include/starkernel/session.h: vm_id (VMUuid), pinned (int, authoritative
over Stadium's STADIUM_FLAG_PIN per H.10), parent (VMUuid), name (fixed
64-byte buffer), identity (embedded VMIdentity, reusing the existing
type rather than inventing a new one). No logic yet, no callers -- next
step wires the session-slot array and register/find functions.
Verified 3-arch boot to ok> (amd64/aarch64/riscv64).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
One coding task per item across 7 phases, each gated by the mandatory
3-arch QEMU boot acceptance test. Two corrections found while grounding
this against live code: capsule_birth.c's generic admission path already
admits every VM as a Stadium patron (just hardcoded unpinned), so no new
stadium_birth_hermes()/_artemis() functions are needed -- the real task
is making that path pin Hera/Hermes/Artemis specifically. VMIdentity
(vm_identity.h) already exists fully built; Session.identity reuses it
directly rather than inventing a new type.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Gap #1: closed set of four cards (VM/word/block/message), no new cards
for now. Gap #2: creator-ceiling enforcement is a birth-time snapshot
copy, not a live ACL-RECHECK-time check -- reversed mid-conversation once
the live-check approach's cross-VM dictionary-lookup problem surfaced;
final rationale is child program stability (a program developed against
one ACL set must not have it silently changed by later parent changes).
Gap #3: Zuse's eligibility list persists in the existing growable
metadata-fence mechanism as a simple owner_pubkey[32] list, no extra
per-entry metadata.
Every design-level question in this refactor is now closed. What remains
is pure implementation (5 items) and two intentional deferrals.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Full sweep across the whole session-refactor design pass -- 7 of the
original 10 §H.9 gaps are now closed. Compiles the current accurate
picture: one genuinely open design question (more cards beyond the four
named?), two small undecided pieces (creator-ceiling enforcement location,
Zuse eligibility-list storage), five pure implementation gaps (design
fully decided, nothing coded), and two explicitly-deferred non-goals
(VM card multi-owner, elevation trigger alternates). Also updates H.10's
stale "elevation trigger not yet usable" note now that H.7 confirmed the
messaging substrate is already real.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
No actual conflict between D.2 and H.1. D.2 (2026-08-27, "session ending =
VM detach, reuse COOL") was scoped to user sessions, written before
Hera/Hermes/Artemis were framed as having their own persistent sessions.
D.2 is a special case of H.1's broader pinned/non-pinned model, fully
subsumed: it describes exactly how the non-pinned branch behaves. Pinned
sessions have no D.2-style "ending" at all -- permanent by construction.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Checked MSG-SEND directly -- no ACL check anywhere in it today, on common
or a private topic alike. Confirmed: private-topic traffic stays ungated;
trust is established once at the handshake (CH-REQUEST/CH-ACCEPT/
CH-CONFIRM) and never re-checked per-message afterward. The message
card's scope is now final: it gates the single CH-REQUEST call, nothing
else in the messaging path.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Checked capsules/common/messaging.4th directly instead of assuming: the
H.7 messaging protocol isn't design vision, it's already substantially
built -- COMMON-CH, CH-REQUEST, CH-ACCEPT, CH-CONFIRM, CH-CLOSE, and
local ACK/NACK (MSG-ACK-LAST/MSG-NACK-LAST/MSG-NACKED) all exist and work
today. CH already is the topic, 1:1, no new representation needed.
Upgrades H.7/H.8's framing from "protocol groundwork, not yet real" to
"the mechanism exists, the ACL gate on top of it is the remaining work" --
the message card is now a concrete wiring task, not a hypothetical one.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Checked capsules/ACL.4th directly: both existing modes (STRICT, TTL)
already default to allow=1, matching the word card's default-permissive
baseline. Since each VM has its own dictionary, existing per-word ACL
state is already scoped per-session for free. Confirmed: the word card is
the existing acl_ttl/acl_allow/acl_mode/acl_pinned system reused as-is,
not a new parallel structure. Surfaces one new follow-on gap: creator-
ceiling enforcement at VM birth time isn't designed anywhere yet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Both fields (parent VMUuid, name) confirmed as-is with no adjustment. All
five session fields (vm_id, pinned, parent, name, identity) are now
confirmed, none still model-proposed-only.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Confirmed: an ephemeral user VM's session lands as a non-pinned,
COOL-subject Stadium patron -- synthesized from D.2's earlier finding
(session ending = VM detach, reuse COOL) rather than a fresh decision.
