docs(ci): name the timing fields this forge actually returns

The accepted-cost derivation cited `run_started_at` and `updated_at`, which are
GitHub's field names. This Gitea's runs payload has neither -- it returns
`started_at` and `completed_at`, and omits `run_started_at`, `updated_at` and
`created_at` entirely (confirmed by dumping the keys of a run object). The
figures are unaffected: the script that produced them fell through to the real
fields, so 83 s median over 77-106 s stands.

The derivation was named so a reader could re-take the measurement, and as
written it returned nothing when followed literally, which defeats the only
reason it was there.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LjbhSqQf3pwnPA6MVaWcWL

.gitea/workflows/ci.yaml

@@ -0,0 +1,103 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+    # Spelled out only to keep `edited` in the list — naming any type replaces
+    # the whole default set, so the other three have to be restated.
+    #
+    # `edited`, not `ready_for_review`: draft here is the `WIP:` title prefix,
+    # so un-drafting is a title edit and no ready-for-review action is ever
+    # raised. That edit is what creates the run which lifts the `if:` skip
+    # below (decided once, when a run is CREATED). Do not swap it back and do
+    # not drop this to the bare default — either restores the bug. The accepted
+    # cost and the evidence are in README's Development section.
+    types: [opened, synchronize, reopened, edited]
+
+# gitdan-ci runs four repos' CI on two capacity slots, and the compiler-backed
+# suites below are multi-minute. A superseded run costs a slot in front of
+# somebody's build, so drop it.
+#
+# `push` groups on `github.sha` rather than `github.ref`: a constant per-branch
+# group is what let this Gitea cancel two of daniel/gitdan's merge runs
+# outright while `cancel-in-progress` was gated away from `push` entirely —
+# see the long note in that repo's ci.yaml. Observed on 1.26.0; the instance
+# is 1.27.2 now and whether it persists is unverified, which is why every
+# commit gets its own group instead of trusting the flag.
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && github.ref || github.sha }}
+  cancel-in-progress: true
+
+jobs:
+  selftest:
+    name: shellcheck + selftests
+    # Two clauses, both load-bearing. The second skips draft PRs: Gitea sets
+    # draft:true when the title starts with `WIP:`, so work-in-progress pushes
+    # cost the shared runner nothing until the PR is un-WIP'd. The first is
+    # what keeps that from also skipping pushes to `main` — a `push` event has
+    # no `pull_request` context, so `github.event.pull_request.draft` is empty
+    # there and the negation alone would be unreliable. Never drop it.
+    if: ${{ github.event_name != 'pull_request' || !github.event.pull_request.draft }}
+    runs-on: ubuntu-latest
+    timeout-minutes: 20
+    # Deliberately no `container.volumes:` entry, unlike every repo that
+    # CONSUMES this action. The suites below build throwaway workspaces under
+    # `mktemp -d` and want a cold target dir every time — a persistent cache
+    # would make "did this run rebuild?" unanswerable, which is the question
+    # restore-mtimes-selftest.sh exists to ask. So this job takes no share of
+    # the shared CI cache disk budget.
+    steps:
+      - name: Checkout sources
+        uses: actions/checkout@v4
+
+      - name: Install shellcheck
+        uses: taiki-e/install-action@v2
+        with:
+          tool: shellcheck
+
+      # `hardlink-clone-selftest.sh` and `restore-mtimes-selftest.sh` drive a
+      # real Cargo against a real scratch workspace — they are the only things
+      # here that verify the hardlink-aliasing and mtime-freshness behaviour
+      # against the compiler rather than against a fixture, and selftest.sh's
+      # own header says `--fast` is for iterating, not for signing off a
+      # change. So CI installs a toolchain and runs the full set.
+      #
+      # The scratch workspaces use path dependencies only, so nothing here
+      # reaches crates.io.
+      # Nightly first, stable second, so stable ends up the default and
+      # nightly is only reachable through an explicit `+nightly`.
+      #
+      # `hardlink-clone-selftest.sh`'s last scenario needs a Cargo that
+      # resolves freshness by CONTENT — the mode where the dep-info file
+      # carries per-source checksums, which is the mutation that turns a
+      # hardlink clone into silent stale-artifact reuse rather than a slow
+      # build. This step is what supplies it, and as of 2026-08-26 it does:
+      # the scenario ran and passed against 1.100.0-nightly.
+      #
+      # It briefly did not. Cargo PR #17382 (2026-08-22) demoted
+      # `-Z checksum-freshness` to a gate and gave `build.fingerprint` the
+      # choice, defaulting to `mtime`, so the suite — which set only the gate —
+      # measured a genuine INACTIVE and skipped its strongest scenario. That
+      # read as "upstream withdrew content freshness" and was written up here
+      # as this step buying nothing. It was a moved switch, not a withdrawal;
+      # the suite now exports both and the coverage is back. See
+      # daniel/gitdan#62 for the investigation.
+      - name: Install Rust nightly
+        uses: dtolnay/rust-toolchain@nightly
+      - name: Install Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+
+      # Every script, including the suites themselves. `-x` follows the
+      # `. cache-lib.sh` each one sources, which is where most of the logic
+      # being checked actually lives; without it shellcheck reports SC1091 and
+      # analyses each file with a hole in it.
+      - name: shellcheck
+        run: shellcheck -x --source-path=scripts scripts/*.sh
+
+      # One command, not six: selftest.sh is the entry point a developer runs,
+      # so a suite added there is gated here without a matching edit in this
+      # file.
+      - name: Selftests
+        run: bash scripts/selftest.sh

README.md

@@ -29,16 +29,63 @@ on the runner having a single execution slot.
 
 One thing neither project had, and the reason the clone is not a plain
 `cp -al`: **a build inside a hardlink clone does mutate the directory it was
-cloned from.** Cargo writes real artifacts by replacing them, but writes its
+cloned from.** rustc renames its own outputs into place, but Cargo writes its
 metadata — and build scripts write their `OUT_DIR` — with a plain truncating
-write, straight through the shared inode. Under
+write, straight through the shared inode. When Cargo resolves freshness by
-`CARGO_UNSTABLE_CHECKSUM_FRESHNESS` the file that gets corrupted is
+content the file that gets corrupted is its dep-info fingerprint —
-`.fingerprint/<unit>/dep-<target>`, which holds the per-source checksums that
+`.fingerprint/<unit>/dep-<target>`, or
-decide freshness, and the failure is silent stale-artifact reuse rather than a
+`build/<pkg>/<hash>/fingerprint/dep-<target>` under Cargo's build-dir layout v2
-slow build. `scripts/hardlink-clone-selftest.sh` reproduces it as an explicit
+— which holds the per-source checksums that decide freshness, and the failure
-control and asserts the fix. The fix is to hardlink the artifacts (the GB) and
+is silent stale-artifact reuse rather than a slow build.
-real-copy the metadata (the MB) — about 3.7% of a Bevy-sized target directory,
+
-against 100% for a full copy.
+One more family joins them, and it is not metadata: **anything the linker
+writes**. rustc writes an `.rlib` or `.rmeta` to a temporary and renames it
+into place, but a linked executable is written through whatever inode is
+already at its path — so a `cargo test --no-run` in a raw `cp -al` clone
+rewrites the source's own test binary. Measured 2026-08-27 on cargo 1.93.1
+stable, 1.96.0-nightly, 1.98.0-nightly and 1.100.0-nightly, under both
+build-dir layouts.
+
+`scripts/hardlink-clone-selftest.sh` reproduces both as explicit controls and
+asserts the fix. The fix is to hardlink what rustc renames into place — the
+`.rlib`, `.rmeta` and `incremental/` bulk — and real-copy the metadata and the
+linker outputs.
+
+**The selection names that set directly, under either build-dir layout.**
+Layout v2 regroups everything per build unit under
+`build/<pkg>/<hash>/{fingerprint,out,run}/`, artifacts included, so there is no
+`.fingerprint` and no `deps` to key off and `build/` is no longer a proxy for
+"metadata" — it is the whole tree. The one place the two layouts genuinely
+differ is that under v2 a build script's `OUT_DIR` and a compile unit's rlib
+are both a directory called `out`.
+
+**Ambiguity there resolves toward unsharing**, because over-unsharing costs
+bytes and under-unsharing costs corruption. An `out` directory stays shared
+only when two independent signals agree it is a compile unit's: it holds an
+`.rlib`/`.rmeta` of its own, *and* its unit carries no record of a build-script
+execution beside it (`run/` under v2, a loose `root-output` under v1). The
+execution record alone is not enough — Cargo writes it only after the script
+succeeds, so a build script that populates `OUT_DIR` and then fails leaves a
+unit that reads as a compile unit. v2 is the nightly default and stabilises in
+cargo 1.100.0 on 2026-11-12.
+
+### What it costs
+
+Real-copied share of a 5.5 GB Bevy target directory, before and after the
+linker-output rule landed:
+
+| tree | before | after |
+|---|---|---|
+| **excluding `incremental/`** — the figure to plan against, since the quick-start below sets `CARGO_INCREMENTAL: 0` | **36.4%** | **57.0%** |
+| whole tree, `incremental/` included (a local dev checkout, not CI) | 9.0% | 14.1% |
+
+The first row is the one a CI consumer gets. The increase is the linker-output
+rule, not the layout work: on a scratch crate the layout fix alone takes v2
+from 99.996% to 0.2%.
+
+The copy is paid per clone and does not amortise — a fresh `cp -al` leaves
+every file with `nlink >= 2`, so the `-links +1` filter cannot skip anything —
+and a clone happens twice per job, once seeding and once publishing.
 
 ---
 
@@ -98,10 +145,14 @@ env:
   CARGO_INCREMENTAL: 0                  # per-run bloat on a persistent volume
   CARGO_PROFILE_DEV_DEBUG: line-tables-only
   CARGO_PROFILE_TEST_DEBUG: line-tables-only
-  # Nightly only. Content-addressed freshness instead of mtime-based — a
+  # Nightly only, and BOTH are needed. Content-addressed freshness instead of
-  # strictly stronger guarantee, complementary to the mtime restore (which
+  # mtime-based — a strictly stronger guarantee, complementary to the mtime
-  # still covers directory-form `rerun-if-changed` build-script watches).
+  # restore (which still covers directory-form `rerun-if-changed` build-script
+  # watches). Since cargo PR #17382 (2026-08-22) the `-Z` gate below only
+  # unlocks the feature; `build.fingerprint` selects it and defaults to
+  # `mtime`, so the gate on its own is accepted and does nothing.
   CARGO_UNSTABLE_CHECKSUM_FRESHNESS: "true"
+  CARGO_BUILD_FINGERPRINT: "content"
 ```
 
 ---
@@ -287,6 +338,69 @@ Lowering either here needs no coordination: the arbiter then only defers a
 reclamation this side would already have permitted, which costs disk rather
 than correctness.
 
+### Names this repo doesn't reclaim, but the arbiter depends on
+
+The five names above are leftovers — dead trees gitdan's arbiter finds and
+deletes. Four other dot-prefixed names change that arbiter's *behaviour*
+without ever being a leftover: it reads them to make a correctness decision
+and never reclaims them. They are a real cross-repo dependency too, on the
+name's spelling rather than on the tree's lifetime — gitdan's
+`DEPENDED-UPON NAMES CONTRACT` block (in `scripts/ci-cache-reclaim.sh`) is the
+canonical description; this is the producer's half for the two we own.
+
+| Name | Produced here by | Consequence of an unannounced rename |
+|---|---|---|
+| `.ci-lock-<id>` | `scripts/cache-lock.sh` (acquire/release) and `cache-lib.sh`'s `write_cache_lock()` | The arbiter's lock check silently stops matching — a live job's staging tree loses its liveness guard and becomes an ordinary reclaim candidate while still in use |
+| `.cache-last-used` | `cargo-cache/action.yml`, stamped every run | The arbiter falls back to directory mtime — silently reordering its eviction order, and possibly failing to recognise the directory as a cache dir at all |
+
+Two more names are in gitdan's list — `.gitea-last-used` and `.ci-keep` — and
+neither is produced by anything in this repo, so there is no producer-side
+half to write here. `.gitea-last-used` is a naming convention individual
+repos used before adopting this shared action; nothing here writes it, and
+gitdan's script reads it only for compatibility with directories created
+under that older scheme. `.ci-keep` is a per-repo, hand-placed opt-out a
+consuming repo's own workflow drops directly into a cache directory it wants
+exempted from eviction — never something this action or its scripts write.
+
+**Adding a fifth depended-upon name counts as much as renaming one of the two
+above.** If a future change here makes gitdan's arbiter start depending on the
+spelling of some new dot-prefixed name — a name it reads for a decision but
+never reclaims — that is exactly this category, and it needs the matching
+`DEPEND_*` entry on gitdan's side before it ships, not after.
