diff --git a/disk/README.md b/disk/README.md index 43c633f6..8972082a 100644 --- a/disk/README.md +++ b/disk/README.md @@ -7,7 +7,27 @@ architectures' kernel QEMU boots — not `scripts/rundisk.sh`, which targets a separate, currently-unused `disks/` (plural) directory for the *hosted* VM's `--disk-img=` flag instead. -- `artemis.img` — standard Artemis persistence test disk. **Reformatted +- `artemis.img` — standard Artemis persistence test disk. **Rebuilt fresh at + 1 GiB, 2026-09-17** (FABRIC-3.md §XXXV.6): the prior 30 MiB image was + three orders of magnitude too small for the block-backed per-ISA DoE + persistence design (§XXXV.2 — measured 516 MB for just 8 trials of raw + per-tick data against a 30 MiB device). Rather than migrate the existing + image's top-of-device system-metadata fence to new geometry (fence + addressing is relative to `total_devblocks`, so a resize alone would shift + every existing fence consumer onto the wrong physical bytes), Captain Bob + ratified building a brand-new blank image and re-minting Zuse + all + thumbdrive identities fresh against it, since re-minting invalidates the + prior identities regardless of whether a migration is attempted. Blank + (all zero) at creation, same as every other fresh-format fixture below — + not yet formatted or re-minted; that is a separate, later live-boot step. + The superseded 30 MiB image is preserved, not deleted, as + `artemis-30mb-pre-1gib-backup.img` (below). +- `artemis-30mb-pre-1gib-backup.img` — the pre-2026-09-17 `artemis.img` + (30 MiB), preserved rather than deleted when `artemis.img` was rebuilt + fresh at 1 GiB (see above). Kept for reference / possible forensic + recovery of the prior Zuse/thumbdrive identity state; disposal is a + separate, still-open decision (FABRIC-3.md §XXXV.7), not made here. +- (superseded, kept for history) `artemis.img` was previously **reformatted 2026-08-02**: this image had been stuck in a corrupted state (valid `LithosAnanke` magic header, but data not matching what `ART-READ-TEST` expects) since before this repo's own git history @@ -19,7 +39,8 @@ VM's `--disk-img=` flag instead. file and letting a normal boot format+write-test it; verified `PASS: persist-read` on amd64, aarch64, and riscv64 against the same image afterward (cross-arch resume, matching the arch-neutral on-disk - format `.claude/ARTEMIS.md` specifies). + format `.claude/ARTEMIS.md` specifies). This history now applies to + `artemis-30mb-pre-1gib-backup.img`, not to the current `artemis.img`. - `artemis-debug-roundtrip.img` — round-trip regression fixture created during that investigation. Known-good: format → self-test → write-test → reboot → resume → `PASS: persist-read`, confirmed 3 times in a row. diff --git a/disk/artemis-30mb-pre-1gib-backup.img b/disk/artemis-30mb-pre-1gib-backup.img new file mode 100644 index 00000000..3a26b9b8 Binary files /dev/null and b/disk/artemis-30mb-pre-1gib-backup.img differ diff --git a/disk/artemis.img b/disk/artemis.img index 3f74c3ec..4fce05a4 100644 Binary files a/disk/artemis.img and b/disk/artemis.img differ