Thumbdrive detach is the explicit trigger. No exception case for a
temporarily-pinned user VM.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Decided: yes, FORTH wrappers for the block-card ACL primitives, mirroring
the word-level ACL split (raw C accessors, policy composed in FORTH).
zuse_cert_seed's C-only precedent doesn't apply -- that's key material,
block ACL fields are ordinary ACL state like DictEntry's. BMAPFMT's
field/API design (owner_fp, acl_allow, acl_ttl, flags bits, FORTH
wrappers) is now fully decided across this and the prior three passes --
only the actual code edit remains.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
blk_meta_t's existing unused flags field (uint64_t) uses orthogonal bits,
not a mutually-exclusive enum: bit 0 = CLAIMED, bit 1 = MIGRATING (serves
MIGSM), bit 2 = STALE (serves UNCLEAN), 61 bits reserved. BMAPFMT's field
design (owner_fp, acl_allow, acl_ttl, and now these flags bits) is fully
decided; only the actual code edit to block_subsystem.h/.c remains.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Resolves part of section F.4's deferred item #1 (allow-list/grant shape
beyond the single fast-deny bit): blk_meta_t's acl_reserved[7] becomes a
4-byte acl_ttl (mirroring DictEntry's countdown shape) plus 3 bytes still-
open slack, reusing the same ACL-TTL/Zuse-eligibility-list mechanism as
the word card. Kept lean -- no acl_mode/acl_pinned mirror, since a block
isn't pinned/strict the way a word is. Updated both F.4 (the field design)
and H.6 (the block card) to match.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Decided: session_set_pinned()/session_is_pinned() (or equivalent) are the
sole read AND write path for pin state -- nothing, including existing
Stadium code, touches STADIUM_FLAG_PIN on the patron header directly
anymore. Not just a write-side guard.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Checked the load-bearing assumption underneath the word card's
creator-ceiling invariant and the H.5 elevation trigger, both of which
live on DictEntry ACL fields -- confirmed each VM gets its own separate
memory/dictionary buffer at birth (vm_bootstrap.c:181), so DictEntry ACL
state is already naturally scoped per-session. No conflict, no redesign.
Also notes two smaller, lower-risk open items: pin-authority sync between
Session and Stadium's flag, and the elevation trigger's dependency on
not-yet-real Hermes messaging.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Trigger: live message-based request (ELEVATE-REQUEST on common or a
private topic, Zuse's session grants/NACKs, ACL-ALLOW!/ACL-TTL! write on
receipt), gated by a Zuse-held eligibility list keyed by owner_pubkey.
Grounded against capsules/ACL.4th directly -- ACL-TTL is a recheck-cadence
cache, not a grant mechanism; reuse means calling its ACL-ALLOW!/ACL-TTL!
primitives from new Zuse-triggered code, not new C primitives. Live-console
sudo-style grant and pre-signed capability tickets remain deferred, not
rejected, alternates.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Design capture from today's session-refactor discussion, back to Hera:
session-as-Stadium-patron struct shape (references the patron by VMUuid,
pin-authoritative, identity embedded), and the identity ACL "stack of
cards" model with all four named dimensions scoped so far (VM, word,
block, message) plus the PubSub/topics messaging-protocol groundwork the
message card depends on. Closes with a numbered gap-analysis list (§H.9)
of everything still explicitly open.
Capture-only, matching this document's own established discipline -- no
struct written, no code changed. Mirrors §D's Tripod-vision capture in
style and structure.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
fb_scroll_rows() hardcoded the pixel distance it physically shifts the
framebuffer by as char_rows * 16 * scale -- the bitmap-font (font_8x16.c)
cell height -- regardless of which glyph mode vt100.c actually had active.
In TTF mode (the REPL's default, cell height 24px via VT100_TTF_CELL_H_PX)
this meant every scroll_up(1) call physically shifted the framebuffer by
only 16px while the text model (g_vt.rows, py_of()) placed each row 24px
apart. That 8px-per-scroll shortfall compounds with every subsequent
scroll: a few scrolls barely show it, but enough scrolls -- or scrolling
quickly, which is just many scrolls in a short span -- accumulates into
visible pixel overlap between rows, with newer lines drawn on top of the
tail end of older ones.
fb_scroll_rect() (the box-confined scroll added later for 4.4t) already
carried a doc comment calling this out explicitly, describing its own
explicit pixel_rows parameter as the fix for fb_scroll_rows()'s "fixed
16px-row assumption" -- fb_scroll_rows() itself was just never updated to
match.
Fixed by changing fb_scroll_rows()'s parameter from an implicit char_rows
count to an explicit pixel_rows count (matching fb_scroll_rect()'s
existing convention), and having its one caller (vt100.c's scroll_up())
pass lines * cell_h() -- the real active cell height -- instead of a raw
line count for the callee to guess at.
Verified: booted amd64 to the REPL (TTF mode active per sk_repl()'s own
console_fb_enable_ttf() call), let boot chatter + WORDS output scroll the
screen through thousands of accumulated scroll_up() calls, then measured
every visible line's y-position via a QMP screendump. Spacing held at a
perfectly consistent 24px (TTF cell height) top to bottom with zero drift
-- the old hardcoded-16px bug could not have produced that after this many
scrolls. Re-verified boot to ok> on all three architectures
(amd64/aarch64/riscv64) per repo acceptance policy.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
sk_repl_idle() (called every ~1s from sk_console_readline()'s idle loop)
opens with console_ensure_line_start(), which unconditionally forces a
newline whenever the console isn't at a line boundary -- including
mid-edit, after characters have been typed and echoed but before Enter.