+
+### The directory LAYOUT is part of that contract as well
+
+Names are one half; where they sit is the other. gitdan's arbiter walks a
+volume's `_data` tree to `CI_CACHE_MAX_DEPTH`, which is **2** — deliberately
+tight, because a deeper walk starts meeting Cargo's own
+`incremental/<crate>-<hash>` directories, which match the same name shape it
+uses to recognise a cache dir and must never be evicted individually. So:
+
+```
+_data/target-<key>                    depth 1 — no lineage
+_data/<lineage>/target-<key>          depth 2 — a lineage
+_data/<a>/<b>/target-<key>            depth 3 — INVISIBLE to the arbiter
+```
+
+That budget is the whole reason `cache-lineage` is one path component and not
+a path. Nesting deeper is not an error anywhere: the caches work, the
+in-workflow prune pass keeps managing them, and the one script whose job is
+the shared disk budget across every repo simply never sees them again.
+
+It is also why the fix for daniel/gitdan#60 nests rather than suffixing the
+cache key. A `target-<key>-<lineage>` name would be read as dead by
+`prune-cache.sh`'s liveness pass — which classifies by recomputing
+`target-<cache_key(branch)>` for every branch on origin — and evicted
+unconditionally on every run; and it falls out of the arbiter's own
+`BRANCH_DIR_RE` too, so the same directories would never be candidates there
+either. Nesting leaves both matchers reading exactly the names they already
+read, one level down.
+
+One known rough edge, on gitdan's side and cosmetic: that script logs an
+eviction as `<volume>/<basename>`, so a nested `target-<key>` and a flat one
+of the same key are indistinguishable in its output. It evicts the right
+directory; the line just doesn't say which.
+
 ---
 
 ## Inputs
@@ -296,6 +410,7 @@ than correctness.
 | input | default | meaning |
 |---|---|---|
 | `cache-root` | `/cache` | mount point of the persistent volume inside the job container |
+| `cache-lineage` | *(empty)* | one directory level under `cache-root`, for a second job building the same ref for a different target or profile — see [Multiple jobs in one workflow](#multiple-jobs-in-one-workflow) |
 | `protected-branches` | `dev main` | refs that publish snapshots and are never evicted |
 | `min-free-percent` | `10` | prune when free space drops below this |
 | `restore-mtimes` | `true` | restore tracked-file mtimes from git history |
@@ -304,7 +419,7 @@ than correctness.
 | `own-ref` | *(auto)* | override; defaults to `github.head_ref`, else `github.ref_name` |
 | `base-ref` | *(auto)* | override; defaults to `github.base_ref` (empty on push) |
 | `seed-fallback-dir` | *(empty)* | absolute path to seed from when no snapshot exists — for migrating off an existing flat cache |
-| `watermark-file` | `.ci-watermark-<job>-sha` | must differ per job when two jobs share one cache key |
+| `watermark-file` | `.ci-watermark-<job>-sha` | must differ per job when two jobs share one target directory; the default already does |
 | `lock-id` | `<job>-<run_id>` | identifies this job's cache lock |
 | `stale-lock-seconds` | `7200` | age past which another job's lock is treated as abandoned |
 
@@ -320,6 +435,7 @@ Exports to the job environment: `CARGO_TARGET_DIR`, `CARGO_CACHE_ROOT`,
 | input | default | meaning |
 |---|---|---|
 | `cache-root` | `/cache` | must match the consume action |
+| `cache-lineage` | *(empty)* | must match the consume action; a mismatch fails the step rather than publishing the wrong tree |
 | `protected-branches` | `dev main` | refs that publish snapshots |
 | `mode` | `publish` | `publish`, or `release-lock` for the `if: always()` step |
 | `own-ref` | *(auto)* | override; defaults to `github.head_ref`, else `github.ref_name` |
@@ -336,13 +452,70 @@ of a merge-preview build, which is not what `dev` is.
 
 ## Multiple jobs in one workflow
 
-Jobs sharing a cache key (a `ci` job and a `wasm` job on the same branch, say)
+Two jobs building the same ref — a `ci` job and a `wasm` job, say — are two
-each need their **own** watermark file. A shared one breaks the moment two
+consumers of one cache key, and the cache key alone is not enough to keep them
-jobs run in sequence within one trigger: job A advances the watermark to HEAD,
+apart.
-and job B then reads that just-advanced value, computes an empty diff, and
+
-loses the merge protection entirely. The default (`.ci-watermark-<job>-sha`)
+**Give each its own lineage.** A cache key names a *ref*; what a target
-already gives each job its own; only override `watermark-file` if you also
+directory holds is the product of a ref and a build configuration. Left to the
-override `lock-id`, and then keep both distinct per job.
+key alone, both jobs export the same `CARGO_TARGET_DIR`, and Cargo's
+build-directory lock is exclusive — so on a runner with more than one slot the
+second job sits on `Blocking waiting for file lock on build directory` for the
+length of the first, occupying a capacity slot while doing nothing
+(daniel/gitdan#60). `cache-lineage` is that second dimension:
+
+```yaml
+      - name: Restore the Cargo cache
+        uses: https://gitdan.com/daniel/gitdan-actions/cargo-cache@v1
+        with:
+          cache-lineage: wasm32          # the `ci` job sets none
+
+      # ... build steps ...
+
+      - name: Record watermark, publish cache snapshot
+        uses: https://gitdan.com/daniel/gitdan-actions/cargo-cache-publish@v1
+        with:
+          cache-lineage: wasm32          # the SAME value, or the step fails
+```
+
+A lineage nests one directory level under the cache root
+(`<cache-root>/<lineage>/target-<key>`), so each lineage gets its own target
+dirs, its own snapshots, and its own prune pass. Everything else works as it
+already did, one level down: a PR branch in a lineage layers over **that
+lineage's** base snapshot, the publisher branch publishes into it, and a prune
+pass run inside it never sees a sibling lineage's caches.
+
+Setting no lineage resolves to the cache root unchanged, byte for byte, so a
+workflow that does not use one keeps the exact directories it already has on
+the volume.
+
+**Both actions need the same value.** `cargo-cache-publish` derives both ends
+of the snapshot swap from its own `cache-root`, so a publish step left at the
+default while its consume step nested would republish a *different* lineage's
+live target dir over that lineage's snapshot, on every push, with nothing in
+the log to say so. The publish action therefore compares its own inputs
+against the `CARGO_CACHE_ROOT` the consume step exported and fails the step on
+a mismatch. (The `mode: release-lock` call is exempt: it releases a lock on
+`$CARGO_TARGET_DIR` and never touches a cache root, so it takes no lineage.)
+
+**Some lineage names are refused.** A lineage is one path component, drawn
+from `[A-Za-z0-9._-]`, and several otherwise-reasonable names are rejected at
+resolve time because a *reader elsewhere* would stop seeing the caches
+underneath them: a Cargo profile name (`debug`, `release`, `doc`, …) is one
+gitdan's arbiter never descends into, a `target-`/`snapshot-` prefix makes the
+lineage directory itself an eviction candidate for this repo's own prune pass,
+and a hex-suffixed name is read by that arbiter as a per-branch cache dir in
+its own right. `validate_cache_lineage()` in `scripts/cache-lib.sh` states each
+rejection with the reader that imposes it.
+
+**Watermarks are still per job.** Two jobs in one lineage — or one job before
+lineages were introduced — each need their **own** watermark file. A shared one
+breaks the moment two jobs run in sequence within one trigger: job A advances
+the watermark to HEAD, and job B then reads that just-advanced value, computes
+an empty diff, and loses the merge protection entirely. The default
+(`.ci-watermark-<job>-sha`) already gives each job its own; only override
+`watermark-file` if you also override `lock-id`, and then keep both distinct
+per job.
 
 ---
 
@@ -415,14 +588,69 @@ not automatic.
 ## Development
 
 ```bash
-bash scripts/selftest.sh          # everything (~1 min; needs cargo)
+shellcheck -x --source-path=scripts scripts/*.sh
+bash scripts/selftest.sh          # everything (needs cargo)
 bash scripts/selftest.sh --fast   # fixture-only suites, no compiler
 ```
 
+Both run in CI — `.gitea/workflows/ci.yaml`, one job, on pushes to `main` and
+on PRs that were non-draft when the run was created. It installs shellcheck
+and both a stable and a nightly Rust toolchain (nightly so
+`hardlink-clone-selftest.sh` can run its content-freshness scenario, which as
+of 2026-08-26 a nightly does enable — 1.100.0-nightly (787af2b8c 2026-08-25)
+resolves freshness by content given both `CARGO_UNSTABLE_CHECKSUM_FRESHNESS`
+and `CARGO_BUILD_FINGERPRINT: content`, per cargo PR #17382; the suite still
+settles that by experiment on every run and skips the scenario loudly when it
+cannot measure) and references no credentials; the scratch workspaces the
+compiler-backed suites build use path dependencies only, so nothing reaches
+crates.io. It runs the full suite rather than `--fast`,
+because the two compiler-backed suites are the ones that check this scheme
+against real Cargo instead of against a fixture.
+
+**Draft (`WIP:`-titled) PRs skip it; un-drafting un-skips them, through
+`edited`** — no empty commit needed. The skip is decided when a run is
+*created*, so lifting it needs an event that creates one, and un-drafting on
+this Gitea is a title edit: `edited` is in the workflow's `pull_request` types
+for exactly that reason. `ready_for_review` held that slot first and never
+fired — this Gitea has no draft column and no ready-for-review event at all,
+draft being computed from the title prefix — so a PR opened as `WIP:` carried
+its skip decision all the way to merge unless some later push happened to
+create a run. Do not swap the type back and do not drop the `types:` list to
+its bare default; either restores the bug.
+
+**Accepted cost: a body edit on an already-non-draft PR now triggers a real
+run.** Gitea populates no `changes` field for a title-or-body edit, unlike
+GitHub, so the workflow cannot tell the edit that un-drafts a PR from an
+ordinary body PATCH — a closing-reference fix-up, say. The price is around
+**90 seconds** of a runner shared across four repos on two capacity slots: the
+last twelve non-skipped runs of this job, `started_at` to `completed_at` off
+the Actions API, are 83 s median over 77–106 s. `timeout-minutes: 20` is a
+ceiling for a hung suite, not a duration. It is also bounded: the workflow's
+`concurrency:` block groups `pull_request` runs on `github.ref` with
+`cancel-in-progress: true`, so a burst of edits collapses to one run rather
+than N. A still-draft PR pays nothing extra — the `if:` guard skips those
+exactly as before.
+
+The evidence for `edited` comes from `daniel/emowheel`, which hit the identical
+bug, shipped the identical wrong fix, and corrected it in commit `08da820` (see
+`daniel/emowheel#71`): Gitea 1.27.2's `HookIssueAction` enum has no
+`ready_for_review` entry and no notifier emits one, while
+`issue.IsWorkInProgress` derives draft from the title. Live since: emowheel PR
+#141 was un-drafted at 19:01:13 on 2026-08-31 and run 2801 was created two
+seconds later on the **same** head SHA as the two runs skipped before it — a
+run created by the title edit alone, with no push, that executed and passed.
+`daniel/gitdan` ported the same one-word change. That it behaves the same way
+*here* has not been demonstrated in this repo; the next `WIP:` PR opened
+against `main` is the test.
+
+This repo is consumed by three other repos' CI at `@v1`, a moving tag, so a
+change here reaches all of them at once. That is what the gate is for.
