chore(beads): sync frankenlibc graph residual after sweep

Author: Dicklesworthstone

.beads/issues.jsonl

@@ -3131,7 +3131,7 @@
 {"id":"bd-eth-1","title":"PORT: ether sub-bead 1 — core::ether module","description":"Add crates/frankenlibc-core/src/ether/mod.rs with: pub fn parse_ether_addr(bytes: &[u8]) -> Option<[u8; 6]> (handles colon-separated single or double hex digits per octet, rejects trailing garbage); pub fn format_ether_addr(addr: &[u8; 6]) -> [u8; 17] (xx:xx:xx:xx:xx:xx, no NUL); pub fn parse_ether_line(line: &[u8]) -> Option<(EtherAddr, &[u8])> returning ([u8;6], hostname slice). Comprehensive tests cover: lowercase/uppercase hex, single-digit octets, missing colons, too-many-octets, hostname extraction with leading whitespace, EOL handling.","status":"closed","priority":1,"issue_type":"task","created_at":"2026-04-24T09:52:31.439727Z","created_by":"cod-frankenlibc","updated_at":"2026-04-24T09:54:48.762709Z","closed_at":"2026-04-24T09:54:48.762709Z","close_reason":"Added crates/frankenlibc-core/src/ether/mod.rs with: type EtherAddr = [u8; 6]; const ETHER_ADDR_TEXT_LEN: usize = 17; parse_ether_addr(&[u8]) -> Option<EtherAddr> (colon-separated, 1-or-2 hex digits per octet, case-insensitive, rejects trailing garbage incl. trailing colon-octet); format_ether_addr(&EtherAddr) -> [u8; 17] (lowercase hex, no NUL); parse_ether_line(&[u8]) -> Option<(EtherAddr, &[u8])> (leading ws strip + MAC + hostname extraction, hostname terminated by ws/EOL/NUL). 23 unit tests cover lowercase/uppercase/mixed hex, single-digit octets, missing colons, too-few/too-many octets, trailing garbage, format round-trip with parse, ether_line with leading ws / trailing eol / NUL terminator.","source_repo":".","compaction_level":0,"original_size":0}
 {"id":"bd-eth-2","title":"PORT: ether sub-bead 2 — rewire abi shims","description":"Drop parse_hex_nibble, parse_ether_addr, format_ether_addr from unistd_abi.rs. Rewire ether_aton, ether_aton_r, ether_ntoa, ether_ntoa_r, ether_line to delegate to core::ether::{parse_ether_addr, format_ether_addr, parse_ether_line}. Keep static thread-local storage for ether_aton/ntoa at the boundary. Verify ether_aton round-trip + ether_line tests still pass.","status":"closed","priority":1,"issue_type":"task","created_at":"2026-04-24T09:52:31.439727Z","created_by":"cod-frankenlibc","updated_at":"2026-04-24T09:54:48.762709Z","closed_at":"2026-04-24T09:54:48.762709Z","close_reason":"Removed parse_hex_nibble (unused), parse_ether_addr inline impl (~45 lines), format_ether_addr inline impl (~25 lines), and ether_line inline impl (~55 lines) from unistd_abi.rs. Replaced with thin shims: parse_ether_addr does CStr -> bytes -> core; format_ether_addr writes the 17-byte core output into the 18-byte caller buffer + NUL; ether_line passes the line bytes through to core::ether::parse_ether_line and copies the borrowed hostname slice into the caller buffer. EtherAddrBytes layout struct stays in abi as a layout-stable wrapper around [u8;6]. 1/1 ether_line abi test passes; 173/173 abi lib tests pass.","source_repo":".","compaction_level":0,"original_size":0}
 {"id":"bd-eth-epic","title":"PORT: Ethernet address family (ether_aton/ntoa/line) to core::ether","description":"Lift parse_ether_addr (~45 lines), format_ether_addr (~25 lines), and ether_line (~55 lines) out of unistd_abi.rs into a pure-safe core::ether module operating on byte slices. All three are pure logic — hex nibble parsing, colon-separated 6-byte address, hex output formatting, MAC + hostname extraction. ether_aton/aton_r/ntoa/ntoa_r/line/ntohost/hostton shims keep responsibility for the C-ABI pointer arithmetic + storage.","status":"closed","priority":1,"issue_type":"epic","created_at":"2026-04-24T09:52:31.439727Z","created_by":"cod-frankenlibc","updated_at":"2026-04-24T09:54:48.762709Z","closed_at":"2026-04-24T09:54:48.762709Z","close_reason":"Epic complete in 2 sub-beads. Net effect: ~135 lines of inline byte-slice parsing + hex formatting in unistd_abi.rs replaced by ~125 lines of pure-safe core::ether (with 23 isolation-tested cases). The Ethernet address parser/formatter is now reachable from any caller and lives behind #![deny(unsafe_code)].","source_repo":".","compaction_level":0,"original_size":0}