This fired on every elapsed SK_IDLE_BEAT_INTERVAL regardless of whether
sk_repl_idle() had anything to print, visually snapping the in-progress
input line to a fresh empty line -- indistinguishable from Enter having
been pressed. Most noticeable on the space key since it's the most common
key hit during a pause.
Gate the idle beat on n == 0 (no in-progress edit), mirroring the n > 0
guard the prompt reanchor logic just below already uses. Deferring the
xhci/block-sync idle service by at most one more interval while a line
is being edited is within its own documented "coarse cadence, cheap
early-exit" tolerance.
Verified: reproduced via QMP send-key against a live amd64 QEMU boot
(multi-character line typed with pauses across several idle intervals
stayed intact after the fix, where it previously broke on each interval).
Re-verified boot to ok> on all three architectures (amd64/aarch64/riscv64)
per repo acceptance policy.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Clarified 2026-08-29 with the author: "no hurry" is wrong for #1 and #2.
- #1 (Hosted StarForth): the provisional will expire before conversion. Its
priority claim is time-boxed (provisional filed ~Dec 2025, window ~Dec
2026); if the non-provisional isn't filed claiming benefit before the
window lapses the priority is lost and the same subject matter can't be
re-staked by refiling thereafter. Hard clock regardless of December scope.
- #2 (full StarshipOS): the December deliverable that keeps the flagship
covered; pending counsel confirmation it may also be the conversion
vehicle for #1's provisional.
Open point to resolve: is #2 (or the December delivery) the conversion vehicle
for #1, or is #1 converted by a filing separate from #2? Decision record only;
no filing made. Authoritative form in ROADMAP.md; FABRIC-3.md tracks it.
The three-product split is treated as three patent applications, each matched
to a trademark (decided 2026-08-29):
- Patent 1 + mark StarForth -> hosted/interpreted runtime
- Patent 2 + marks StarshipOS (canonical spelling; noted as "StarshopOS"
in the decision words) + LithosAnanke -> full standalone OS
- Patent 3 + mark Compudynamics -> Zynq steady-state machinery
with sealed executions, HOL-proven, anchored by the physics-adaptive runtime
Flagged open for counsel (not assumed/filed): the repo already carries a
"Patent pending" USPTO provisional filed Dec 2025 for the Compudynamics
physics-adaptive runtime (docs/patent/). Whether patent 3 is a continuation/
refinement of that provisional, and whether the three product applications are
additive to or fold it in, must be reconciled. Records intent only; no legal
text drafted or filed. Authoritative form in ROADMAP.md; FABRIC-3.md tracks it.
The Zynq FPGA is not a fourth platform to port to — it is the pivot that
forces the project into three distinct products, each with its own host,
delivery, and proof character (decided 2026-08-29):
1. Hosted StarForth - the existing hosted/interpreted StarForth runtime
(3-arch acceptance-tested), delivered as a portable embedded runtime.
2. A full StarshipOS - the standalone OS built on LithosAnanke
(LithosAnanke -> StarshipOS), a self-booting OS on general silicon
(SER5/RasPi/Milk-V line).
3. Hardware steady-state machinery with sealed executions, HOL-proven -
the FPGA-native product: hardware-enforced sealed executions and
steady-state machinery machine-checked in a proof assistant (HOL);
delivery is the bitstream + HOL proof artifacts, not just an OS port.
The coupling is the point: product 3 is born on the FPGA, and its existence
is what cleanly separates 1 from 2 from 3. Product 1 ships hosted on an OS,
product 2 as an OS on general silicon, product 3 as proven hardware. Per-
product gates and the even-major line carrying the three as separate tracks
are scoped at v2.5.0 close / during the coloring-in phase. Complements, and
does not retract, the existing hardware bare-metal release policy.
Authoritative form in ROADMAP.md; FABRIC-3.md tracks the same horizon.
The next big milestone after the three-board bare-metal cut (v2.5.0) is
transferring the battle-tested amd64/aarch64/riscv64 story to a Zynq (AMD
Xilinx) FPGA SoC — configurable silicon with soft/hard CPU cores, PL fabric,
and a non-standard memory map, a genuinely larger step than any prior board
(expected on a new even-major line). Between v2.5.0 and the Zynq sits a
"coloring in" period: hardening that thickens the shape of what exists rather
than adding silicon (USB BOT/xHCI + block robustness, live-entropy and
Zuse-cert hardening on real ASICs, SMP/multi-core + IRQ routing from the HAL
notes, driver breadth) so the FPGA carries a production-honest shape forward.
Authoritative form in ROADMAP.md "Beyond v2.5.0"; FABRIC-3.md tracks the same
horizon in the post-release section and re-points G.6's Next at the
v2.2.0/v2.4.0/v2.5.0 cadence.
Real silicon arrives incrementally (Beelink SER5 in hand now; RasPi 5 + Milk-V
orderable around Mon 2026-08-31), so the hardware release is split per board
in hand, each an even-minor LTS point-in-time cut on the same line:
- v2.2.0 amd64 bare metal — Beelink SER5 (two 16GB sticks: one boots the
thumbdrive image, one mints a real Zuse user)
- v2.4.0 aarch64 bare metal — Raspberry Pi 5
- v2.5.0 all three bare — adds Milk-V (riscv64)
Each closes only when its board's items are proven live (boot to ok>, block-
fence Zuse true, backend entropy non-deterministic), never build-only.