+
 | suite | covers |
 |---|---|
-| `hardlink-clone-selftest.sh` | that a build in a clone cannot mutate its source — with a control proving a raw `cp -al` does. Needs a real compiler. |
+| `cache-root-selftest.sh` | that a lineage nests one level and nothing else moves: no lineage resolves byte-for-byte to the cache root, two lineages on one cache key get disjoint target dirs, seed/publish/prune all stay inside their own lineage, a PR layers over its own lineage's base snapshot — **and one rejection per lineage name a reader elsewhere would stop seeing**, plus the publish-side mismatch guard |
-| `seed-target-dir-selftest.sh` | seed-source preference, lock-file stripping, two jobs racing on one cache key, **and one scenario per check a hardlink clone is validated against**: a source rotated wholesale, a subtree silently lost from the walk, a copy that reports failure over a tree both other checks read as whole, and a source identity that resolved at neither end — plus a staging tree that could not be privately owned being discarded rather than published |
+| `hardlink-clone-selftest.sh` | that a build in a clone cannot mutate its source — with a control proving a raw `cp -al` does. Needs a real compiler, **and a nightly that actually resolves freshness by content for its last scenario**: the source's-next-build check reasons about content rather than mtime, so under mtime freshness it would assert a bug. Whether the toolchain does is settled by experiment on a throwaway crate, not by asking it — accepting `-Z checksum-freshness` stopped implying it on 2026-08-22, when cargo PR #17382 demoted the flag to a gate and gave `build.fingerprint` (default `mtime`) the choice; the suite now exports both and 1.100.0-nightly measures ACTIVE again. The experiment reports **three** outcomes, not two: active, measured-inactive, and *not measured*. Its answer codes are `0` and `3`, deliberately clear of every status bash generates for its own errors — so nothing that goes wrong inside the probe, including an expansion failure no guard can catch, can be read as an answer. The scenario is skipped for the last two alike, but a failure to measure is never reported as a measurement. The control also reports which mutation families the running Cargo exhibits — a note, not an assertion, since that set moves upstream. It also pins the SELECTION itself against both of Cargo's build-dir layouts, from two file-only fixtures that need no compiler — so the layout the installed Cargo does not happen to write is still covered — and asserts the partition in both directions: every file Cargo rewrites in place is privately owned, and every `.rlib`/`.rmeta` still shares its inode. The second half is what the old suite never checked beyond `shared > 0`, and it is what a layout change silently inverts. |
+| `seed-target-dir-selftest.sh` | seed-source preference, lock-file stripping, two jobs racing on one cache key, **and one scenario per check a hardlink clone is validated against**: a source rotated wholesale, a subtree silently lost from the walk, a copy that reports failure over a tree both other checks read as whole, and a source identity that resolved at neither end — plus a staging tree that could not be privately owned being discarded rather than published, and the publisher's log showing it waited on the consumer's own reader-lock marker before reclaiming a rotated snapshot |
 | `publish-snapshot-selftest.sh` | the atomic swap, that a live consumer survives a republish, and the publisher's side of the rotation race: deferred reclamation under a live reader, and its sweep once the reader is gone |
 | `prune-cache-selftest.sh` | liveness, protection, locking, eviction order, self-clear, **and that a cache a job claims *inside* the check-to-unlink window survives it** — against a real scratch `origin` |
 | `restore-mtimes-selftest.sh` | the merge hazard and the watermark that closes it, including the two-jobs-one-namespace case. Needs a real compiler. |
@@ -442,7 +670,7 @@ key and asserts only what must be true whichever of them wins the rename.
 which places the interference inside the window rather than hoping it lands
 there. `prune-cache-selftest.sh` scenario 12 stubs `du`, so the pass's own
 measurement publishes a reader marker strictly between its check and its
-unlink. `seed-target-dir-selftest.sh` uses the shape four times over, on two
+unlink. `seed-target-dir-selftest.sh` uses the shape five times over, on two
 different commands. Scenarios 8a, 8b and 8c stub `cp`, so the consumer's own
 clone is what rotates the snapshot underneath it, loses a subtree of its own
 source, or reports a failure over a tree that is in fact whole — each strictly
@@ -453,9 +681,14 @@ could not be READ rather than on one that changed, and the only way to make
 that the sole witness is to fail the identity reads while the copy between
 them succeeds. 8a also starts a second real process — the actual
 `publish-snapshot.sh` — but the stub is what fixes where its swap lands; the
-concurrency is incidental to the determinism. Every stub asserts that it
+concurrency is incidental to the determinism. Scenario 11 reuses 8a's exact
-fired, because a scenario whose interference silently did not happen passes
+stub and the same forced rotation, but reads a different witness: not the
-for the wrong reason.
+consumer's own checks, but a line in the *publisher's* log reporting that it
+waited on the reader marker this consumer's clone wrote — `reader_lock_acquire`
+is exercised by 8a already, but nothing asserts it actually fired until this
+scenario reads that line back. Every stub asserts that it fired, because a
+scenario whose interference silently did not happen passes for the wrong
+reason.
 
 **A synthetic stand-in for the other side, where that artefact *is* the
 contract.** `publish-snapshot-selftest.sh` scenarios 6 to 8 hold a
@@ -487,6 +720,15 @@ how both `[ "$cp_rc" -eq 0 ]` and `[ "$i_before" != missing ]` sat unpinned
 (issue #5) while looking well covered. Isolating a check means constructing
 the state only it can see, not the state that trips several at once.
 
+Scenario 11 applies the same discipline to a witness outside the clone
+entirely: not which of several checks inside `hardlink_clone_into` caught a
+fault, but whether `reader_lock_acquire`'s marker was observed by anything
+outside it at all. The consumer's own log and exit status are silent either
+way — a run with the marker deleted still succeeds — so what is asserted is
+one line in the *publisher's* log reporting that it waited. Deleting
+`reader_lock_acquire` (issue #10) leaves that line unwritten without failing
+anything else in the suite.
+
 The action YAML holds no logic beyond wiring; everything testable lives in
 `scripts/`. A composite action needs `shell: bash` on every `run:` step, and
 the actions reach their shared scripts through

cargo-cache-publish/action.yml

@@ -10,6 +10,15 @@ inputs:
     description: 'Mount point of the persistent cache volume. Must match the consume action.'
     required: false
     default: '/cache'
+  cache-lineage:
+    description: >-
+      Must match the cargo-cache step in this job. Checked rather than
+      assumed: this action derives both ends of the snapshot swap from its own
+      cache-root, so a publish step left at the default while its consume step
+      nested would republish a DIFFERENT lineage's live target dir over that
+      lineage's snapshot, on every push, silently. A mismatch fails the step.
+    required: false
+    default: ''
   protected-branches:
     description: >-
       Space-separated refs that publish snapshots. A run whose own ref is not
@@ -67,6 +76,11 @@ runs:
     # Everything here reads the environment the consume action exported, so a
     # workflow that forgets to run cargo-cache first fails loudly here rather
     # than silently publishing a snapshot of the wrong directory.
+    #
+    # The cache root is taken from that same environment for the same reason,
+    # and this action's own cache-root/cache-lineage inputs are checked against
+    # it rather than used. The two have always had to agree; until a lineage
+    # existed they always did, because nobody overrode the default.
     - id: resolve
       shell: bash
       env:
@@ -92,6 +106,18 @@ runs:
         : "${CARGO_CACHE_KEY:?cargo-cache-publish: CARGO_CACHE_KEY not set by the cargo-cache action}"
         echo "active=yes" >> "$GITHUB_OUTPUT"
 
+        # Only in publish mode. The `release-lock` call is an `if: always()`
+        # step that consuming workflows invoke with `mode:` and nothing else —
+        # it releases $CARGO_CACHE_LOCK_ID on $CARGO_TARGET_DIR and never
+        # touches a cache root at all, so holding its default inputs to the
+        # consume step's would fail the cleanup step of every job that sets a
+        # lineage, for a value it does not use.
+        if [ "${{ inputs.mode }}" = "publish" ]; then
+          : "${CARGO_CACHE_ROOT:?cargo-cache-publish: CARGO_CACHE_ROOT not set by the cargo-cache action}"
+          bash "${CARGO_CACHE_SCRIPTS}/cache-root.sh" verify \
+            "${{ inputs.cache-root }}" "${{ inputs.cache-lineage }}" "$CARGO_CACHE_ROOT"
+        fi
+
         OWN_REF="${{ inputs.own-ref }}"
         [ -n "$OWN_REF" ] || OWN_REF="${{ github.head_ref || github.ref_name }}"
 
@@ -131,7 +157,7 @@ runs:
       run: |
         set -euo pipefail
         bash "${CARGO_CACHE_SCRIPTS}/publish-snapshot.sh" \
-          "$CARGO_CACHE_KEY" "${{ inputs.cache-root }}" \
+          "$CARGO_CACHE_KEY" "$CARGO_CACHE_ROOT" \
           "${{ github.job }}-${{ github.run_id }}-$$"
 
     # Released in both modes. In `publish` mode this is the normal end-of-job

cargo-cache/action.yml

@@ -10,6 +10,18 @@ inputs:
     description: 'Mount point of the persistent cache volume inside the job container.'
     required: false
     default: '/cache'
+  cache-lineage:
+    description: >-
+      Distinguishes two jobs that build the SAME ref for different targets or
+      profiles (a host build and a wasm32 build, say) and would otherwise
+      resolve to one CARGO_TARGET_DIR and serialise on Cargo's exclusive
+      build-directory lock. Names one directory level under cache-root:
+      <cache-root>/<lineage>/target-<key>. Must be a single path component;
+      several names are refused outright because a reader elsewhere would stop
+      seeing the caches under them (see validate_cache_lineage in
+      scripts/cache-lib.sh). Pass the same value to cargo-cache-publish.
+    required: false
+    default: ''
   protected-branches:
     description: >-
       Space-separated refs that publish snapshots and are never evicted.
@@ -54,8 +66,9 @@ inputs:
   watermark-file:
     description: >-
       Name of this job's build-watermark file inside the target dir. MUST be
-      distinct per job when two jobs share one cache key. Defaults to
+      distinct per job when two jobs share one target directory — which two
-      .ci-watermark-<job>-sha.
+      jobs no longer need to do; cache-lineage gives them separate ones.
+      Defaults to .ci-watermark-<job>-sha, already distinct per job.
     required: false
     default: ''
   lock-id:
@@ -74,6 +87,12 @@ outputs:
   cache-key:
     description: 'Sanitized cache key for this run''s own ref.'
     value: ${{ steps.resolve.outputs.cache-key }}
+  cache-root:
+    description: >-
+      Resolved cache root — cache-root, plus the lineage directory when one is
+      set. Every directory this action reads or writes is under it. Also
+      exported as CARGO_CACHE_ROOT.
+    value: ${{ steps.resolve.outputs.cache-root }}
   seeded-from:
     description: 'Where the target dir came from: own | base-snapshot | own-snapshot | fallback-dir | concurrent-peer | cold.'
     value: ${{ steps.seed.outputs.seeded-from }}
@@ -95,6 +114,16 @@ runs:
     # `base_ref` is populated only for pull_request events. A push run has
     # nothing to layer over: its own ref IS the reference branch. It
     # publishes, it does not consume.
+    #
+    # The cache root is resolved first because every path below hangs off it.
+    # A lineage nests one directory level (`<root>/<lineage>/target-<key>`),
+    # which is what lets two jobs on ONE ref hold two build directories and so
+    # not serialise on Cargo's exclusive lock. It is resolved through
+    # cache-root.sh rather than interpolated here so the name is validated —
+    # several otherwise-reasonable lineage names put their whole subtree out of
+    # reach of a pass that has to see it. Everything downstream reads the
+    # resolved value, and it is exported as CARGO_CACHE_ROOT so the publish
+    # action can check it agrees with its own inputs.
     - id: resolve
       shell: bash
       run: |
@@ -102,6 +131,8 @@ runs:
         SCRIPTS=$(cd "${{ github.action_path }}/.." && pwd)/scripts
         [ -d "$SCRIPTS" ] || { echo "::error::cargo-cache: scripts/ not found at $SCRIPTS"; exit 1; }
         echo "CARGO_CACHE_SCRIPTS=${SCRIPTS}" >> "$GITHUB_ENV"
+        CACHE_ROOT=$(bash "${SCRIPTS}/cache-root.sh" resolve \
+          "${{ inputs.cache-root }}" "${{ inputs.cache-lineage }}")
         OWN_REF="${{ inputs.own-ref }}"
         [ -n "$OWN_REF" ] || OWN_REF="${{ github.head_ref || github.ref_name }}"
         BASE_REF="${{ inputs.base-ref }}"
@@ -116,7 +147,7 @@ runs:
           echo "cache: own ref '${OWN_REF}' -> ${OWN_KEY} (no base ref — this ref publishes, it does not consume)"
         fi
 
-        TARGET_DIR="${{ inputs.cache-root }}/target-${OWN_KEY}"
+        TARGET_DIR="${CACHE_ROOT}/target-${OWN_KEY}"
         WATERMARK="${{ inputs.watermark-file }}"
         [ -n "$WATERMARK" ] || WATERMARK=".ci-watermark-${{ github.job }}-sha"
         LOCK_ID="${{ inputs.lock-id }}"
@@ -128,10 +159,11 @@ runs:
           echo "base-key=${BASE_KEY}"
           echo "lock-id=${LOCK_ID}"
           echo "watermark-file=${WATERMARK}"
+          echo "cache-root=${CACHE_ROOT}"
         } >> "$GITHUB_OUTPUT"
         {
           echo "CARGO_TARGET_DIR=${TARGET_DIR}"
-          echo "CARGO_CACHE_ROOT=${{ inputs.cache-root }}"
+          echo "CARGO_CACHE_ROOT=${CACHE_ROOT}"
           echo "CARGO_CACHE_KEY=${OWN_KEY}"
           echo "CARGO_CACHE_LOCK_ID=${LOCK_ID}"
           echo "CI_WATERMARK_FILE=${WATERMARK}"
@@ -155,7 +187,7 @@ runs:
         bash "${SCRIPTS}/seed-target-dir.sh" \
           "${{ steps.resolve.outputs.cache-key }}" \
           "${{ steps.resolve.outputs.base-key }}" \
-          "${{ inputs.cache-root }}" \
+          "${{ steps.resolve.outputs.cache-root }}" \
           "${{ github.job }}-${{ github.run_id }}-$$" \
           "${{ inputs.seed-fallback-dir }}" \
           "${{ steps.resolve.outputs.lock-id }}"
@@ -165,6 +197,14 @@ runs:
     # touched unconditionally every run: a run that hits the cache for every
     # crate may write nothing at all inside the tree, which would make a
     # just-used directory look stale to the eviction pass.
+    #
+    # `.cache-last-used` IS A CROSS-REPO CONTRACT NAME with daniel/gitdan's
+    # `scripts/ci-cache-reclaim.sh`, which reads this exact marker for its own
+    # LRU ordering and to recognise a directory as a cache dir at all. Rename
+    # it here without a matching change there and that script silently falls
+    # back to directory mtime for both — see README.md's "Scratch names in a
+    # cache root" section and the `DEPENDED-UPON NAMES CONTRACT` block in
+    # gitdan's script.
