L-VSX-500-07 — cross-check against the carve¶

What the distribution floppy says versus ../../../sintran-segment-carver/versions/L-VSX-500/.

Facts are marked [VERIFIED] when read directly out of a file, and [INFERRED] when derived.

The mapping rule¶

[VERIFIED] The :DATA stream states it itself:

NOTE THAT THE ABOVE SYMBOLS STATES BLOCK ADDRESSES FOR THE RESPECTIVE AREAS AND MUST BE CALCULATED FROM PAGE NUMBER

So the layout parameters are SEGFIL page numbers — the same quantity as madr in segment-facts.json.

[INFERRED] but exact on 28 of 32 parameters: take the leading additive terms in octal and discard the trailing -N (a length / in-page offset, not part of the address), then subtract the image base 0o200:

madr = (sum of leading additive terms, octal) - 0o200

Agreement¶

[VERIFIED] 28 exact matches against carved madr:

parameter	raw	page	madr	carved segment
`PCCS` common code/start	`200`	0o200	0	`S3IMAGE`
`PSY` system segment	`355-62`	0o355	0o155	`S3SSYS`
`PSPD` spooling datafields	`360-72`	0o360	0o160	`S3SSPD`
`PXP` ext common	`361-13`	0o361	0o161	`S3SECOM`
`PRP` / `PMP` / `PIP`	`363` / `446` / `531-15`		0o163 / 0o246 / 0o331	`S3SRPIT` / `S3SMPIT` / `S3SIPIT`
`P5P` / `P5M`	`200+414-13` / `200+421-20`		0o414 / 0o421	`S3S5PIT` / `S3SSM5`
`PSTB` segment table	`200+501`		0o501	`S3SSGT`
`PU1` / `PU2`	`200+521` / `200+561`		0o521 / 0o561	`S3SU110` / `S3SU120`
`PFS` / `PCS` / `PSM`	`200+621` / `706` / `773`		0o621 / 0o706 / 0o773	`S3SFS` / `S3SCP` / `S3SSM`
`PXK` / `PXR` / `PXF`	`200+1037` / `1067` / `1137`		0o1037 / 0o1067 / 0o1137	`S3SXMK` / `S3SXROU` / `S3SXMFI`
`PNM` / `PDM` / `PNS` / `PNN`	`200+1225` / `1233` / `1237` / `1323`			`S3SDNAM` / `S3SDMWD` / `S3SNKSE` / `S3SNKNA`
`PEC` / `PED` / `PPM` / `PEX` / `PBO` / `PMT`	`200+1423`…`1767`			`S3SERWC` / `S3SERWD` / `S3SPRMA` / `S3SEVMS` / `S3SBOPC` / `S3SMTSE`

Actionable findings¶

1. Segment 61 has a copy-pasted description — fix¶

[VERIFIED] segment-facts.json gives segments 60 and 61 the same description: "Save of XMSG kernel". The floppy separates them:

parameter	page	segment	correct meaning
`PXK=200+1037`	0o1037	`S3SXMK`	XMSG kernel
`PXR=200+1067`	0o1067	`S3SXROU`	XMSG XROUT

Segment 61's description should be "Save of XMSG XROUT", per the macro legend (PXRO → XROUT) in inputs/distribution-layout-params.txt.

2. Thirty medium-confidence segments can be promoted — update¶

[VERIFIED] 30 records carry confidence: "medium" solely because the OCR'd manual §8.3 name disagreed with the live LIST-SEGMENT name — e.g. manual #74 name=S3IDMWD against live S3SXMK, because the manual's rows are shifted by one in that region (EXTRACTING-SEGMENTS.md §8 documents the OCR damage).

The floppy is a third, independent, non-OCR witness and it backs the live names. Those 30 can go medium → high, citing this file.

3. A parity/OCR corruption in the repo's symbol table — fix¶

[VERIFIED] SINTRAN/NPL-SOURCE/SYMBOLS/L07/SYMBOL-1-LIST.SYMB.TXT line 6140 reads ENT0-011302 (hyphen). The floppy has ENT0=011302. This is the only differing line across all five comparable tables — the rest are byte-identical, which is itself a useful confirmation that the repo's symbol tables are sound.

4. An open conflict — flag, do not "fix"¶

[VERIFIED] PRD is documented in the floppy legend as "RESIDENT DATA PART ONE" and resolves to page 0o300 → madr 0o100. The carve names madr 0o100 S3SDPIT — "Save of DPIT".

These disagree. Do not rename anything on this basis. It may be that the resident-data area and the DPIT save area overlap, or that the -2 suffix in PRD=300-2 changes the interpretation. Record as [OPEN].

5. Parameters below the image base — unresolved¶

[VERIFIED] PPOP=1, PEP=100-14, PRL=137-14 sit below the 0o200 base, so the -0o200 rule does not apply. Their nearest carved neighbours S3SERRP (0o77) and S3SRTC (0o136) are off by exactly one page. [INFERRED] these address a bootstrap / error-program / RT-loader pre-image area with a different convention. Not resolved.

Not worth doing¶

Re-importing the symbol tables. Already in the repo byte-identical and already applied — 98,120 placements across 70 segments-ref bundles. Source breakdown: SYMBOL-1-LIST 80,284 · N500-SYMBOLS 7,988 · SYMBOL-2-LIST 3,947 · RTLO 2,809 · FILSYS 1,788 · XMSG 1,304.

The bigger prize, not yet attempted¶

The remaining ~1.08 MB of SINTRAN-L-1:DATA after the 7,457-byte header is the as-shipped, un-patched system in BPUN loadable form (first control byte at offset 7897; 273,595 of 1,095,538 bytes have the high bit set).

The carve is of an installed and site-patched system — it contains a segment S3PATCH at madr 0o4506. Decoding the 22 )9READ blocks and diffing them against the carved .bin files would separate what ND shipped from what this installation became, byte by byte.

Nothing in the current carve can do that. It needs a BPUN block decoder for the )9READ stream; the block format inside the stream is not yet verified, so a clean diff is not guaranteed.