Records the decision in the authoritative Release Versioning Policy
(docs/lithosananke/ROADMAP.md, "Board-by-board hardware rollout, decided
2026-08-29") and re-maps G.4/G.5 in FABRIC-3.md onto the new cadence, noting
v2.0.0 was cut (tag v2.0.0).
First real per-arch RNG backend, added to the v2.0.0 unified entry point in
src/starkernel/rng/rng.c, #if-guarded to amd64: CPUID.01H:ECX[30] RDRAND
detection + inline-asm rdrand draws feeding rdrand_fill() (whole-byte
emission from the low end; a partial final draw is discarded -- throwing
away entropy is always safe).
Probe order honors the release policy: virtio-rng is tried first, so the
QEMU path stays on virtio-rng unchanged; RDRAND is the fallback only real
hardware (which has no virtio-rng device) reaches. QEMU-verified both ways
on amd64: with virtio-rng present -> "rng: backend = virtio-rng" (unchanged);
with virtio-rng absent and RDRAND exposed (-cpu max) -> "rng: backend =
rdrand" + "entropy: ready" + Zuse attach confirmed. rdrand_fill()'s exact
logic host-proven: fills 32-byte/16-byte buffers and yields differing draws
run-to-run (non-deterministic). aarch64/riscv64 builds unaffected (guarded
off). riscv64 Zkr and aarch64 peripheral-RNG backends remain parked for their
real boards.
FABRIC-3.md G.4 amd64 slice marked BUILT + QEMU-verified.
New `make -f Makefile.starkernel thumbdrive` goal builds a generic
UEFI-bootable GPT disk image (disk image -> GPT + one FAT32 "STARKERNEL"
EFI System partition with EFI/BOOT/BOOT<ARCH>.EFI + startup.nsh) that can be
written directly to a USB thumbdrive with dd and boots on any real amd64 UEFI
firmware (Beelink SER5 path) as well as under QEMU. The monolithic loader
embeds the whole kernel, so the ESP needs only the UEFI fallback boot path.
Verified under QEMU by attaching the image as a USB mass-storage device (not
cdrom): OVMF BDS auto-selected "UEFI QEMU QEMU USB HARDDRIVE" (Boot0002),
then kernel booted normally -- LithosAnanke v2.0.0, POST 1012/0/0 + ok>,
rng: backend = virtio-rng + entropy: ready, Artemis ready, and the USB BOT/
xHCI storage path enumerated (READ CAPACITY10 -> MSC device ready -> READ10
CSW PASS). This mirrors the real-hardware SER5 flow: firmware boots the USB
thumbdrive's ESP, and the OS's own storage rides the same USB BOT/xHCI path.
Works for all three arches via the arch-mapped EFI boot name.
The §G v2.0.0 QEMU release-gate punch list is complete (G.1 xHCI stall
recovery, G.2 unified rng_get_bytes() entropy entry point, G.3 NVRAM
de-scoped). Bump the kernel version to 2.0.0 (even major = LTS, per the
Release Versioning Policy: X.0.0 = QEMU release, X.5.0 = hardware bare-metal
release). Update README and .claude/CLAUDE.md version references and the
Makefile.starkernel version-roadmap comment.
Verified: all three arches (amd64/aarch64/riscv64) build clean with v2.0.0
embedded; QEMU amd64 boot shows "LithosAnanke v2.0.0", POST 1012/0/0 + ok>,
rng: backend = virtio-rng + entropy: ready, and Zuse identity confirmed from
the attached thumbdrive.
The QEMU-verifiable slice of the real-hardware RNG driver (per FABRIC-3.md
§G.2). New include/starkernel/rng.h + src/starkernel/rng/rng.c provide the
single entropy entry point: rng_init() probes the backend set (v2.0.0:
virtio-rng only) and, on no backend, prints a loud boot-time warning while
rng_get_bytes() returns RNG_ERR_NO_BACKEND - never silently degrading to a
deterministic seed. The backend-selection switch in rng.c is the exact seam
v2.5.0's per-arch drivers (amd64 RDRAND, riscv64 Zkr, aarch64 peripheral) plug
into without touching the call path.
Consumers route through the unified layer instead of virtio-rng directly:
capsule_mint.c (identity seed + drive_uuid) and kernel_main.c phase 8
(rng_init()). virtio_rng.c stays as the sole backend. Built clean on
amd64/aarch64/riscv64. QEMU amd64 boot: POST 1012/0/0 + ok>, "rng: backend =
virtio-rng" + "entropy: ready", Zuse identity confirmed from thumbdrive -
mint/cert behavior unchanged.
FABRIC-3.md §G.2 v2.0.0 slice marked BUILT+VERIFIED.
Root cause of the G.1 follow-up boot-time attach race: on pathological
controller behavior the xhci_poll_events() drain loop had no hard ceiling.