     - shell: bash
       run: |
         set -euo pipefail
@@ -198,7 +238,7 @@ runs:
         set -euo pipefail
         SCRIPTS=$(cd "${{ github.action_path }}/.." && pwd)/scripts
         bash "${SCRIPTS}/prune-cache.sh" \
-          "${{ inputs.cache-root }}" \
+          "${{ steps.resolve.outputs.cache-root }}" \
           "${{ steps.resolve.outputs.target-dir }}" \
           "${{ inputs.protected-branches }}" \
           "${{ inputs.min-free-percent }}"

scripts/cache-lib.sh

@@ -114,6 +114,106 @@ cache_key() {
 target_dir_for() { printf '%s/target-%s' "$1" "$2"; }
 snapshot_dir_for() { printf '%s/snapshot-%s' "$1" "$2"; }
 
+# ---------------------------------------------------------------------------
+# Cache lineages
+# ---------------------------------------------------------------------------
+#
+# A cache key names a REF. What a target directory holds is the product of a
+# ref and a BUILD CONFIGURATION, and the two are not the same thing: emowheel
+# builds the same ref twice on every push, once for the host and once for
+# wasm32, in two jobs that run concurrently. Keyed on the ref alone both
+# resolve to one CARGO_TARGET_DIR, and Cargo's build-directory lock is
+# exclusive — so the second job sits on `Blocking waiting for file lock on
+# build directory` for the length of the first, holding a runner capacity slot
+# while doing nothing (daniel/gitdan#60).
+#
+# A lineage is that second dimension, and it is expressed as ONE DIRECTORY
+# LEVEL above the per-ref directories rather than as a suffix on the key:
+#
+#   <cache-root>/target-<key>              no lineage (the flat layout)
+#   <cache-root>/<lineage>/target-<key>    a lineage
+#
+# Nesting rather than suffixing is what keeps every existing reader correct
+# without teaching any of them a new name shape. `prune-cache.sh` resolves
+# liveness by recomputing `target-<cache_key(branch)>` for every branch on
+# origin and evicting whatever does not match — a suffixed `target-<key>-wasm32`
+# matches nothing, so it would be classified dead and unconditionally evicted
+# on every single run. The host-level arbiter in daniel/gitdan reads the same
+# shape (its BRANCH_DIR_RE), and a suffixed name falls out of it too: not
+# evicted there, but never a candidate either, so a whole lineage becomes
+# invisible to the global disk budget. Nesting leaves both matchers reading
+# exactly the names they already read, one directory deeper.
+#
+# ONE LEVEL, AND NOT TWO. The arbiter walks a volume to CI_CACHE_MAX_DEPTH,
+# which is 2 — `_data/target-<key>` and `_data/<lineage>/target-<key>`. It is
+# kept tight there on purpose (a deeper walk starts meeting Cargo's own
+# `incremental/<crate>-<hash>` directories, which match the same name shape and
+# must never be evicted individually), so a lineage is a single path component
+# and validate_cache_lineage refuses one containing a slash.
+
+# Directory names daniel/gitdan's ci-cache-reclaim.sh refuses to descend into
+# (its CI_CACHE_NODESCEND_NAMES). A lineage named one of these puts its whole
+# subtree outside the global arbiter's reach: the caches accumulate and the one
+# script whose job is the shared disk budget cannot see them.
+CACHE_LINEAGE_RESERVED_NAMES="debug release deps incremental build .fingerprint tmp examples doc"
+
+# The shape that same script reads as a per-branch cache directory (its
+# BRANCH_DIR_RE). A lineage matching it is taken for a cache dir in its own
+# right — never descended into, and an eviction candidate whole, which is the
+# entire lineage rather than one ref's share of it.
+CACHE_LINEAGE_BRANCH_DIR_RE='^.+-[0-9a-f]{7,40}$'
+
+# Every rejection below names the reader that imposes it, because that is the
+# only way the constraint survives: none of these is a filesystem limit, and a
+# name that trips one produces no error anywhere — it produces a lineage that
+# silently stops being pruned, or silently stops being reclaimed.
+validate_cache_lineage() {
+  local lineage="$1" reserved
+  [ -n "$lineage" ] || return 0
+
+  case "$lineage" in
+    */*)
+      echo "::error::cache lineage '${lineage}' must be a single path component: the host-level arbiter walks a cache volume to depth 2, so <cache-root>/<lineage>/target-<key> is as deep as a cache directory may sit and still be reclaimable" >&2
+      return 1
+      ;;
+    .*)
+      echo "::error::cache lineage '${lineage}' must not start with a dot: every dot-prefixed entry under a cache root belongs to the leftover-naming contract (see the top of this file), and a lineage is not garbage to be reclaimed" >&2
+      return 1
+      ;;
+    target-* | snapshot-*)
+      echo "::error::cache lineage '${lineage}' must not start with 'target-' or 'snapshot-': prune-cache.sh globs both prefixes at the cache root, so the lineage directory itself would become an eviction candidate" >&2
+      return 1
+      ;;
+    *[!A-Za-z0-9._-]*)
+      echo "::error::cache lineage '${lineage}' may contain only [A-Za-z0-9._-] — the same charset cache_key() sanitises a ref down to" >&2
+      return 1
+      ;;
+  esac
+
+  for reserved in $CACHE_LINEAGE_RESERVED_NAMES; do
+    if [ "$lineage" = "$reserved" ]; then
+      echo "::error::cache lineage '${lineage}' is one of the Cargo directory names daniel/gitdan's ci-cache-reclaim.sh never descends into (CI_CACHE_NODESCEND_NAMES) — every cache under it would be invisible to the host-level disk budget" >&2
+      return 1
+    fi
+  done
+
+  if [[ $lineage =~ $CACHE_LINEAGE_BRANCH_DIR_RE ]]; then
+    echo "::error::cache lineage '${lineage}' ends in a hex suffix, which is the shape daniel/gitdan's ci-cache-reclaim.sh reads as a per-branch cache directory — it would treat the lineage directory as one cache and evict the whole thing" >&2
+    return 1
+  fi
+
+  return 0
+}
+
+# The cache root a lineage's directories actually live under. An empty lineage
+# resolves to the cache root unchanged, byte for byte: that is what makes this
+# a no-op for every consumer that does not set one, rather than a migration.
+cache_root_for() {
+  local root="$1" lineage="${2:-}"
+  validate_cache_lineage "$lineage" || return 1
+  printf '%s%s' "$root" "${lineage:+/$lineage}"
+}
+
 # ---------------------------------------------------------------------------
 # Disk accounting
 # ---------------------------------------------------------------------------
@@ -210,33 +310,123 @@ _unshare_files() {
   # The inner shell propagates a failure of any individual copy-and-rename out
   # through xargs (which exits 123 if any invocation exits 1-125), so a
   # partially-unshared tree is reported rather than silently accepted.
+  # shellcheck disable=SC2016  # the quoted program is for the INNER shell: $f
+  # is its loop variable, $rc its accumulator and $$ its pid. Expanding any of
+  # them here is what the single quotes exist to prevent.
   find "$@" -links +1 -print0 2>/dev/null |
     xargs -0 -r -n 64 bash -c 'rc=0; for f; do cp -p -- "$f" "$f.unshare.$$" && mv -f -- "$f.unshare.$$" "$f" || rc=1; done; exit $rc' _
 }
 
+# True when a directory holds a compiled library artifact of its own.
+#
+# The glob is left unquoted and unmatched-glob-safe on purpose: with nullglob
+# off an unmatched pattern stays literal and the `-e` test fails, which is the
+# answer wanted.
+_holds_compiled_artifact() {
+  local f
+  for f in "$1"/*.rlib "$1"/*.rmeta; do
+    [ -e "$f" ] && return 0
+  done
+  return 1
+}
+
+# The directories `unshare_mutable_paths` replaces, under either layout.
+#
+# All four names are pruned, so nothing selected here can contain anything else
+# selected here and the caller never unshares a subtree twice.
+#
+# `out` is the one that needs deciding rather than naming, and it is the whole
+# difficulty of layout v2: a compile unit's rlib and a build script's OUT_DIR
+# are both a directory called `out`, one directory apart, and they need
+# opposite treatment.
+#
+# AMBIGUITY RESOLVES TOWARD UNSHARING, and that direction is the rule rather
+# than a default: over-unsharing costs bytes, under-unsharing costs corruption.
+# So an `out` directory is left shared only when TWO independent signals agree
+# it is a compile unit's artifact directory, and either one missing is enough
+# to real-copy it:
+#
+#   1. it holds an `.rlib`/`.rmeta` of its own — the artifact whose sharing is
+#      the entire point of the clone; and
+#   2. its unit directory has no record of a build-script execution beside it
+#      (`run/` under layout v2, a loose `root-output` under v1).
+#
+# Signal 2 alone was the first cut of this and it is NOT sufficient, because
+# Cargo writes `root-output` only AFTER the script exits successfully. A build
+# script that populates `OUT_DIR` and then FAILS leaves a unit with no record
+# at all, which reads as "compile unit" — and the old `-name build` selection
+# covered that state by real-copying `build/` wholesale, so trusting signal 2
+# alone was a regression against it. Reproduced on cargo 1.93.1 stable: the
+# clone's build wrote through the shared inode into the source's OUT_DIR.
+# Signal 1 closes it, because a failed build script's OUT_DIR holds no rlib.
+#
+# The residual is a build script that writes a file NAMED `*.rlib`/`*.rmeta`
+# into `OUT_DIR` and has never once succeeded. Nothing bounds that away; it is
+# simply far narrower than what it replaces.
+#
+# Requiring signal 1 also means a bin, test or build-script COMPILE unit's
+# `out` is real-copied rather than shared — at no cost in bytes, since
+# everything in one is an executable or a `*.d`, and both are privately owned
+# by the file rules below either way.
+_mutable_dirs() {
+  local root="$1" d unit
+  while IFS= read -r d; do
+    if [ "${d##*/}" = out ]; then
+      unit="${d%/out}"
+      if ! [ -d "$unit/run" ] && ! [ -e "$unit/root-output" ] \
+         && _holds_compiled_artifact "$d"; then
+        continue
+      fi
+    fi
+    printf '%s\n' "$d"
+  done < <(find "$root" -type d \
+    \( -name .fingerprint -o -name fingerprint -o -name run -o -name out \) \
+    -prune -print 2>/dev/null)
+}
+
 # THE load-bearing function of this whole design.
 #
 # A hardlink clone is only safe if every write the clone's build performs
-# lands on a NEW inode, leaving the source's data untouched. That is true for
+# lands on a NEW inode, leaving the source's data untouched. That is true of
-# compilation artifacts — rustc and the linker replace `deps/*.rlib`,
+# rustc's own outputs — it writes an `.rlib` or `.rmeta` to a temporary and
-# `*.rmeta`, and binaries rather than truncating them in place — and it is
+# renames it into place — and it is NOT true of the metadata Cargo and build
-# NOT true for the metadata Cargo and build scripts write with a plain
+# scripts write with a plain truncating write, nor of anything the LINKER
-# truncating write. Measured directly (Linux, ext4, cargo 1.9x nightly:
+# produces. Measured directly (Linux, ext4: `cp -al` a warm target dir, change
-# `cp -al` a warm target dir, change a source file, build in the clone, diff
+# a source file, build in the clone, diff the source) the following files in
-# the source) the following files in the SOURCE were mutated through the
+# the SOURCE were mutated through the shared inode:
-# shared inode:
 #
-#   <profile>/.fingerprint/<unit>/dep-<target>   (only under
+#   <profile>/.fingerprint/<unit>/dep-<target>   (build-dir layout v1) — or,
-#                                                CARGO_UNSTABLE_CHECKSUM_FRESHNESS,
+#   <profile>/build/<pkg>/<hash>/fingerprint/dep-<target>
-#                                                where this file carries the
+#                                               (build-dir layout v2; see the
-#                                                per-source blake3 checksums)
+#                                                dated note below for which
-#   <profile>/build/<pkg>/output, root-output   (Cargo build-script metadata)
+#                                                Cargo writes which). Only
+#                                                when Cargo resolves freshness
+#                                                by CONTENT, where this file
+#                                                carries the per-source blake3
+#                                                checksums.