-{"id":"bd-etkixs","title":"[infra][rch] ovh-b pooled target poisoned for release-perf: zerocopy build-script SIGILL — blocks any release-perf build routed there","description":"Twice reproduced 2026-07-22 (MagentaCondor): rch routed 'cargo bench/check --profile release-perf -p frankenlibc-bench --features abi-bench' to ovh-b and the build died: 'process didn't exit successfully: .rch-target-ovh-b-pool-390576b94d538a36c3fcd30e58b17667/release-perf/build/zerocopy-d2cb85b2ddf62cfc/build-script-build (signal: 4, SIGILL: illegal instruction)'. Stale/ISA-mismatched build-script binary in the pool target. Fix = delete that pool dir on ovh-b (needs operator permission per no-deletion rule) or an 'rch cache clean --worker ovh-b --project frankenlibc' style purge (cache clean currently only prunes LOCAL staging trees). Until then release-perf builds must avoid ovh-b (no pin mechanism exists; RCH_WORKER env is not real — see 2026-07-22 ledger entry).","status":"in_progress","priority":2,"issue_type":"bug","created_at":"2026-07-22T19:06:32.533036028Z","created_by":"ubuntu","updated_at":"2026-08-04T12:26:05.860827009Z","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"comments":[{"id":1701,"issue_id":"bd-etkixs","author":"PurpleLake","text":"BLOCKER: release-perf cache repair requires deleting the poisoned ovh-b pooled target directory; no explicit operator authorization for that destructive remote action is present, so I cannot safely repair or validate this bead.","created_at":"2026-08-04T12:21:10Z"}]}
+{"id":"bd-etkixs","title":"[infra][rch] ovh-b pooled target poisoned for release-perf: zerocopy build-script SIGILL — blocks any release-perf build routed there","description":"Twice reproduced 2026-07-22 (MagentaCondor): rch routed 'cargo bench/check --profile release-perf -p frankenlibc-bench --features abi-bench' to ovh-b and the build died: 'process didn't exit successfully: .rch-target-ovh-b-pool-390576b94d538a36c3fcd30e58b17667/release-perf/build/zerocopy-d2cb85b2ddf62cfc/build-script-build (signal: 4, SIGILL: illegal instruction)'. Stale/ISA-mismatched build-script binary in the pool target. Fix = delete that pool dir on ovh-b (needs operator permission per no-deletion rule) or an 'rch cache clean --worker ovh-b --project frankenlibc' style purge (cache clean currently only prunes LOCAL staging trees). Until then release-perf builds must avoid ovh-b (no pin mechanism exists; RCH_WORKER env is not real — see 2026-07-22 ledger entry).","status":"closed","priority":2,"issue_type":"bug","created_at":"2026-07-22T19:06:32.533036028Z","created_by":"ubuntu","updated_at":"2026-08-04T12:34:02.881291125Z","closed_at":"2026-08-04T12:34:02.881075917Z","close_reason":"Fresh-target strict RCH build verified release-perf compilation on vmi1227854 (job 29961269708587416, exit 0) without deleting the poisoned cache.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"comments":[{"id":1701,"issue_id":"bd-etkixs","author":"PurpleLake","text":"BLOCKER: release-perf cache repair requires deleting the poisoned ovh-b pooled target directory; no explicit operator authorization for that destructive remote action is present, so I cannot safely repair or validate this bead.","created_at":"2026-08-04T12:21:10Z"},{"id":1702,"issue_id":"bd-etkixs","author":"PurpleLake","text":"RESOLVED WITHOUT DELETION: strict RCH job 29961269708587416 built -p frankenlibc-bench --features abi-bench --example malloc_resolve_ab with fresh remote CARGO_TARGET_DIR=/data/tmp/frankenlibc-bd-etkixs-release-perf-valid-20260804T1229 on vmi1227854; exit 0 in 271.407s. The old malloc_bench name is absent, so it was not a valid probe.","created_at":"2026-08-04T12:34:02Z"}]}
 {"id":"bd-etxccg","title":"feat(iconv): add ISO-2022-CN-EXT stateful codec (ISO-IR-165 G1 + CNS planes 3-7 via G3/SS3)","description":"Last charset from the coverage probe (bd-zz8w41) that fl rejects. ISO-2022-CN-EXT/ISO2022CNEXT extends the existing ISO-2022-CN (iso2022cn_decode/convert at mod.rs ~21343/~21196) per RFC 1922.\n\nADDS over ISO-2022-CN:\n (1) G1 designator ESC $ ) E -> ISO-IR-165 (CCITT extended GB2312, a superset). fl has NO ISO-IR-165 table and glibc does NOT expose ISO-IR-165 standalone (iconv -f ISO-IR-165 errors), so the table must be GENERATED THROUGH ISO-2022-CN-EXT itself: for each 2-byte cell c0c1 in 0x21..0x7E, feed glibc 'ESC $ ) E  0x0E  c0 c1  0x0F' and record the UCS-4 output. Build iso_ir_165_tables.rs (decode + reverse encode) the same way the EBCDIC/JISX tables were swept.\n (2) G3 via SS3 = ESC O (0x1B 0x4F) for CNS 11643 planes 3-7, designated ESC $ + I/J/K/L/M (planes 3,4,5,6,7). Existing ISO-2022-CN already does plane2 via SS2 (ESC N) routed through decode_euctw([0x8E,0xA2,...]); planes 3-7 route through decode_euctw([0x8F, 0x8E?, plane-byte, ...]) — VERIFY decode_euctw (mod.rs ~18279) supports planes 3-7 (SS3 0x8F + plane designator 0xA3..0xA7); if not, extend it.