ERDP is written back only when the loop exits, so the controller cannot
reclaim event TRBs mid-drain; if it keeps producing events the head can
chase the software dequeue pointer forever. xhci_poll_events() never returns,
sk_repl_idle() never reaches its bot_msc_attach_pending check, and a fresh
USB BOT device that finished SET_CONFIGURATION is left flagged-but-never-
attached while the guest appears hung.
Fix: bound the drain to a full ring (XHCI_EVT_RING_MAX_DRAIN = 256), so
xhci_poll_events() always terminates and always writes ERDP each call.
Unprocessed events keep their cycle bit and are re-read next poll; nothing
is dropped. On the healthy path one drain processes only the one-or-few
events the controller posts per chained command, so the bound never triggers
except in the pathological case it breaks.
Beyond the G.1 additions: a new macro in include/starkernel/xhci.h and a
bounded loop in src/starkernel/usb/xhci.c. Builds clean on amd64. Verified
across six consecutive fresh QEMU boots (previously intermittently hung).
Full BOT-spec stall recovery per FABRIC-3.md F.14: new STALL_ERROR handling,
Reset Endpoint + Set TR Dequeue Pointer commands, CLEAR_FEATURE(ENDPOINT_HALT),
escalating to Bulk-Only Mass Storage Reset, capped retries
(XHCI_BOT_STALL_MAX_RECOVERIES=2) mirroring bot_tur_retries, clean terminal
failure via xhci_stall_fail().
Purely additive recovery path off the non-success transfer-event branch; the
normal path is unchanged. Builds clean on amd64/aarch64/riscv64. QEMU amd64
boot regression passes: zero stalls, BOT attach (READ CAPACITY10 -> READ10 ->
home-blocks) completes, normal-path xHCI trace identical to baseline. Live
stall injection is not provable under qemu-xhci; deferred to v2.5.0 hardware.
FABRIC-3.md G.1 documented; ROADMAP release-versioning policy folded in.
Code review fixes, all compile clean (hosted gcc + aarch64/riscv64 kernel flags):
- repl.c (H1): reentrancy guards on the MSG-TICK idle pump. sk_repl_idle()
now defers when Hera is mid-interpret (g_mama_interpreting) or when its
own vm_interpret is on the stack (g_idle_pump_active), so a blocking
KEY/EXPECT/QUERY inside a dispatched line can no longer re-enter the
interpreter and clobber the in-flight input buffer.
- virtio_rng.c: clamp device-returned used_len to VRNG_BUF_SIZE before the
caller's data_buf copy, closing a device-controlled OOB read.
- block_subsystem.c: first-write path now keys off created_time==0 instead
of dead magic==0 so fresh blocks get a real created_time stamp; first_free/
last_allocated fixed to absolute Forth LBNs (set in blk_compute_fresh_geometry
from slot->start_lbn, no longer the wrong physical-BAM-index values from
compute_totals_from_B); physical-bounds guard on blk_meta_zone_read/write
prevents unsigned underflow on a corrupt fence >= device size.
- capsule_zuse_boot.c / capsule_wirebind.c: identity seed validated magic ->
version -> CRC-64 (compute_crc64 over offsetof(crc)) before trusting it,
so a corrupt/format-mismatched record is refused, never loaded.
- log.h / starkernel/log.h: unused LOG_LINE_MAX 256 renamed LOG_MSG_LINE_MAX
to lift the include-order collision with vm.h's LOG_LINE_MAX 64; stale
include-order comments dropped (kernel_main.c, shim.c, capsule_birth.c).
- FABRIC-3.md: three stale-doc carry-forward items closed [x] with cbe7b49
notes.
Real KEY/?TERMINAL/QUERY/EXPECT bodies (console WIP):
- repl.h/repl.c: sk_console_getkey()/sk_console_key_available()/
sk_console_readline() public bodies; non-destructive peek buffers the
found byte so a following KEY returns it.
- shim.c: getchar()/fgetc()/fgets()/sf_terminal_ready() routed through the
real console paths instead of stubs; sf_terminal_ready() in platform_io.h
with sf_terminal_ready() implemented for the hosted build (linux/io.c,
POSIX select on fd 0) wired into Makefile.
- io_words.c: ?TERMINAL now returns actual terminal-readiness, not constant 0.
Artifacts: minted disk/artemis.img + rebuilt lfs kernel; BLOCK_MAP.md,
doe csv + qemu log regenerated.
capsule_mint_identity()'s MINT_DEFAULT_PERSONALITY only ever defined
WELCOME, unlike capsule_console.c's own CONSOLE_IDENTITY_SRC. The moment
WIREBIND made a freshly-minted identity's VM live, Hera's per-idle-tick
MSG-TICK pump (repl.c Phase C) began erroring on it every tick forever --
UNKNOWN WORD: 'MSG-TICK'. Every identity minted before this fix would hit
the same infinite error-spam on going live, not just this one.
Fixed by loading common:messaging.4th + calling MSG-CD-INIT before
WELCOME, mirroring capsule_console.c's own pattern exactly.