+#   <profile>/build/<pkg>/output, root-output   (Cargo build-script metadata;
+#                                                `<pkg>/<hash>/run/root-output`
+#                                                under layout v2)
 #   <profile>/build/<pkg>/out/**                (whatever the build script
 #                                                writes into OUT_DIR — build
 #                                                scripts overwhelmingly use a
 #                                                plain fs::write)
 #   <profile>/deps/*.d, <profile>/*.d           (Cargo's post-processed
 #                                                dep-info)
+#   <profile>/deps/<test>-<hash>                (a linked TEST binary; under
+#                                                layout v2,
+#                                                `build/<pkg>/<hash>/out/`)
+#
+# THE LINKED-OUTPUT CASE IS NOT LAYOUT-SPECIFIC AND WAS NOT PART OF THIS
+# FUNCTION UNTIL 2026-08-27 (gitdan-actions#14). A `cargo test --no-run` inside
+# a raw `cp -al` clone rewrote the source's own test binary in place on cargo
+# 1.93.1 stable, 1.96.0-nightly, 1.98.0-nightly (layout v1) and 1.100.0-nightly
+# (layout v2) alike. Cargo re-creates the path first when it also has to uplift
+# the result — a bin target's `deps/<bin>-<hash>` has a hardlink twin at
+# `<profile>/<bin>` — and other crate shapes relinked to a fresh inode for
+# reasons this measurement did not pin down. Since the safe cases could not be
+# enumerated, every executable is treated as mutable; `.rlib`, `.rmeta` and
+# `incremental/` are what stay shared, and they are the bytes worth sharing.
 #
 # The checksum-freshness case is not a cosmetic one. Reproduced end to end:
 # branch B clones base's cache, builds its own content, and thereby rewrites
@@ -246,10 +436,54 @@ _unshare_files() {
 # sources, reports `Fresh`, and reuses a binary built from the PRE-merge code.
 # That is silent stale-artifact reuse — a wrong answer, not a slow one.
 #
-# So: hardlink the artifacts (the GB), real-copy the metadata (the MB).
+# WHAT UPSTREAM CHANGED, AND WHAT IT DID NOT (measured 2026-08-26, daniel/gitdan#62).
-# Measured on a 6.9 GB Bevy workspace target dir, the unshared set is
+#
-# .fingerprint 22 MB + build/ 237 MB + a handful of dep-info files — about
+# An earlier revision of this comment recorded that the dep-info write was
-# 3.7% of the tree, against 100% for a plain `cp -a`.
+# "NOT reproduced on 1.100.0-nightly (2026-08-25) — upstream appears to have
+# stopped writing it in place". That reading was wrong, and the way it was
+# wrong is the reason this paragraph is dated. Two unrelated upstream changes
+# landed within days of each other, and between them they moved both the
+# switch that turns the behaviour on and the path it writes to:
+#
+#   1. The ON-SWITCH MOVED. cargo PR #17382 `feat(config): Add build.fingerprint`
+#      (merged 2026-08-22) demoted `-Z checksum-freshness` to a gate: it now
+#      only UNLOCKS the feature, and `build.fingerprint` SELECTS it, defaulting
+#      to `"mtime"`. So `CARGO_UNSTABLE_CHECKSUM_FRESHNESS=true` on its own is
+#      accepted and does nothing, which is exactly the "flag accepted, mtime
+#      anyway" result that was mistaken for a withdrawal. Content freshness
+#      needs BOTH, and with both it is entirely intact:
+#
+#          CARGO_UNSTABLE_CHECKSUM_FRESHNESS=true CARGO_BUILD_FINGERPRINT=content
+#
+#      Measured on cargo 1.100.0-nightly (e8cb624d5 2026-08-22): with the gate
+#      alone a `cp -al` clone mutates only the build/ and *.d families; add
+#      `CARGO_BUILD_FINGERPRINT=content` and the source's dep-info file is
+#      mutated through the shared inode again. Same toolchain, same clone, one
+#      env var apart. The hazard was never removed — it was switched off.
+#
+#   2. THE PATH MOVED. Build-dir layout v2 (`-Z build-dir-new-layout`, cargo
+#      1.91) became the nightly default in cargo 1.99 (PR #17258) and was
+#      stabilised by PR #17354, merged 2026-08-18, shipping in cargo 1.100.0
+#      stable on 2026-11-12. Under v2 there is no `<profile>/.fingerprint` and
+#      no `<profile>/deps` at all: everything is regrouped per build unit under
+#      `<profile>/build/<pkg>/<hash>/{fingerprint,out,run}/`, artifacts
+#      included. Bracketed locally: cargo 1.97.1 and 1.98.0-nightly write v1,
+#      1.100.0-nightly writes v2.
+#
+# Until 2026-08-27 the selection was `-name .fingerprint -o -name build`,
+# which under a v2 Cargo matched nothing on its first clause and the entire
+# tree on its second, because the artifacts moved under `build/` too. The guard
+# held by accident and the saving did not: on one scratch crate (serde +
+# serde_json + regex plus a build script), same sources both ways —
+#
+#     cargo 1.98.0-nightly (layout v1)   64.9 MB unshared of 165.1 MB — 39.3%
+#          1.100.0-nightly (layout v2)  110.4 MB unshared of 110.4 MB — 99.996%
+#
+# The selection now names the mutable set directly rather than by the container
+# it used to live in, so it holds under both layouts; the same crate measures
+# 45.3% (v1) and 38.4% (v2), both dominated by the linked-output rule above
+# rather than by the layout. On a real 5.5 GB Bevy target dir the whole change
+# moves the real-copied share from 9.0% to 14.1%.
 #
 # `incremental/` is deliberately left shared: rustc writes each incremental
 # session to a fresh `s-*-working` directory and finalises it with a rename,
@@ -261,10 +495,9 @@ unshare_mutable_paths() {
   [ -d "$root" ] || return 0
   # The list is materialised in full before anything is replaced: each
   # replacement deletes and recreates a directory, and a live `find` walk over
-  # a tree being mutated underneath it is a needless hazard. `-prune` keeps a
+  # a tree being mutated underneath it is a needless hazard.
-  # match's own contents out of the list.
   local -a dirs=()
-  mapfile -t dirs < <(find "$root" -type d \( -name .fingerprint -o -name build \) -prune -print 2>/dev/null)
+  mapfile -t dirs < <(_mutable_dirs "$root")
   for d in "${dirs[@]}"; do
     [ -n "$d" ] || continue
     unshare_subtree "$d" || {
@@ -276,6 +509,37 @@ unshare_mutable_paths() {
     echo "::error::unshare_mutable_paths: failed to unshare dep-info files under ${root}" >&2
     return 1
   }
+  # Layout v1's build-script run metadata, which v2 groups under `run/` and v1
+  # leaves loose in the run unit's directory. `invoked.timestamp` is empty and
+  # carries its meaning in its mtime, which a shared inode carries too.
+  _unshare_files "$root" -type f \
+    \( -name output -o -name root-output -o -name stderr -o -name invoked.timestamp \) || {
+    echo "::error::unshare_mutable_paths: failed to unshare build-script run metadata under ${root}" >&2
+    return 1
+  }
+  # Linked outputs. Unlike an rlib or an rmeta — which rustc writes to a
+  # temporary and renames into place — an executable or shared object is
+  # written by the LINKER, and the linker writes THROUGH an existing inode.
+  # Measured 2026-08-27 on cargo 1.93.1 stable, 1.96.0-nightly, 1.98.0-nightly
+  # (layout v1) and 1.100.0-nightly (e8cb624d5, layout v2), mold and the
+  # default linker alike: a `cargo test --no-run` binary in a `cp -al` clone
+  # rewrote the SOURCE's copy of itself in place, under both layouts.
+  #
+  # What separates that from the executables measured INTACT is not
+  # established. Every intact case observed was one Cargo has to re-create
+  # anyway to maintain an uplift hardlink — a bin target's
+  # `deps/<bin>-<hash>`, twinned at `<profile>/<bin>`. Whether the twin is the
+  # mechanism or a correlate of it was not determined, and the rule below does
+  # not depend on the answer: exempting twinned executables would recover no
+  # bytes this function newly copies. See gitdan-actions#17.
+  #
+  # The executable bit is the discriminator because it is the linker's own
+  # output that is at risk, not the directory it happens to land in — `.rlib`,
+  # `.rmeta` and `incremental/` stay shared and they are the bytes that matter.
+  _unshare_files "$root" -type f -perm -u+x || {
+    echo "::error::unshare_mutable_paths: failed to unshare linked outputs under ${root}" >&2
+    return 1
+  }
   _unshare_files "$root" -maxdepth 3 -type f -name '.rustc_info.json' || {
     echo "::error::unshare_mutable_paths: failed to unshare .rustc_info.json under ${root}" >&2
     return 1

scripts/cache-lock.sh

@@ -14,6 +14,19 @@
 # until STALE_LOCK_SECONDS, after which it is treated as abandoned by a job
 # the runner killed before it reached its own release step. Honouring a lock
 # forever would let one crashed job pin a directory permanently.
+#
+# `.ci-lock-<id>` IS A CROSS-REPO CONTRACT NAME with daniel/gitdan's
+# `scripts/ci-cache-reclaim.sh`: that host-level arbiter reads this same
+# marker (its `leftover_is_locked()`) to decide whether a leftover it is about
+# to reclaim is still held open by a live job. `cache-lib.sh`'s
+# `write_cache_lock()` writes the identical shape independently of this
+# script. Rename or reshape this marker here without a matching change on
+# gitdan's side and the arbiter's lock check silently stops matching — a live
+# job's staging tree loses its liveness guard and becomes an ordinary reclaim
+# candidate while still in use. See the "Scratch names in a cache root" section
+# in README.md, and daniel/gitdan's `DEPENDED-UPON NAMES CONTRACT` block in
+# `scripts/ci-cache-reclaim.sh`, for the full arrangement — this repo's half
+# is: don't rename `.ci-lock-` without telling that script.
 set -euo pipefail
 
 MODE="${1:?usage: cache-lock.sh acquire|release <dir> <lock-id>}"

scripts/cache-root-selftest.sh

@@ -0,0 +1,194 @@
+#!/usr/bin/env bash
+# Regression test for cache-root.sh and the lineage rules in cache-lib.sh —
+# the fix for daniel/gitdan#60, where two jobs building the same ref for
+# different targets resolved to one CARGO_TARGET_DIR and serialised on Cargo's
+# exclusive build-directory lock.
+#
+#   1. NO LINEAGE CHANGES NOTHING — the effective root is the cache root byte
+#      for byte, so every consumer that does not set a lineage keeps the exact
+#      directories it already has on the volume. This is the whole of the
+#      migration story for lublub, zemyna and emowheel's `ci` job, so it is
+#      asserted rather than assumed.
+#   2. TWO LINEAGES, ONE REF, TWO TARGET DIRS — the bug itself. The two jobs
+#      keep one cache key (they are the same ref) and still get directories
+#      that are neither equal nor nested one inside the other, which is what
+#      Cargo's per-directory lock needs in order not to serialise them.
+#   3. THE WHOLE PIPELINE MOVES TOGETHER — seed, publish and prune all operate
+#      inside the lineage root. A pass run in one lineage must not evict, or
+#      even see, a sibling lineage's caches or the flat layout's.
+#   4. BASE SEEDING IS PER LINEAGE — a PR branch layers over ITS OWN lineage's
+#      base snapshot, not over whatever the flat root happens to hold. This is
+#      the property that keeps a warm start for both jobs rather than one.
+#   5. A NAME NO READER CAN HANDLE IS REFUSED AT RESOLVE TIME — one assertion
+#      per constraint, each named for the reader that imposes it. None of these
+#      is a filesystem limit: every one of them produces a working directory
+#      that some pass silently stops seeing, which is the failure mode this
+#      whole scheme exists to avoid rather than to relocate.
