vms-4a6: resolve the main executable's own .vms$wimp at activation - #694
Merged
Merged
Conversation
resolve_weak_imports() iterated only g_prods[] (loaded producers); the main executable (g_exe) was never in it, so an executable's OWN weak-by-name imports were silently dropped -- an executable's resolvable weak import stayed 0 and its first call jumped through a null import-GOT cell to address 0 (the C++ first-light crash class, vms-70d/cc1: whole-archived libstdc++/libgcc weak-references a universal a producer exports). Factor the per-image .vms$wimp walk into resolve_wimp_for(base, wimp_addr) and run it for every producer AND for g_exe. Register the executable's mapped .vms$wimp (g_exe.wimp_addr) in activate_symbol_vector while the section headers are readable. Fail-honest is preserved unchanged: a weak import no loaded producer exports still resolves to 0. Strong (.vms$imp) binding is untouched. New gate run_exe_weak_import_activation.sh drives the symmetric case the producer test (run_weak_import_activation.sh) never covered: an executable whose own .vms$wimp names a universal reached only via a transitively-loaded producer. Present -> bound (exit 5), absent -> 0 (fail-honest, exit 0). Verified causally: reverting the imgact.c change makes the positive case return 0 (unbound). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This was referenced Aug 21, 2026
baron-3dl
added a commit
that referenced
this pull request
Aug 22, 2026
) Bump OVMX_PRODUCT_VERSION V0.5 -> V0.5-1. Point release packing the V0.5-x train: the NetBSD/VAX executive Files-11 ODS-2 ACP flip completing the flip on all three substrates (#690, vms-d9c green), Alpha authentic binary-SYSUAF login + standing green-by-SHA CI gate (#691/#692), C++ first-light 3-way-gated (#695), the executable-own .vms$wimp binding fix (#694), and the 1.0=hardening roadmap re-scope (#693). Release notes: docs/release-notes-0.5-1.md. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
baron-3dl
added a commit
that referenced
this pull request
Aug 22, 2026
* release: OVMX V0.5-1 — 3-substrate flip complete + C++ first-light Bump OVMX_PRODUCT_VERSION V0.5 -> V0.5-1. Point release packing the V0.5-x train: the NetBSD/VAX executive Files-11 ODS-2 ACP flip completing the flip on all three substrates (#690, vms-d9c green), Alpha authentic binary-SYSUAF login + standing green-by-SHA CI gate (#691/#692), C++ first-light 3-way-gated (#695), the executable-own .vms$wimp binding fix (#694), and the 1.0=hardening roadmap re-scope (#693). Release notes: docs/release-notes-0.5-1.md. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * roadmap: reconcile for V0.5-1 (regenerate doc + build json from rd/tags) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
resolve_weak_imports() iterated only loaded producers; the main executable (g_exe) was never in the set, so an executable's OWN weak-by-name imports were silently dropped → an executable's resolvable weak import stayed 0 and its first call jumped through a null GOT cell to 0x0.
Factor the per-image .vms$wimp walk into resolve_wimp_for(base, wimp_addr); run it for every producer AND for g_exe; register g_exe.wimp_addr in activate_symbol_vector. Fail-honest preserved (absent → 0); strong .vms$imp binding untouched. New causal gate run_exe_weak_import_activation.sh (wired into the weak-import-activate job): present→bound (exit 5), absent→0; reverting imgact.c makes the positive case return 0.
Note: this is NOT the cpptest keystone. The vms-4a6 investigation (descriptor dump + live gdb) found the original 'unbound weak malloc' premise was a misattribution — malloc is a STRONG import (.vms$imp[17], DECC$SHR) and its cell binds fine; cpptest's real crash is DECC$SHR/musl pthread-ABI (vms-ad70, GCC lane). This closes a separate, real executable-.vms$wimp binding gap that exe-weak-import binding needs.
🤖 Generated with Claude Code