180  count=2 | od -An -tx1   ->  66 12   (= 063022B = CPUST)
dd if=006-S3FS.bin    bs=1 skip=29732 count=2 | od -An -tx1   ->  22 4e   (= 021116B, CPUST entry: STD I 116)

Instruction walkthrough

Full listing: 262B-GetSystemInfo.ASM.

Prologue (063022B-063027B). STD I 116 / RADD CLD SL DA / RADD CLD SB DD / SAB 43 save the caller's registers and set up the local frame base; JPL I 113 (-> 063141B) calls the standard FILSYS entry helper, then STX ,B 11 stashes the caller's X.

Parameter check (063030B-063045B). Loads ,B 2 (the caller's Number parameter), and on one path branches to the error return (JMP I 110 -> 063142B); it then range-checks against the table bounds LDT 105 / LDT 103 and jumps to the error path 063144B when out of range. (The manual states Number should be 0; the bytes range-check it.)

System-info block copy (063046B-063066B). The routine builds source/destination pointers (RADD CLD SA DL / RADD CLD SA DD, count in T via SAT 25 / SAT 37) and executes two MOVEW block-move instructions (063054B, 063063B) that copy the resident system-info words (system number, CPU type, version, instruction set, patch indicator, generation time) into the caller's 12-word buffer. The worker closes cleanly at 063301B.

The overall shape - prologue, parameter validation, then MOVEW copies into the caller buffer - is byte-proven; the exact field layout of the copied block is inferred from the manual.

Parameter / register contract

Reg / field Dir Meaning Verdict
,B 2 (Number) in selector; manual says always 0; range-checked in bytes VERIFIED (bytes); meaning inferred (manual)
caller buffer out 12-word (24-byte) system-info block filled by two MOVEW copies VERIFIED (bytes: MOVEW at 063054B/063063B); field layout inferred (manual)
X in saved at 063027B (STX ,B 11) VERIFIED (bytes)
return out via saved link; error path at 063142B/063144B VERIFIED (bytes)

Pseudo-code (for an emulator)

See 262B-GetSystemInfo.pseudo.c - the prologue, the parameter range-check and the two MOVEW block copies are byte-verified; the exact system-info field layout is inferred from the manual.

Honest caveats

What is byte-proven: GOTAB[262B] = MFELL; MCTAB[262B] = 063022B = CPUST; the CPUST entry bytes at 063022B in 006-S3FS match the disassembly (021116B = STD I 116); the worker is a real handler that validates Number, then copies a system-info block into the caller's buffer with two MOVEW instructions. The MCTAB identification is not a one-slot coincidence: 216 of 256 slots land exactly on named L07 symbols, and known anchors (MON 005B -> RDISK, MON 200B -> XMSG 007516B, MON 144B -> MAGTP) all match.

What is NOT proven: the precise field-by-field layout and byte order of the 12-word block (system number / CPU type / version / instruction set / patch indicator / generation time) - that ordering comes from the manual, not from these bytes.

Correction to earlier work. The old folder read the disproven GOTAB and claimed GOTAB[262B] = 066262B -> F1737 -> CPUST with the last hop unproven. The real dispatch is GOTAB[262B] = MFELL -> MCTAB[262B] = CPUST, fully byte-proven. The NC oracle depends on this call; its worker is now byte-anchored at 063022B.

Method: ../../../../../EXTRACTING-RESIDENT-CODE.md - master map: ../../MON-CALL-INDEX.md.