+#   6. A PUBLISH THAT DISAGREES WITH ITS CONSUME STEP FAILS LOUDLY — the
+#      footgun the lineage input introduces. cargo-cache-publish derives both
+#      ends of the snapshot swap from its own `cache-root`, so a publish step
+#      left at the default while its consume step nested would republish the
+#      OTHER lineage's live target dir over that lineage's snapshot, silently.
+set -euo pipefail
+script_dir=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+. "$script_dir/cache-lib.sh"
+
+scratch=$(mktemp -d)
+trap 'rm -rf "$scratch"' EXIT
+root="$scratch/cache"; mkdir -p "$root"
+pass_count=0
+fail() { echo "ASSERTION FAILED: $*" >&2; exit 1; }
+ok() { pass_count=$((pass_count + 1)); echo "PASS: $*"; }
+
+resolve() { bash "$script_dir/cache-root.sh" resolve "$@"; }
+
+# A tree that looks enough like a Cargo target dir for the pipeline scripts,
+# with a marker naming which lineage produced it — scenario 4 turns on reading
+# that marker back out of a seeded directory.
+make_tree() {
+  local d="$1" marker="$2"
+  mkdir -p "$d/debug/deps" "$d/debug/.fingerprint/x"
+  echo "$marker" > "$d/debug/deps/libx.rlib"
+  echo "$marker" > "$d/lineage-marker"
+  echo "$marker" > "$d/debug/.fingerprint/x/dep-lib-x"
+}
+
+echo "=== 1. no lineage changes nothing ==="
+
+[ "$(resolve /cache)" = /cache ] || fail "an omitted lineage changed the root"
+[ "$(resolve /cache '')" = /cache ] || fail "an empty lineage changed the root"
+ok "no lineage resolves to the cache root unchanged"
+
+KEY=$(cache_key dev)
+[ "$(target_dir_for "$(resolve /cache '')" "$KEY")" = "/cache/target-$KEY" ] \
+  || fail "the flat target-dir name moved"
+[ "$(snapshot_dir_for "$(resolve /cache '')" "$KEY")" = "/cache/snapshot-$KEY" ] \
+  || fail "the flat snapshot name moved"
+ok "the flat layout's directory names are untouched"
+
+echo
+echo "=== 2. two lineages, one ref, two target dirs ==="
+
+HOST_ROOT=$(resolve /cache)
+WASM_ROOT=$(resolve /cache wasm32)
+[ "$WASM_ROOT" = /cache/wasm32 ] || fail "lineage root resolved to '$WASM_ROOT'"
+HOST_DIR=$(target_dir_for "$HOST_ROOT" "$KEY")
+WASM_DIR=$(target_dir_for "$WASM_ROOT" "$KEY")
+[ "$HOST_DIR" != "$WASM_DIR" ] || fail "both lineages resolved to $HOST_DIR"
+case "$WASM_DIR" in "$HOST_DIR"/*) fail "the wasm target dir sits inside the host one" ;; esac
+case "$HOST_DIR" in "$WASM_DIR"/*) fail "the host target dir sits inside the wasm one" ;; esac
+ok "one cache key ($KEY), two disjoint target dirs: $HOST_DIR and $WASM_DIR"
+
+echo
+echo "=== 3. the whole pipeline moves together ==="
+
+DEAD=$(cache_key feat/dead)
+lin="$root/wasm32"
+mkdir -p "$lin"
+make_tree "$root/target-$DEAD" flat
+make_tree "$root/native/target-$DEAD" native
+make_tree "$lin/target-$DEAD" wasm32
+bash "$script_dir/seed-target-dir.sh" "$KEY" "" "$lin" tag-3 > "$scratch/seed3.log" 2>&1 \
+  || { cat "$scratch/seed3.log"; fail "seed inside a lineage root failed"; }
+[ -d "$lin/target-$KEY" ] || fail "seed did not create $lin/target-$KEY"
+[ -d "$root/target-$KEY" ] && fail "seed created a directory at the flat root as well"
+ok "seed creates its directory under the lineage root and nowhere else"
+
+make_tree "$lin/target-$KEY" wasm32
+bash "$script_dir/publish-snapshot.sh" "$KEY" "$lin" tag-3 > "$scratch/pub3.log" 2>&1 \
+  || { cat "$scratch/pub3.log"; fail "publish inside a lineage root failed"; }
+[ -d "$lin/snapshot-$KEY" ] || fail "publish did not create $lin/snapshot-$KEY"
+[ -d "$root/snapshot-$KEY" ] && fail "publish created a snapshot at the flat root as well"
+ok "publish writes its snapshot under the lineage root and nowhere else"
+
+# Free space far below the threshold, so pass 2 evicts every eligible
+# directory it can see. What it can see is the point of the scenario.
+CACHE_LIVENESS=false CACHE_DF_OVERRIDE="1000000 1000" \
+  bash "$script_dir/prune-cache.sh" "$lin" "$lin/target-$KEY" 'dev main' 10 \
+  > "$scratch/prune3.log" 2>&1 || { cat "$scratch/prune3.log"; fail "prune inside a lineage root failed"; }
+[ -d "$lin/target-$DEAD" ] && { cat "$scratch/prune3.log"; fail "prune left its own lineage's evictable cache in place"; }
+[ -d "$root/target-$DEAD" ] || fail "prune reached out of its lineage and evicted the flat root's cache"
+[ -d "$root/native/target-$DEAD" ] || fail "prune reached into a sibling lineage and evicted its cache"
+ok "prune under disk pressure evicts inside its own lineage only"
+
+echo
+echo "=== 4. base seeding is per lineage ==="
+
+BASE=$(cache_key dev)
+PR=$(cache_key feat/pr)
+rm -rf "$lin" "$root/snapshot-$BASE"
+mkdir -p "$lin"
+make_tree "$root/snapshot-$BASE" flat-base
+make_tree "$lin/snapshot-$BASE" wasm32-base
+bash "$script_dir/seed-target-dir.sh" "$PR" "$BASE" "$lin" tag-4 > "$scratch/seed4.log" 2>&1 \
+  || { cat "$scratch/seed4.log"; fail "seeding a PR branch inside a lineage failed"; }
+[ -d "$lin/target-$PR" ] || fail "the PR branch's lineage target dir was not created"
+got=$(cat "$lin/target-$PR/lineage-marker")
+[ "$got" = wasm32-base ] || fail "the PR branch layered over '$got', not its own lineage's base snapshot"
+grep -q 'base snapshot' "$scratch/seed4.log" || { cat "$scratch/seed4.log"; fail "seed did not report a base-snapshot clone"; }
+ok "a PR branch layers over its own lineage's base snapshot ($got)"
+
+echo
+echo "=== 5. a name no reader can handle is refused ==="
+
+reject() {
+  local lineage="$1" want="$2" desc="$3" out
+  if out=$(resolve /cache "$lineage" 2>&1); then
+    fail "lineage '$lineage' was accepted (resolved to '$out') — $desc"
+  fi
+  case "$out" in
+    *"$want"*) ;;
+    *) fail "lineage '$lineage' was rejected without naming '$want': $out" ;;
+  esac
+  ok "rejected '$lineage' — $desc"
+}
+
+reject 'a/b' 'single path component' "the arbiter walks a volume to depth 2"
+reject '.hidden' 'dot' "every dot-prefixed name under a cache root is leftover-contract territory"
+reject 'target-x' 'prune-cache.sh' "prune-cache.sh globs target-* at the cache root"
+reject 'snapshot-x' 'prune-cache.sh' "prune-cache.sh globs snapshot-* at the cache root"
+reject 'wasm 32' 'A-Za-z0-9._-' "a cache key is sanitised to that charset and a lineage sits beside one"
+reject 'release' 'CI_CACHE_NODESCEND_NAMES' "the arbiter never descends into a Cargo profile name"
+reject 'doc' 'CI_CACHE_NODESCEND_NAMES' "same, for the docs profile directory"
+reject 'lineage-deadbeef' 'per-branch cache directory' "the arbiter reads a hex-suffixed name as one cache dir"
+
+for good in wasm32 web android host wasm32.release lineage_2; do
+  out=$(resolve /cache "$good") || fail "lineage '$good' was rejected: $out"
+  [ "$out" = "/cache/$good" ] || fail "lineage '$good' resolved to '$out'"
+done
+ok "ordinary lineage names still resolve"
+
+echo
+echo "=== 6. a publish that disagrees with its consume step fails loudly ==="
+
+verify() { bash "$script_dir/cache-root.sh" verify "$@"; }
+
+verify /cache wasm32 /cache/wasm32 > /dev/null 2>&1 \
+  || fail "verify rejected a publish step that agrees with its consume step"
+verify /cache '' /cache > /dev/null 2>&1 \
+  || fail "verify rejected an unmigrated consumer's matching default pair"
+ok "verify accepts a publish step whose inputs match what the consume step exported"
+
+# The exact shape of the mistake: the consume step nested, the publish step
+# kept the default. Left unchecked this republishes the host lineage's live
+# target dir over the host lineage's snapshot, from the wasm job.
+if out=$(verify /cache '' /cache/wasm32 2>&1); then
+  fail "verify accepted a publish step that resolved to /cache while the job exported /cache/wasm32"
+fi
+case "$out" in
+  *'/cache/wasm32'*) ;;
+  *) fail "the mismatch error does not quote what the consume step exported: $out" ;;
+esac
+case "$out" in
+  *'SAME cache-root and cache-lineage'*) ;;
+  *) fail "the mismatch error does not say what to do about it: $out" ;;
+esac
+ok "verify rejects a publish step that forgot the lineage, and says so"
+
+if verify /cache 'a/b' /cache/a/b > /dev/null 2>&1; then
+  fail "verify accepted an invalid lineage as long as both sides agreed on it"
+fi
+ok "verify validates the lineage as well as comparing it"
+
+echo
+echo "cache-root-selftest: ${pass_count} assertions passed"

scripts/cache-root.sh

@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+# Resolves — and cross-checks — the cache root a job's directories live under.
+#
+#   cache-root.sh resolve <cache-root> [lineage]
+#   cache-root.sh verify  <cache-root> <lineage> <exported-root>
+#
+# `resolve` prints the effective root: the cache root unchanged when no lineage
+# is given, or `<cache-root>/<lineage>` when one is. Invalid lineage names are
+# rejected here rather than downstream — see validate_cache_lineage() in
+# cache-lib.sh, where every rejection names the reader that imposes it.
+#
+# `verify` is the publish side's guard. cargo-cache-publish resolves the same
+# two inputs the consume action was given and compares the result against the
+# CARGO_CACHE_ROOT the consume step exported into the job environment. The two
+# actions have always had to agree — `cache-root`'s description in the publish
+# action says "must match the consume action" — and until a lineage existed
+# they always did, because nobody overrode the default. A disagreement is not
+# a harmless no-op: publish-snapshot.sh takes the root as an argument and
+# derives BOTH ends of the swap from it, so a publish step that kept the
+# default while its consume step nested would read `<root>/target-<key>` — the
+# OTHER lineage's live target dir — and republish it over `<root>/snapshot-<key>`,
+# which is that lineage's snapshot. Two jobs would then be publishing one
+# snapshot from one tree on every push, and nothing in either action would say
+# so. Hence: fail the job, loudly, rather than resolve the ambiguity in
+# either direction.
+#
+# A thin CLI over cache-lib.sh, kept as its own entry point for the same
+# reason branch-cache-key.sh is: an out-of-band job that needs to find a
+# lineage's directories should resolve the path the way the action does
+# instead of reimplementing the rule.
