Skip to content

MON 317B (octal) - UECOM (ExecuteCommand)

Executes a SINTRAN III command line as if it had been typed at the terminal (without the leading @). The command text is passed as one string. Errors print a message and return - they do not terminate the calling program.

Status: byte-verified end to end. Every layer of the dispatch - the level-14 entry, the GOTAB slot, the MFELL level switch, the MCTAB slot and the UECOM worker body - is carved and reproduced from bytes. All addresses/values are octal.

Dispatch path

flowchart LR
    A["User program<br/>MON 317B"] --> B["ENT14 level-14 entry<br/>026-S3IMPIT :072167B"]
    B --> C["GOTAB[317B] = MFELL<br/>MGOTA=071233B :071552B"]
    C --> D["MFELL level switch<br/>026-S3IMPIT :072114B"]
    D --> E["CALLP on the monitor level<br/>:032201B"]
    E --> F["MCTAB[317B] = UECOM<br/>MCTAB=005620B :006137B"]
    F --> G["UECOM worker<br/>003-S3CP :050701B"]
    G --> H["shared command body<br/>upcase + SINTRAN decoder"]
    class A blue
    class B,C,D,E,F teal
    class G,H 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. The link earlier documents called the "uncarved CALLPROC bridge" is MFELL at 072114B, and it is carved: it is a program-level switch (IRW 20 DP writes CALLP into the monitor level's P register), not a subroutine call. See Honest caveats.

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
level-14 monitor entry 026-S3IMPIT.asm 072167B-072260B 33006 ENT14 VERIFIED
GOTAB base 026-S3IMPIT.asm 071233B 32468 (slot) MGOTA VERIFIED
GOTAB[317B] slot 026-S3IMPIT.asm 071552B = 072114B 32468 -> MFELL VERIFIED
MFELL level switch 026-S3IMPIT.asm 072114B-072123B 32920 MFELL VERIFIED
CALLP literal 026-S3IMPIT.asm 072161B = 032201B 33014 CALLP VERIFIED
MCTAB[317B] slot 044-S3IDPIT (not yet in segments-ref) 006137B = 050701B 2238 -> UECOM VERIFIED
UECOM worker body 003-S3CP (not yet in segments-ref) 050701B-050725B 17282 UECOM VERIFIED
shared command body 003-S3CP (not yet in segments-ref) 050740B-050771B 17314 (shared) VERIFIED

044-S3IDPIT and 003-S3CP are carved but have not yet been promoted into segments-ref/; their canonical bins are tools/sintran-segment-carver/versions/L-VSX-500/segments/044-S3IDPIT.bin and .../003-S3CP.bin. Promoting them is the follow-up action (see Honest caveats).

Verify by hand (from the carver's versions/L-VSX-500/segments/ directory): 

dd if=026-S3IMPIT.bin bs=1 skip=32468 count=2 | od -An -tx1   ->  74 4c   (= 072114B = MFELL)
dd if=044-S3IDPIT.bin bs=1 skip=2238  count=2 | od -An -tx1   ->  51 c1   (= 050701B = UECOM)
dd if=003-S3CP.bin    bs=1 skip=17282 count=2 | od -An -tx1   ->  cc 61   (= 146141B, UECOM entry)

Instruction walkthrough

Full listing: 317B-ExecuteCommand.ASM.

The dispatch (072253B-072260B). T holds the MON number (masked to 8 bits by the constant 000377B at 072266B, so GOTAB is exactly 256 words). The fetch 057020 LDX I ,X 20 resolves through the literal at 072276B = 071233B = MGOTA, giving X := MEM[MGOTA + N] = GOTAB[N]; then 126000 JMP ,X jumps straight to it. The fetch is bracketed by BSET ZRO SSPTM / BSET ONE SSPTM, i.e. the table word is read through a different page-table mapping than the code that runs next. This is a direct jump - no call.

GOTAB is not the monitor-call table. Only 32 of its 256 slots are real resident handlers (MON 1B read, 2B write, 21B-24B, 63B, 163B, 200B XMSG, 310B, 346B-377B); those arm the B-level through IOB14 = 071660B. The other 224 slots all hold MFELL, including MON 317B.

