Update the r5.01 2ine documentation for the working pipeline: gspcl chains the C stages, the tools behave normally, and the fork's loader carries the 16-bit stack-unmap guard. Only tools/tms34010-sdk/2INE-501.md, tools/tms34010-sdk/OS2-501.md and the two bad-apple notes change.
Update the r5.01 2ine documentation for the working pipeline: `gspcl` chains the C stages, the tools behave normally, and the fork's loader carries the 16-bit stack-unmap guard. Only `tools/tms34010-sdk/2INE-501.md`, `tools/tms34010-sdk/OS2-501.md` and the two `bad-apple` notes change.
gspac and gspcg returning 139 was a real SIGSEGV, not a strange exit
code: they free their automatic data segment (DGROUP), which also holds
the running 16-bit stack, just before DosExit; lxFreeSegment unmapped it
and the 16->32 bridge faulted returning through it. Record the root
cause, the loader fix (PR #1, commit cad0c71), and that gspcl now chains
the whole pipeline. Keep the "check output files" rule for unpatched
loader builds.
The docs now describe the working r5.01 pipeline: gspcl chains the C
stages, every tool exits 0 (or 1 for a real error), and there is no
"check output files instead of exit codes" rule. The old 139/SIGSEGV
story stays in git history.
opencode.siil.arti.ee
changed title from docs(tms34010-sdk): what the 2ine 139 exit status was to docs(tms34010-sdk): the r5.01 2ine pipeline and gspcl2026-11-16 19:26:23 +02:00
Reviewed (base 17dfd24 → head 147bc84). Approved, with one merge-order note.
I reproduced the claims against a loader built from 2ine PR #1 (cad0c71), using the r5.01 tools from tools/tms34010-sdk/...Code Generation Tools (1 of 2) r5.01...zip:
gspcl -k -o2 -c HELLO.C → status 0, lowercase hello.obj + hello.asm — §7 is right, including the case transform and the -c/-z split.
base loader (3bcdfc3): gspcl → 1 after gspac only, and gspac itself → 139after writing a valid .if. So the behaviour this PR says is gone really is gone, and with it the old "exit 139 but still write valid files" caveat.
The one thing to be careful about is ordering: 2INE-501.md §1 says the fork's main carries the stack-unmap guard (cad0c71). Today main is 3bcdfc3; the guard lives only on PR #1. I am merging 2ine PR #1 first, so the sentence becomes true rather than aspirational.
Two nits, not blockers and both predating or adjacent to this diff:
§7 "chains the full C pipeline" — gspcl chains compile → optimize → codegen → assemble; it links only with -z and never runs gsprom. §7's own -c/-z note already says this; the table row at :25 could say so too.
§1 "this document's status update (3bcdfc3)" — 3bcdfc3 is the fork's TMS34010.md status update, not this document's. It reads as a cross-repo reference that isn't.
Reviewed (base `17dfd24` → head `147bc84`). **Approved**, with one merge-order note.
I reproduced the claims against a loader built from 2ine PR #1 (`cad0c71`), using the r5.01 tools from `tools/tms34010-sdk/...Code Generation Tools (1 of 2) r5.01...zip`:
* `gspcl -k -o2 -c HELLO.C` → status **0**, lowercase `hello.obj` + `hello.asm` — §7 is right, including the case transform and the `-c`/`-z` split.
* missing source / unopenable output → **1**; `-z` links and writes `hello.out`.
* base loader (`3bcdfc3`): `gspcl` → **1** after `gspac` only, and `gspac` itself → **139** *after* writing a valid `.if`. So the behaviour this PR says is gone really is gone, and with it the old "exit 139 but still write valid files" caveat.
The one thing to be careful about is ordering: `2INE-501.md` §1 says the fork's `main` carries the stack-unmap guard (`cad0c71`). Today `main` is `3bcdfc3`; the guard lives only on PR #1. I am merging 2ine PR #1 first, so the sentence becomes true rather than aspirational.
Two nits, not blockers and both predating or adjacent to this diff:
* §7 "chains the full C pipeline" — `gspcl` chains compile → optimize → codegen → assemble; it links only with `-z` and never runs `gsprom`. §7's own `-c`/`-z` note already says this; the table row at `:25` could say so too.
* §1 "this document's status update (`3bcdfc3`)" — `3bcdfc3` is the fork's `TMS34010.md` status update, not this document's. It reads as a cross-repo reference that isn't.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Update the r5.01 2ine documentation for the working pipeline:
gspclchains the C stages, the tools behave normally, and the fork's loader carries the 16-bit stack-unmap guard. Onlytools/tms34010-sdk/2INE-501.md,tools/tms34010-sdk/OS2-501.mdand the twobad-applenotes change.docs(tms34010-sdk): what the 2ine 139 exit status wasto docs(tms34010-sdk): the r5.01 2ine pipeline and gspclReviewed (base
17dfd24→ head147bc84). Approved, with one merge-order note.I reproduced the claims against a loader built from 2ine PR #1 (
cad0c71), using the r5.01 tools fromtools/tms34010-sdk/...Code Generation Tools (1 of 2) r5.01...zip:gspcl -k -o2 -c HELLO.C→ status 0, lowercasehello.obj+hello.asm— §7 is right, including the case transform and the-c/-zsplit.-zlinks and writeshello.out.3bcdfc3):gspcl→ 1 aftergspaconly, andgspacitself → 139 after writing a valid.if. So the behaviour this PR says is gone really is gone, and with it the old "exit 139 but still write valid files" caveat.The one thing to be careful about is ordering:
2INE-501.md§1 says the fork'smaincarries the stack-unmap guard (cad0c71). Todaymainis3bcdfc3; the guard lives only on PR #1. I am merging 2ine PR #1 first, so the sentence becomes true rather than aspirational.Two nits, not blockers and both predating or adjacent to this diff:
gspclchains compile → optimize → codegen → assemble; it links only with-zand never runsgsprom. §7's own-c/-znote already says this; the table row at:25could say so too.3bcdfc3)" —3bcdfc3is the fork'sTMS34010.mdstatus update, not this document's. It reads as a cross-repo reference that isn't.