+set -euo pipefail
+. "$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)/cache-lib.sh"
+
+MODE="${1:-}"
+case "$MODE" in
+  resolve)
+    [ $# -ge 2 ] && [ $# -le 3 ] || {
+      echo "::error::cache-root.sh resolve: expected <cache-root> [lineage]" >&2
+      exit 1
+    }
+    [ -n "$2" ] || { echo "::error::cache-root.sh: cache-root must not be empty" >&2; exit 1; }
+    cache_root_for "$2" "${3:-}"
+    ;;
+  verify)
+    [ $# -eq 4 ] || {
+      echo "::error::cache-root.sh verify: expected <cache-root> <lineage> <exported-root>" >&2
+      exit 1
+    }
+    [ -n "$2" ] || { echo "::error::cache-root.sh: cache-root must not be empty" >&2; exit 1; }
+    expected=$(cache_root_for "$2" "$3")
+    if [ "$expected" != "$4" ]; then
+      echo "::error::cache-root.sh: this step resolves its cache root to '${expected}' (cache-root '$2', cache-lineage '$3') but the cargo-cache step in this job exported '$4'. Pass the SAME cache-root and cache-lineage to both actions." >&2
+      exit 1
+    fi
+    echo "cache root: ${expected} (agrees with the cargo-cache step in this job)"
+    ;;
+  *)
+    echo "::error::cache-root.sh: unknown mode '${MODE}' (expected resolve or verify)" >&2
+    exit 1
+    ;;
+esac

scripts/hardlink-clone-selftest.sh

@@ -5,8 +5,10 @@
 #
 # That assumption is FALSE for a plain `cp -al`. Measured, and asserted below
 # as an explicit control: build in a raw `cp -al` clone and the source's
-# `.fingerprint/<unit>/dep-*` (under CARGO_UNSTABLE_CHECKSUM_FRESHNESS),
+# dep-info file (`.fingerprint/<unit>/dep-*` under Cargo's build-dir layout
-# `build/<pkg>/output`, `build/<pkg>/out/**` and `deps/*.d` all change,
+# v1, `build/<pkg>/<hash>/fingerprint/dep-*` under v2 — and under content
+# freshness only, see the probe below), `build/<pkg>/output`,
+# `build/<pkg>/out/**` and `deps/*.d` all change,
 # because Cargo and build scripts write those with a plain truncating write
 # rather than the write-then-rename Cargo uses for real artifacts.
 #
@@ -26,8 +28,6 @@ set -euo pipefail
 script_dir=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 . "$script_dir/cache-lib.sh"
 
-command -v cargo >/dev/null || { echo "SKIP: no cargo on PATH"; exit 0; }
-
 scratch=$(mktemp -d)
 trap 'rm -rf "$scratch"' EXIT
 pass_count=0
@@ -57,6 +57,14 @@ version = "0.1.0"
 edition = "2021"
 [workspace]
 TOML
+  # A binary as well as a library, because the two are written differently and
+  # only one of them is safe to share: rustc writes an rlib to a temporary and
+  # renames it into place, while the LINKER writes an executable through the
+  # existing inode. Without a bin target this suite never relinks anything and
+  # cannot see that difference.
+  cat > "$dir/src/main.rs" <<'RS'
+fn main() { println!("{}", probe::f()); }
+RS
   cat > "$dir/build.rs" <<'RS'
 use std::{env, fs, path::PathBuf};
 fn main() {
@@ -68,27 +76,391 @@ fn main() {
 RS
 }
 
+
+# ---------------------------------------------------------------------------
+# Both build-dir layouts, without a compiler
+# ---------------------------------------------------------------------------
+#
+# Every other scenario in this file runs whichever layout the installed Cargo
+# happens to write, so on any one machine it exercises exactly ONE of the two —
+# and gitdan-ci's is v1. The two fixtures below reproduce both directory shapes
+# from files alone, clone them through the real `hardlink_clone_into()`, and
+# assert file by file which side of the partition each one lands on.
+#
+# Shapes taken from a scratch crate (serde + serde_json + regex, plus a build
+# script) built on 2026-08-27: cargo 1.98.0-nightly (a335d47ff 2026-06-26)
+# writes v1, cargo 1.100.0-nightly (e8cb624d5 2026-08-22) writes v2.
+#
+# v2 is where the partition is easy to get wrong, and the fixtures are built to
+# say so: a build script's OUT_DIR and a compile unit's rlib are BOTH a
+# directory called `out`, one directory apart, and they need opposite
+# treatment.
+mkfile() { mkdir -p "$(dirname "$1")"; printf '%s' "$2" > "$1"; }
+
+# Big enough that the byte-fraction assertion below measures something.
+artifact_bytes=$(head -c 4096 /dev/zero | tr '\0' 'A')
+
+# Spec lines are `<shared|private>|<path relative to the tree root>`.
+assert_partition() {
+  local label="$1" src="$2" spec="$3"
+  local clone="${src}-clone" want rel si ci total=0 copied=0 sz
+  hardlink_clone_into "$src" "$clone" "selftest-$label" \
+    || fail "$label: hardlink_clone_into refused the destination"
+  while IFS='|' read -r want rel; do
+    [ -n "${rel:-}" ] || continue
+    [ -e "$clone/$rel" ] || fail "$label: $rel is missing from the clone"
+    si=$(stat -c '%i' "$src/$rel"); ci=$(stat -c '%i' "$clone/$rel")
+    case "$want" in
+      shared)
+        [ "$si" = "$ci" ] \
+          || fail "$label: $rel was real-copied, but it is an artifact and must stay shared" ;;
+      private)
+        [ "$si" != "$ci" ] \
+          || fail "$label: $rel still shares an inode with the source, so a build in the clone can rewrite it" ;;
+      *) fail "$label: unknown spec verb '$want'" ;;
+    esac
+  done <<< "$spec"
+  ok "$label: every file landed on the right side of the partition"
+
+  # The cost model, asserted rather than assumed. Selecting too much is not a
+  # correctness bug, which is exactly why nothing caught layout v2 taking the
+  # selection from 39.3% of one scratch crate's tree to 99.996% of it
+  # (gitdan-actions#14):
+  # a hardlink clone that real-copies everything is a `cp -a` with extra steps.
+  # The bound is loose on purpose. It is not a budget — the honest figure moves
+  # with how much of a tree is linker output, and these fixtures are mostly
+  # that by construction — it is a floor under "still a hardlink clone at all".
+  while IFS= read -r rel; do
+    # `.cargo-*lock*` is stripped from every clone by design, so it has no
+    # counterpart to compare against.
+    [ -e "$clone/$rel" ] || continue
+    sz=$(stat -c '%s' "$src/$rel")
+    total=$((total + sz))
+    [ "$(stat -c '%i' "$src/$rel")" = "$(stat -c '%i' "$clone/$rel")" ] || copied=$((copied + sz))
+  done < <(cd "$src" && find . -type f -printf '%P\n')
+  [ "$total" -gt 0 ] || fail "$label: fixture has no bytes to measure"
+  [ $((copied * 100 / total)) -lt 90 ] \
+    || fail "$label: the clone real-copied $((copied * 100 / total))% of its source's bytes — the hardlink saving is gone"
+  ok "$label: clone real-copies $((copied * 100 / total))% of ${total} B (${copied} B), the rest is shared"
+}
+
+# Layout v2: no `.fingerprint`, no `deps`. Everything regroups per build unit
+# under `build/<pkg>/<hash>/{fingerprint,out,run}`, artifacts included — which
+# is what took `-name build` from "the metadata" to "the whole tree".
+#
+# THE THREE UNIT KINDS ARE THE POINT. `1bf...` is a compile unit: its `out`
+# holds the rlib. `08c...` is the build script's own compile unit: its `out`
+# holds the build-script binary. `da9...` is the build-script RUN unit: its
+# `out` IS the OUT_DIR, and it is the only one of the three whose `out` is
+# mutable. The `run/` directory beside it is the structural difference.
+v2="$scratch/layout-v2"
+mkfile "$v2/CACHEDIR.TAG" 'Signature: 8a477f597d28d172'
+mkfile "$v2/.rustc_info.json" '{"rustc_fingerprint":1}'
+mkfile "$v2/debug/.cargo-lock" ''
+mkfile "$v2/debug/libprobe.rlib" "$artifact_bytes"
+mkfile "$v2/debug/libprobe.d" '/probe/src/lib.rs:'
+for f in dep-lib-probe lib-probe lib-probe.json invoked.timestamp; do
+  mkfile "$v2/debug/build/probe/1bf5493368dce3cd/fingerprint/$f" "$f"
+done
+mkfile "$v2/debug/build/probe/1bf5493368dce3cd/out/libprobe-1bf5493368dce3cd.rlib" "$artifact_bytes"
+mkfile "$v2/debug/build/probe/1bf5493368dce3cd/out/libprobe-1bf5493368dce3cd.rmeta" "$artifact_bytes"
+mkfile "$v2/debug/build/probe/1bf5493368dce3cd/out/probe-1bf5493368dce3cd.d" '/probe/src/lib.rs:'
+for f in build-script-build-script-build build-script-build-script-build.json \
+         dep-build-script-build-script-build invoked.timestamp; do
+  mkfile "$v2/debug/build/probe/08c7dda6eacd6dca/fingerprint/$f" "$f"
+done
+mkfile "$v2/debug/build/probe/08c7dda6eacd6dca/out/build_script_build" "$artifact_bytes"
+chmod +x "$v2/debug/build/probe/08c7dda6eacd6dca/out/build_script_build"
+mkfile "$v2/debug/build/probe/08c7dda6eacd6dca/out/build_script_build.d" '/probe/build.rs:'
+# A test binary: the same `out` directory as the rlib above, and the largest
+# thing in a real tree that a linker writes.
+for f in dep-test-lib-probe test-lib-probe test-lib-probe.json invoked.timestamp; do
+  mkfile "$v2/debug/build/probe/6a091d813b2be60d/fingerprint/$f" "$f"
+done
+mkfile "$v2/debug/build/probe/6a091d813b2be60d/out/probe-6a091d813b2be60d" "$artifact_bytes"
+chmod +x "$v2/debug/build/probe/6a091d813b2be60d/out/probe-6a091d813b2be60d"
+mkfile "$v2/debug/build/probe/6a091d813b2be60d/out/probe-6a091d813b2be60d.d" '/probe/src/lib.rs:'
+for f in run-build-script-build-script-build run-build-script-build-script-build.json; do
+  mkfile "$v2/debug/build/probe/da96cf45111f80dd/fingerprint/$f" "$f"
+done
+mkfile "$v2/debug/build/probe/da96cf45111f80dd/out/gen.txt" 'generated from 24 bytes'
+for f in invoked.timestamp root-output stdout stderr; do
+  mkfile "$v2/debug/build/probe/da96cf45111f80dd/run/$f" "$f"
+done
+# A build script that wrote into OUT_DIR and then FAILED: Cargo records the
+# run only on success, so this unit has `out/` populated and no `run/` at all.
+# Reading "no execution record" as "compile unit" left this shared, which is
+# the one state the old `-name build` selection covered and the first cut of
+# this one did not.
+mkfile "$v2/debug/build/probe/f00ded00f00ded00/out/gen.txt" 'half-written'
+mkfile "$v2/debug/incremental/probe-abc/s-xyz/dep-graph.bin" "$artifact_bytes"
+
+assert_partition "layout v2" "$v2" "$(cat <<'SPEC'
+private|.rustc_info.json
+private|debug/libprobe.d
+private|debug/build/probe/1bf5493368dce3cd/fingerprint/dep-lib-probe
+private|debug/build/probe/1bf5493368dce3cd/fingerprint/lib-probe
+private|debug/build/probe/1bf5493368dce3cd/fingerprint/lib-probe.json
+private|debug/build/probe/1bf5493368dce3cd/fingerprint/invoked.timestamp
+private|debug/build/probe/1bf5493368dce3cd/out/probe-1bf5493368dce3cd.d
+private|debug/build/probe/08c7dda6eacd6dca/fingerprint/dep-build-script-build-script-build
+private|debug/build/probe/08c7dda6eacd6dca/fingerprint/invoked.timestamp
+private|debug/build/probe/08c7dda6eacd6dca/out/build_script_build.d
+private|debug/build/probe/da96cf45111f80dd/fingerprint/run-build-script-build-script-build
+private|debug/build/probe/da96cf45111f80dd/out/gen.txt
+private|debug/build/probe/da96cf45111f80dd/run/root-output
+private|debug/build/probe/da96cf45111f80dd/run/stdout
+private|debug/build/probe/da96cf45111f80dd/run/stderr
+private|debug/build/probe/da96cf45111f80dd/run/invoked.timestamp
+private|debug/build/probe/f00ded00f00ded00/out/gen.txt
+shared|debug/libprobe.rlib
+shared|debug/build/probe/1bf5493368dce3cd/out/libprobe-1bf5493368dce3cd.rlib
+shared|debug/build/probe/1bf5493368dce3cd/out/libprobe-1bf5493368dce3cd.rmeta
+private|debug/build/probe/08c7dda6eacd6dca/out/build_script_build
+private|debug/build/probe/6a091d813b2be60d/out/probe-6a091d813b2be60d
+private|debug/build/probe/6a091d813b2be60d/out/probe-6a091d813b2be60d.d
+private|debug/build/probe/6a091d813b2be60d/fingerprint/dep-test-lib-probe
+shared|debug/incremental/probe-abc/s-xyz/dep-graph.bin
+SPEC
+)"
+# Layout v1: one `.fingerprint` and one `deps` per profile; `build/<pkg>-<hash>`
+# holds the build script's compiled binary in one unit directory and its run
+# metadata plus OUT_DIR in another.