Found and verified minting the first real second identity end to end:
disk/captain-bob.img, hot-attached via QMP alongside an already-
authenticated Zuse session, MINTed, WIREBIND-attached, switched into
live with USE, WELCOME confirmed. Zero MSG-TICK errors over 18+ seconds
of idle-loop cycling after the fix (was immediate and continuous before).
Full three-architecture regression clean.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KBjfeLPo71sUQ8zC7V7P5m
Documents the mismatched-marker/blank-drive root cause, the LOG_INFO-vs-
LOG_WARN filtering trap that made the debug probes look like they weren't
firing, the three-architecture acceptance results, and the answered "why
not just reboot after MINT" design question (both branches already share
install_and_activate(), so there's no duplicate path to reconcile).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KBjfeLPo71sUQ8zC7V7P5m
xhci_bringup() now scans for already-connected ports at bring-up
(xhci_scan_ports_for_already_connected()), not just later hotplug events,
so a USB device present on the QEMU command line at launch is detected.
Makefile.starkernel attaches disk/zuse.img on the xhci0 bus by default in
all three arch qemu targets (ZUSEDISK=, empties for a bare boot).
capsule_mint_identity() gained a drive_known_blank param to skip a fully
redundant second homeblocks_sig_check() when the caller already confirmed
HOMEBLOCKS_SIG_BLANK itself.
Root-caused what looked like a hang after the drive attached: Artemis's
fence still carried a genesis marker from before a mid-session reformat,
while the reformatted disk/zuse.img read back BLANK -- a mismatched pair
capsule_zuse_boot_try_attach() correctly declined to act on, leaving the
boot idling at a plain ok> with nothing left to log (indistinguishable
from a hang under slow TCG). Fixed by zeroing both disk/artemis.img and
disk/zuse.img at their original sizes, giving a matched blank pair.
Verified full three-architecture acceptance: amd64 fresh genesis-mint,
aarch64/riscv64 clean reload against the same now-minted images.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KBjfeLPo71sUQ8zC7V7P5m
Two items, closed together per direct instruction.
1. Fence-persistence root cause, found and fixed: meta_fence_blocks
(the field gating whether blk_meta_zone_write() can succeed at all)
was carved out of what used to be unused padding in blk_volume_meta_t
-- the code's own comment already documented this. disk/artemis.img
was formatted before that field existed, so its on-disk bytes there
have always read back as 0, and the existing-volume load path
(blk_format_or_load_disk()) never recomputes it -- only a fresh
format does. Every "fence write FAILED" message this entire session,
old block-fence flow and new zuse_genesis_marker_t alike, traces to
this one thing. Patching the field in place without redoing the rest
of the geometry would risk corrupting whatever's already allocated
near the top of the volume, so the only safe fix is a genuine
reformat -- done, with explicit confirmation, since it discards
disk/artemis.img's accumulated persistent test state (regenerated
fresh at next boot regardless, not real data). Verified: fence write
now succeeds with no failure suffix, and the full mint-once ->
reboot -> reattach -> re-authenticate cycle works for the first time
this session ("Zuse: identity confirmed from attached thumbdrive",
ZUSE-SESSION? goes 0 -> -1 without re-minting).
2. BINDSTEP (FABRIC-3.md §F.9): capsule_wirebind_verify_cert() extracted
as a shared function so WIREBIND (the original attach) and BINDSTEP
(every USE of an identity-locked VM) check the exact same thing the
exact same way. mama_word_use() now re-verifies live, not cached,
whenever the target VM has VMIdentity.installed=1 -- reads whatever
drive is CURRENTLY attached, re-verifies its cert, compares owner
pubkey against the target's own installed identity, refuses on any
mismatch or no drive attached. A target with installed=0 (Hera,
Hermes, Artemis, any console VM) stays freely targetable, unchanged.
Two related bugs found and fixed live while testing BINDSTEP, not
assumed away: USE was Mama-only, so a console-paired session (§F.22)
had no way back to Hera at all -- any attempt to call USE from inside
a console VM hit "UNKNOWN WORD: USE", a genuine dead end. Per direct
instruction, USE isn't console-specific -- it should work VM-to-VM
universally, same as VM-EXEC already does -- so it's now registered in
register_child_vm_words() too. That alone wasn't enough: the console
relay (sk_repl_dispatch_line()) would have captured a bare USE call and
sent it to the paired user VM as a message instead of running it.
Fixed with a small suffix-match guard (sk_repl_line_calls_use()) --
real FORTH syntax always puts USE last, so a trailing-token check
reliably recognizes it without needing a full tokenizer, and it always
runs directly, never relayed.