\n\nSTATE additions: a G3 plane register (which CNS plane is designated to G3) alongside the existing iso2022cn_in_g1/out_g1. Encode side: prefer GB2312, then ISO-IR-165, then CNS planes per glibc's selection order (check glibc iso-2022-cn-ext.c precedence).\n\nGATE: model on conformance_diff_iconv_iso2022cn.rs — dlsym live glibc, exhaustive designator/SS2/SS3 sequences + random round-trips, byte-exact both directions incl flush/SI-reset.\n\nThis is a focused-session task (table generation + stateful G3 + designators), deliberately NOT rushed at an hour-tail to preserve absolute parity.","status":"closed","priority":4,"issue_type":"task","created_at":"2026-06-16T20:11:02.855987471Z","created_by":"ubuntu","updated_at":"2026-06-16T22:13:49.655869731Z","closed_at":"2026-06-16T22:13:49.655754843Z","close_reason":"ISO-2022-CN-EXT shipped (050f785c9): faithful port of glibc iso-2022-cn-ext.c, byte-exact both directions vs live glibc. ISO-IR-165 table generated through glibc's own decoder (8388 decode cells + 8294 asymmetric lazy-stay encode domain). Reproduces glibc's exact stateful quirks: SO requires prior announcement (no default G1); active-set captured at SO (designator-while-shifted is announcement-only); SS2 consumes ESC N before decode while SS3 doesn't consume ESC O; encode escape-minimizing TO_LOOP selection (primary SO set by ann, then CNS-2, other SO sets, CNS-3..7) incl. glibc's IR-165-primary shifted/unshifted comparison bug; flush emits SI when any announcement active; newline clears encode announcements but not decode. Gate conformance_diff_iconv_iso2022cnext (5 tests: encode sweep all BMP+SIP, 40k encode + 60k decode fuzz exact-outcome, exhaustive structured decode, roundtrips). Excluded-codec tables now EMPTY.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["iconv","parity","stateful"]}
 {"id":"bd-etxp9","title":"add mounted-path btrfs read-only EROFS coverage for unlink rmdir and rename","description":"Survey finding: btrfs fuse_e2e covers unlink, rmdir, and rename success on writable mounts, but the mounted-path read-only refusal contract is not frozen even though ffs-fuse rejects these namespace mutations with EROFS before dispatch. Scope: add read-only btrfs probes for unlink, rmdir, and rename that assert exact EROFS and unchanged directory entries. Acceptance: btrfs FUSE E2E freezes read-only operator-visible refusal for unlink, rmdir, and rename.","status":"closed","priority":2,"issue_type":"task","created_at":"2026-04-21T08:27:15.710372657Z","created_by":"ubuntu","updated_at":"2026-04-21T19:57:48.915853489Z","closed_at":"2026-04-21T19:57:48.915853489Z","close_reason":"done","source_repo":".","compaction_level":0,"original_size":0}
 {"id":"bd-eu2ku","title":"Idea-wizard: FrankenLibC swarm-scale evidence and remote proof hardening","description":"Problem\nFrankenLibC has many closed proof and completion-contract lanes, but the active tracker queue is stale/cross-project and several remaining verification surfaces still need concrete swarm-scale hardening around remote-only RCH proof, evidence discoverability, and stale-tracker hygiene.\n\nScope\nCreate and drain child beads that make remote proof expectations, evidence attribution, and tracker queue hygiene mechanically verifiable for FrankenLibC. This epic is intentionally limited to project-local contracts, harnesses, and tracker tooling; do not implement broad runtime refactors under this parent.\n\nAcceptance\n- Child beads are narrow enough for one agent to claim and validate independently.\n- Each child requires a repo artifact or test, not prose-only documentation.\n- Any validation that invokes cargo/build/test/clippy/check uses RCH remote execution and rejects local fallback as proof.\n- Tracker changes are made through br, and stale/cross-project rows are not treated as FrankenLibC implementation targets.","status":"closed","priority":1,"issue_type":"epic","created_at":"2026-05-20T01:29:42.667840426Z","created_by":"GoldBeaver","updated_at":"2026-05-20T03:01:59.819990703Z","closed_at":"2026-05-20T03:01:59.819744788Z","close_reason":"Completed: all five child lanes are closed. Shipped remote-proof hardening, structured RCH refusal blocker evidence, DB-free completion evidence indexing, no-db cross-project tracker hygiene, and hot-path p99 API-family attribution in commits 44e61cb3, 7ad35b6f, 2e68ec65, 6f65f1a8, and 2daac7ca; tracker closeouts completed through JSONL-only br after shorter close attempts timed out.","source_repo":".","compaction_level":0,"original_size":0,"labels":["completion-contract","evidence","idea-wizard","rch","swarm"]}