v4 ACL: the countdown at call, as v3 -- the call hook may redirect
Captain Bob ruled "As v3: add at call", "Engine: call redirects" and "Build it". - The engine's call hook returns where P goes. - The kernel checks every call from code defined at run time. A TTL is counted down at the call. At a TTL of 0, or a denied word, the call is sent to (ACL-CALL), with the target on the data stack, which rechecks or refuses it by the word's own name, then goes on to it. - A word being compiled is no longer checked, as in v3. Two things were added while building, both from v3: - (ACL-BUSY) turns the check off while the policy rechecks. Without it the recheck recursed. - (ACL-CALL) does its own refusal and leaves WORD's buffer alone. A first version broke POST's CONVERT.digits. Test first: test_mesh_boot's five new checks failed before; now 120 pass and POST is 990/990. test_exec tests the redirect. test_host_quit's two v4-only expectations now follow v3: a denied word compiles. All 52 tests, test-32 and ASan/UBSan pass. An amd64 session passes 18/18. Five alternating host boots: +1.4% median, inside the noise. DECOMPOSITION.md 5.20 is updated, and the FPGA's call opcode is recorded as owing the same redirect. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017RzZgrf7V7Zk2iczBXDRbm
This commit is contained in:
@@ -892,11 +892,20 @@ is rechecked, otherwise the TTL is counted down; then a denied word is refused w
|
||||
is in the variable `ACL-HOOK`; with none there, the word is allowed and given the longest TTL. `ACL.fth` puts
|
||||
its `ACL-RECHECK` there (v3 finds it by name each time).
|
||||
|
||||
**Where this is weaker than v3.** v3's inner interpreter checks every execution, inside definitions too. v4
|
||||
code is native: a compiled call goes straight to its word, and no opcode asks permission. So v4 checks a
|
||||
word when it is *compiled* as well as when it is interpreted — a denied word cannot be built into anything
|
||||
new — but a call compiled while the word was allowed is not checked again when it runs. Closing that needs
|
||||
the check in the node itself (the `call` opcode), which is a hardware decision and is not taken here.
|
||||
**Where this was weaker than v3, and is not now** (2026-11-19, Captain Bob: "As v3: add at call"; "Engine: call
|
||||
redirects"; `docs/v4.0.0/plans/2026-11-19-acl-at-call.md`).
|
||||
- **The old gap:** v3's inner interpreter checks every execution, inside definitions too. v4 checked a word
|
||||
when it was compiled and when it was interpreted, and a compiled call was never checked again.
|
||||
- **Now, at the call:** the engine's call hook may send a call elsewhere. The kernel checks every call from
|
||||
code defined at run time. A word with TTL left is counted down at the call. At a TTL of 0, or for a denied
|
||||
word, the call goes to `(ACL-CALL)` (`acl.f18`), which rechecks or refuses it by the word's name and goes on
|
||||
to it.
|
||||
- **No compile-time check:** a word being compiled is not checked, as in v3.
|
||||
- **The policy runs unchecked:** while the policy rechecks a word, `(ACL-BUSY)` is 1 and nothing is checked,
|
||||
as v3 runs its policy with the check off.
|
||||
- **A node with no kernel has no check at call:** the hosted node is one. Its compiled calls are not checked.
|
||||
- **On the FPGA,** the `call` opcode must offer its kernel the same redirect. That is to be designed with the
|
||||
FPGA.
|
||||
|
||||
| Word | Fate | Notes |
|
||||
| --- | --- | --- |
|
||||
|
||||
Reference in New Issue
Block a user