Verified live end to end: USE on an unlocked VM works unconditionally;
USE escaping a console back to Hera now works; USE on an identity-
locked VM succeeds while its own drive is attached and is refused
once detached ("USE: FinT~user refused -- no matching identity
currently attached"). Clean 3-architecture regression, including
confirming disk/artemis.img's reformatted geometry loads correctly as
an already-recognized volume ("Artemis: LithosAnanke disk -- resuming")
on aarch64 and riscv64 too, not just the amd64 boot it was reformatted
under.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ZGkimpfyh63EZyRkNbkPD
Assembles pieces already built and individually verified this session
-- CERTVERIFY (vm_identity_from_cert(), Phase A/B), RUNCAP, the
console-VM + user-VM pair (§F.22) -- into one automatic sequence,
replacing the RUNCAP-TEST/PAIR-TEST diagnostic words that exercised
each piece by hand.
New capsule_wirebind_try_attach() (capsule_wirebind.h/.c), called from
sk_repl_idle() alongside capsule_zuse_boot_try_attach() on every
HOMEBLOCKS_SIG_OK attach: sig->cert_offset==0 means this is Zuse's own
genesis-mode drive (no cert region) -- that's already
capsule_zuse_boot_try_attach()'s job, skip. Otherwise, with Zuse already
authenticated this boot (nothing to verify a regular cert against
otherwise), reads the cert devblock(s) and calls vm_identity_from_cert()
against mama_vm's own zuse_cert_pubkey and the drive's own drive_uuid.
On success: reads the drive's own user_identity_seed_t for its
username, births a console VM + RUNCAP-born user VM pair (idempotent --
no-ops if that username is already live this session), installs the
verified VMIdentity onto the user VM, and registers the "<username>~user"
pairing sk_repl_dispatch_line() (repl.c, §F.22) looks for. Deliberately
does NOT auto-USE the new console -- that stays an explicit,
ACL-gated step (BINDSTEP, §F.9), not something a bare attach should
trigger silently.
Verified end-to-end live in QEMU, including a genuine negative case:
attached disk/user1.img (signed by a different, earlier-session Zuse
instance) and got a correct "cert verification FAILED -- drive
refused" -- proof the check is real, not a rubber stamp. Minted a
fresh identity with this boot's own Zuse, reattached, and got
"WIREBIND: SamS attached and ready" printed with zero manual commands,
followed by a working USE + async WELCOME relay end to end (queued,
no UNKNOWN WORD, delivered and executed in the paired user VM on the
next idle tick). Clean 3-architecture regression: Hermes/Artemis both
birth live, no unexpected ACL denials or UNKNOWN WORD.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ZGkimpfyh63EZyRkNbkPD
Console sessions now route through the same general VM-to-VM messaging
system (Phase C) any VM can already use for its own reasons -- not a
synchronous shortcut. Per direct instruction: real async MSG-SEND/
MSG-DELIVER (Option B), not a VM-EXEC-based synchronous relay, because
messaging is a general capability, not a console-specific mechanism.
New CONSOLE-CMD-EVENT message type (common:messaging.4th). New
sk_repl_dispatch_line() (repl.c), called from both sk_repl_step and
sk_repl_run in place of a direct vm_interpret(): if the active VM's own
name has a live "<name>~user" counterpart registered, the raw input
line is wrapped as an S"-embedded CONSOLE-CMD-EVENT MSG-SEND and
interpreted on the console VM instead of being run directly -- the
console's own next MSG-TICK (Hera's idle pump) delivers it into the
paired user VM via VM-EXEC, same mechanism every other message already
uses. Falls back to direct interpretation if there's no pairing, or if
the line contains a `"` (known v1 limitation, warned about explicitly
rather than silently mishandled).
New capsule_console_birth() (capsule_console.h/.c): a bare VM whose
only content is loading common:messaging.4th -- the console side of a
pairing, parallel in shape to RUNCAP's user-VM birth but with fixed
embedded content instead of a devblock read (no identity, no thumbdrive
involved). New PAIR-TEST diagnostic word (mama_forth_words.c, matches
RUNCAP-TEST's own precedent): births both halves of a pairing and
registers the "<name>~user" mapping. Not the real pairing call site --
that's the eventual attach/onboarding flow -- this exists to exercise
the relay live before that flow exists.
Found and fixed a real, serious bug live: console_set_vm_name() stored
the caller's raw pointer instead of copying it. mama_word_use() (USE)
passes a VMRegistryEntry field living on its own stack frame -- once
USE returns, that pointer dangles, corrupting every console tag after
the first USE (observed directly as garbled "[[]" / binary-looking
prefixes instead of "[CaptBob]"). Fixed at the source: console_set_
vm_name() now copies into internal storage. That surfaced a second,
related bug across every console_get_vm_name()-based save/restore call
site in mama_forth_words.c (BIRTH, VM-STEP, VM-EXEC, CONNECT-HERMES,
CONNECT-ARTEMIS): saving just a pointer into the single internal buffer
meant an intervening console_set_vm_name() call silently corrupted the
saved value before the restore ever ran. New console_save_vm_name()
copies into caller-owned storage; every save/restore site updated.