MFELL (072114B) is the hand-off, not an error. It copies the MON number into the monitor level's X (IRW 20 DX), writes CALLP = 032201B into that level's P (IRW 20 DP), and activates the level (MST PID / MST PIE). The monitor level then dispatches through MCTAB = 005620B (alias 9MCTA), the real Monitor-Call TABle: one word per MON number, 216 of 256 slots populated, every populated slot landing exactly on a named L07 symbol. MCTAB[317B] at 006137B = 050701B = UECOM.

UECOM (050701B) shares one body with two siblings. COMSB (MON 070B) at 050673B, UECOM (MON 317B) at 050701B and UELOG (MON 320B) at 050726B are three entries into one command-execution body. Each saves the return link (RADD CLD SL DD), calls a common prologue (JPL I ,B -37), stores the caller's string pointer at ,B -177, writes a mode code to ,B -175 (COMSB=1, UECOM=2 or 4, UELOG=3) and jumps to the shared body at 050740B.

That single fact explains the manual's headline difference: MON 70B (COMND) and MON 317B (UECOM) are the same code with a different mode code, which is why one terminates the caller on error and the other returns.

Shared body (050740B-050771B). Loads the command pointer (LDX ,B -177), walks the text with LBYT/SBYT and upper-cases it in place ('a'=141B .. 'z', AAA -40), then calls the standard SINTRAN command decoder via JPL I 41 at 050746B.

Parameter / register contract

Reg / field Dir Meaning Verdict
A in address of the command string VERIFIED (bytes: 050705B STA ,B -177, and the manual)
T in range-checked at 050710B-050715B; selects mode 4 vs mode 2 VERIFIED (bytes); meaning of the two variants inferred
,B -177 frame caller's command-string pointer VERIFIED (bytes)
,B -175 frame mode code: COMSB=1, UECOM=2/4, UELOG=3 VERIFIED (bytes)
,B -200 frame caller's T VERIFIED (bytes)
command text in/out upper-cased in place before decoding VERIFIED (bytes)
error behaviour out prints a message and returns; does not terminate the caller VERIFIED (official manual ND-860228-2 p.180-181); not byte-proven here
completion out synchronous - returns after the command has run VERIFIED (manual); not byte-proven here

Note the string is modified in place (upper-cased). An emulator that passes a read-only buffer will diverge from real L07.

Pseudo-code (for an emulator)

See 317B-ExecuteCommand.pseudo.c - a pseudo-C model of all four dispatch layers plus the worker. The dispatch chain and the mode-code fork are byte-verified; the command-decoder internals are inferred (that body is reached by JPL I 41 at 050746B and is not carved in this folder).

Honest caveats

What is byte-proven: the whole dispatch chain. GOTAB[317B] = MFELL; MFELL loads CALLP and switches program level; MCTAB[317B] = UECOM = 050701B; the UECOM entry bytes at 050701B in 003-S3CP match the disassembly; COMSB/UECOM/UELOG share one body and are distinguished only by the mode code. The MCTAB identification is not a one-slot coincidence: 216 of 256 slots land exactly on named L07 symbols (RDISK, WDISK, CIBUF, OPFIL, MAGTP, DEBUG, CPUST, MOINF, UECOM, ...), and MON 200B -> 007516B is XMSG exactly where it must be.

What is NOT proven: - The meaning of the two UECOM mode variants (4 vs 2, chosen by a range test on T at 050710B-050715B). The fork is in the bytes; what it selects is not. Do not guess it. - The command decoder body itself (JPL I 41 at 050746B) is not carved here. - The behavioural guarantees (synchronous, non-terminating on error, prompts for missing parameters, output to terminal) come from the official manual, not from these bytes.

Correction to earlier work. An earlier analysis read GOTAB out of the SINTRAN-DATA_commoncode carve at 071233B and reported GOTAB[317B] = 112242B with the handler "not on disk". That was wrong: commoncode's 071233B is not the GOTAB (its slot 0 is 000000, not MFELL; its slot 1 is 120303B, not M1=071633B). The real GOTAB is in the monitor-PIT segments (017-S3SMPIT / 026-S3IMPIT), which is also where ENT14 actually lives. The 112xxx "entry addresses" were unrelated bytes at that address in a different overlay. See ../../MON-CALL-INDEX.md.

Follow-up: promote 003-S3CP and 044-S3IDPIT into segments-ref/ so the two tables in the Code-location table above resolve as links like the others.

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