Verified end-to-end, live in QEMU: typed WELCOME at a paired console
VM -- it did not execute directly (no UNKNOWN WORD), printed ok
immediately (queued, async), and on the next idle tick
"[CaptBob~user] Minted identity -- default personality" appeared on
its own -- genuine delivery and execution in the paired user VM through
the real MSG-SEND/MSG-DELIVER pipeline. Console tags confirmed clean
(no garbling) across all three architectures' full regression boot.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ZGkimpfyh63EZyRkNbkPD
Three tightly-coupled changes, verified together per Captain Bob's own
"getting rid of the emergency cli" direction:
1. Zuse's identity is thumbdrive-resident, never system-resident. New
zuse_genesis_marker_t (magic/version/zuse_pubkey[32]/crc) replaces
zuse_cert_devblock_t's slot in the top-of-device fence -- the system
now remembers only that a root identity exists and its pubkey, never
a seed. zuse_cert_devblock_t is kept in the repo, marked superseded,
no longer written by any code path.
capsule_mint_identity() grows a genesis mode (issuer_vm=NULL): no
cert is built or written (Zuse isn't verified against a separate
signer -- she's recognized by pubkey match against the marker) and
two new optional out-params (out_pubkey/out_seed) let the caller
install the cert immediately after a genesis mint.
New capsule_zuse_boot_try_attach() (capsule_zuse_boot.c), called
from sk_repl_idle() on every fresh USB attach (the only point in the
boot lifecycle a thumbdrive can actually be detected -- attach
polling doesn't exist yet at kernel_main.c's old one-shot mint point,
which is why that whole block is gone): no marker + blank drive ->
genesis-mint; marker present + matching drive -> read its own
user_identity_seed_t, install the cert. Either way, re-runs
ACL-ZUSE-BOOT (zuse.4th) so zuse_session activates exactly like it
always has for a same-boot cert install -- ACL-PIN only blocks
redefinition, not re-execution, so no new C-side auth logic needed.
2. ACL.4th activated (capsules/init.4th) -- inactive all session until
now. Found and fixed a real bug this immediately surfaced: zuse.4th's
ACL-ZUSE-BOOT tried `['] ACL-ZUSE-BOOT ACL-PIN` from inside its own
still-compiling definition -- the word isn't findable yet at that
point, so the whole definition silently failed to compile every
previous boot this session (dormant, since ACL.4th never loaded).
Fixed: pin after the definition closes, not from within it -- it
only needs to happen once anyway, and pinning doesn't block the
re-invocation genesis/attach needs.
3. The unauthenticated emergency-CLI ACL bypass is retired
(repl.c): `emergency_console = is_hera ? (zuse_session ? 0 : 1) : 0`
deleted from both sk_repl_step and sk_repl_run. Every word run from
Hera's own bare prompt now goes through ordinary ACL enforcement;
emergency_console is driven only by the genuine C-level fault
handler again.
Added ZUSE-SESSION? (starforth_words.c), a read-only diagnostic
matching ZUSE-PUBKEY@'s own precedent, to verify the whole chain
directly rather than by inference.
Verified end-to-end live in QEMU: fresh boot, no thumbdrive ->
ZUSE-SESSION? reads 0. Attach a genuinely blank drive via QMP -> genesis
mint fires automatically (no typing) -> ZUSE-SESSION? reads -1 (true).
Hermes/Artemis both birth clean on all three architectures with ACL
now actually enforced for the first time all session -- no denials, no
UNKNOWN WORD beyond the deliberate POST self-test cases.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ZGkimpfyh63EZyRkNbkPD
disk/zuse.img and disk/user1.img now hold real Ed25519 keypairs, a
Zuse-signed cert, and real profile data (Konrad Suse/Zuse and Captain
Bob/CaptBob respectively), minted this session via MINT and verified
byte-exact against the raw devblocks. Keeping them as tracked fixtures
for the Apollo test phase.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ZGkimpfyh63EZyRkNbkPD
Extends user_identity_seed_t (version 2) with fixed-size full_name/
username/email/phone fields -- plenty of unused pad space (4016 bytes)
was already there. Deliberately NOT encoded into the DER cert's Subject
field: that would mean building a real X.509 RDNSequence (OIDs for
commonName/emailAddress, PrintableString/UTF8String tagging), well past
this project's own stated "deliberately not a general ASN.1/X.509
[builder]" scope. This human-readable profile data isn't
security-relevant the way pubkey/serial are (the only two fields
CERTVERIFY/BINDSTEP actually check) -- it travels alongside the keypair
in the plain identity record instead.
capsule_mint_identity() takes full_name/username (required, validated
non-empty and within their fixed field widths) and email/phone (NULL or
empty = null, matching the schema's own nullable convention). The MINT
FORTH word's stack signature grows to 4 string pairs
( fname-c fname-u uname-c uname-u email-c email-u phone-c phone-u -- ok? ).
Verified live in QEMU: minted two real identities with real profile
data -- Zuse (full_name "Konrad Suse", username "Zuse",
zuse@pantheon.org) onto disk/zuse.img, and a regular user (full_name
"Captain Bob", username "CaptBob", capt.bob@pantheon.org) onto
disk/user1.img -- then read the raw devblock bytes back off both images
directly and confirmed every field byte-exact at its correct struct
offset. Clean 3-architecture regression boot confirms no side effects.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019ZGkimpfyh63EZyRkNbkPD