+v1="$scratch/layout-v1"
+mkfile "$v1/CACHEDIR.TAG" 'Signature: 8a477f597d28d172'
+mkfile "$v1/.rustc_info.json" '{"rustc_fingerprint":1}'
+mkfile "$v1/debug/.cargo-lock" ''
+mkfile "$v1/debug/libprobe.rlib" "$artifact_bytes"
+mkfile "$v1/debug/libprobe.d" '/probe/src/lib.rs:'
+mkfile "$v1/debug/deps/libprobe-1bf5493368dce3cd.rlib" "$artifact_bytes"
+mkfile "$v1/debug/deps/libprobe-1bf5493368dce3cd.rmeta" "$artifact_bytes"
+mkfile "$v1/debug/deps/probe-1bf5493368dce3cd.d" '/probe/src/lib.rs:'
+# A test binary, which layout v1 leaves in `deps/` beside the rlibs.
+mkfile "$v1/debug/deps/probe-6a091d813b2be60d" "$artifact_bytes"
+chmod +x "$v1/debug/deps/probe-6a091d813b2be60d"
+mkfile "$v1/debug/deps/probe-6a091d813b2be60d.d" '/probe/src/lib.rs:'
+for f in dep-lib-probe lib-probe lib-probe.json invoked.timestamp; do
+  mkfile "$v1/debug/.fingerprint/probe-1bf5493368dce3cd/$f" "$f"
+done
+mkfile "$v1/debug/build/probe-08c7dda6eacd6dca/build-script-build" "$artifact_bytes"
+mkfile "$v1/debug/build/probe-08c7dda6eacd6dca/build_script_build-08c7dda6eacd6dca" "$artifact_bytes"
+chmod +x "$v1/debug/build/probe-08c7dda6eacd6dca/build-script-build" \
+         "$v1/debug/build/probe-08c7dda6eacd6dca/build_script_build-08c7dda6eacd6dca"
+mkfile "$v1/debug/build/probe-08c7dda6eacd6dca/build_script_build-08c7dda6eacd6dca.d" '/probe/build.rs:'
+for f in invoked.timestamp output root-output stderr; do
+  mkfile "$v1/debug/build/probe-da96cf45111f80dd/$f" "$f"
+done
+mkfile "$v1/debug/build/probe-da96cf45111f80dd/out/gen.txt" 'generated from 24 bytes'
+# The same never-succeeded build script under layout v1.
+mkfile "$v1/debug/build/probe-f00ded00f00ded00/out/gen.txt" 'half-written'
+mkfile "$v1/debug/incremental/probe-abc/s-xyz/dep-graph.bin" "$artifact_bytes"
+
+assert_partition "layout v1" "$v1" "$(cat <<'SPEC'
+private|.rustc_info.json
+private|debug/libprobe.d
+private|debug/deps/probe-1bf5493368dce3cd.d
+private|debug/.fingerprint/probe-1bf5493368dce3cd/dep-lib-probe
+private|debug/.fingerprint/probe-1bf5493368dce3cd/lib-probe
+private|debug/.fingerprint/probe-1bf5493368dce3cd/lib-probe.json
+private|debug/.fingerprint/probe-1bf5493368dce3cd/invoked.timestamp
+private|debug/build/probe-08c7dda6eacd6dca/build_script_build-08c7dda6eacd6dca.d
+private|debug/deps/probe-6a091d813b2be60d
+private|debug/deps/probe-6a091d813b2be60d.d
+private|debug/build/probe-da96cf45111f80dd/invoked.timestamp
+private|debug/build/probe-da96cf45111f80dd/output
+private|debug/build/probe-da96cf45111f80dd/root-output
+private|debug/build/probe-da96cf45111f80dd/stderr
+private|debug/build/probe-da96cf45111f80dd/out/gen.txt
+private|debug/build/probe-f00ded00f00ded00/out/gen.txt
+shared|debug/libprobe.rlib
+shared|debug/deps/libprobe-1bf5493368dce3cd.rlib
+shared|debug/deps/libprobe-1bf5493368dce3cd.rmeta
+private|debug/build/probe-08c7dda6eacd6dca/build-script-build
+private|debug/build/probe-08c7dda6eacd6dca/build_script_build-08c7dda6eacd6dca
+shared|debug/incremental/probe-abc/s-xyz/dep-graph.bin
+SPEC
+)"
+
+
+echo
+command -v cargo >/dev/null || {
+  echo "SKIP: no cargo on PATH — the fixture scenarios above ran, the live-Cargo ones cannot"
+  echo "hardlink-clone-selftest: ${pass_count} assertions passed"
+  exit 0
+}
+
 crate_dir="$scratch/probe"
 mkcrate "$crate_dir"
 cd "$crate_dir"
 
 export CARGO_INCREMENTAL=0
+
+# Every assertion below reads cargo's own words out of a build log
+# (`Compiling libdep`, `Fresh probe`). A CI image that forces colour splices an
+# ANSI reset between the status word and the crate name, at which point every
+# one of those greps silently stops matching and the suite reports the
+# opposite of what happened — observed on gitdan-ci's runner image, where
+# scenario 2 failed while the log it printed plainly showed `Compiling libdep`.
+# Pin the format the assertions are written against.
+export CARGO_TERM_COLOR=never
 # Checksum freshness is where the worst failure lives (the dep-* file carries
 # per-source checksums and is rewritten in place). Only available on nightly;
 # without it the test still covers the build/ and *.d families.
-CHECKSUM_MODE="off"
-if cargo +nightly -V >/dev/null 2>&1; then
-  export CARGO_UNSTABLE_CHECKSUM_FRESHNESS=true
-  CARGO_BIN=(cargo +nightly)
-  CHECKSUM_MODE="on"
-else
-  CARGO_BIN=(cargo)
-fi
-echo "=== checksum-freshness mode: ${CHECKSUM_MODE} ==="
-
 CONTENT_A='pub fn f() -> u32 { 1 }'
 CONTENT_B='pub fn f() -> u32 { 22222 } pub fn g() -> u32 { 7 }'
 
+# Probe the BEHAVIOUR, not the channel and not the flag. Two weaker probes
+# were tried against gitdan-ci's runner and each let the suite assert a
+# property the toolchain did not have:
+#
+#   `cargo +nightly -V`                     — answers "did a proxy called with
+#                                             +nightly exit 0". `-V`
+#                                             short-circuits before `-Z` is
+#                                             even parsed.
+#   `cargo +nightly -Z checksum-freshness   — answers "is this flag still
+#    locate-project`                          accepted", which since cargo PR
+#                                             #17382 (2026-08-22) is a
+#                                             different question from "is
+#                                             content freshness on". That PR
+#                                             demoted the flag to a gate and
+#                                             gave `build.fingerprint` the
+#                                             choice, defaulting to `mtime` —
+#                                             so 1.100.0-nightly accepts the
+#                                             flag and resolves freshness by
+#                                             mtime unless
+#                                             CARGO_BUILD_FINGERPRINT=content
+#                                             is set too. Measured 2026-08-26;
+#                                             see daniel/gitdan#62.
+#
+# The scenario at the end of this file depends on one thing and it is neither
+# of those: that changed content with an OLDER mtime rebuilds. Under mtime
+# freshness the correct answer is Fresh, so under mtime freshness that
+# scenario asserts a bug. So the probe simply performs that experiment, on its
+# own crate and its own target dir, with no clone anywhere near it — which is
+# also what makes it a control rather than a restatement of the scenario: the
+# probe establishes that the toolchain rebuilds on content, the scenario
+# establishes that a hardlink clone did not take that away.
+# THREE OUTCOMES, NOT TWO. An experiment that cannot tell a negative result
+# from a failed measurement is not settling the question, and the two are not
+# interchangeable here: "this toolchain resolves freshness by mtime" is a
+# statement about Cargo, while "a probe build failed" is a statement about this
+# machine. Collapsing them — which an earlier cut of this did, by returning
+# non-zero for both — makes a half-installed toolchain print a confident and
+# wrong explanation and quietly drop a scenario. The scenario still has to be
+# skipped in either case; what must not happen is the log claiming to know why.
+#
+#   0  content freshness measured ACTIVE   — both builds ran, the backdated
+#                                            rebuild recompiled
+#   3  measured INACTIVE                   — both builds ran, the backdated
+#                                            rebuild reported Fresh
+#   anything else  NOT MEASURED            — nothing was learned about the
+#                                            toolchain
+#
+# THE ANSWER CODES ARE 0 AND 3, AND THE GAP IS THE MECHANISM. Bash produces 1
+# for an ordinary command failure, 2 for a usage error, 126/127 for a command
+# it could not run, 128+n for a signal, and — this is the one that matters —
+# 1 for an unbound-variable or other EXPANSION failure, which happens before
+# the command runs and is therefore invisible to a `||` guard and to an ERR
+# trap alike. It never produces 3. So "not an answer code" is decided by a
+# property of the shell rather than by an enumeration of the ways a step can
+# go wrong, and a step added later without a guard, or with a guard that
+# cannot fire, lands on NOT MEASURED by construction.
+#
+# That is the whole reason INACTIVE is not 1. It was, and three review rounds
+# on this function each found a narrower way for a shell-generated 1 to be read
+# as a measurement — an unguarded command, then a typo'd variable name on a
+# line that HAS its guard. Each was closed by narrowing the failure surface,
+# which is a game with no last move. Moving the answer off the codes bash can
+# generate ends it instead: there is no longer a mutation that turns an error
+# into an answer, only mutations that turn an error into a different error.
+#
+# The guards below stay, and so does the trap, but their job is now reporting
+# rather than correctness: they make a failed step land on 2 with its logs
+# printed instead of on some incidental status, which is nicer to debug and
+# lands in the same place either way.
+#
+# One piece of that reporting layer is load-bearing and not obvious. A command
+# on the left of `||` — or in an `if` condition — runs with errexit suppressed,
+# and that suppression propagates into a subshell and is NOT undone by a
+# `set -e` inside it (measured on bash 5.3: an unguarded `false` there falls
+# through to `exit 0`). Calling with errexit disarmed at the site is the only
+# form that lets the subshell re-arm it; hence the `set +e` bracket. The ERR
+# trap is then required on top, because a bare `set -e` abort exits with the
+# FAILING COMMAND's status, and `false` gives 1.
+#
+# WHY NOT-MEASURED SKIPS RATHER THAN FAILS. The scenario it gates is the only thing in
+# this suite that depends on freshness mode; everything else still runs and
+# still catches real regressions. Failing instead would turn a statement about
+# one machine's toolchain into a red gate reading "the hardlink scheme is
+# broken" across the three repos consuming this action — the same category
+# error the three-state split exists to prevent, one level up. What would
+# change the answer is not-measured becoming the everyday CI outcome; it is not
+# — gitdan-ci's outcome is a measurement either way. It reported a measured
+# INACTIVE until 2026-08-26, for the reason recorded at the `export` below, and
+# an ACTIVE once both switches were set.
+CHECKSUM_MODE="off"
+CHECKSUM_REASON="no nightly on PATH accepting -Z checksum-freshness"
+CARGO_BIN=(cargo)
+checksum_freshness_probe() {
+  local d="$scratch/freshness-probe" t="$scratch/freshness-probe-target"
+  mkcrate "$d" || return 2  # 2 is simply "not 0 and not 3"; see the header
+  (
+    set -e
+    trap 'exit 2' ERR
+    cd "$d" || exit 2
+    printf '%s\n' "$CONTENT_A" > src/lib.rs || exit 2
+    CARGO_TARGET_DIR="$t" cargo +nightly build -q > "$scratch/freshness-probe-warm.log" 2>&1 || exit 2
+    printf '%s\n' "$CONTENT_B" > src/lib.rs || exit 2
+    touch -d '@1000000000' src/lib.rs || exit 2
+    CARGO_TARGET_DIR="$t" cargo +nightly build -v > "$scratch/freshness-probe.log" 2>&1 || exit 2
+    # 3, not 1: see the header. This is the only statement in the subshell that
+    # may report a measurement, and it is the only one that may exit 3.
+    if grep -qE '^\s+Fresh probe' "$scratch/freshness-probe.log"; then exit 3; fi
+    exit 0
+  )
+}
+if cargo +nightly -Z checksum-freshness locate-project > /dev/null 2>&1; then
+  export CARGO_UNSTABLE_CHECKSUM_FRESHNESS=true
+  # BOTH, since cargo PR #17382 (2026-08-22): the -Z flag only unlocks the
+  # feature and `build.fingerprint` selects it, defaulting to `mtime`. Setting
+  # the gate alone is what made this suite report a measured INACTIVE on
+  # 1.100.0-nightly and skip its strongest scenario (daniel/gitdan#62). Safe to
+  # export unconditionally — a Cargo that does not know the key ignores it
+  # silently, verified 2026-08-26 on 1.93.1 stable and 1.96.0-nightly, both of
+  # which still measure ACTIVE from the gate alone.
+  export CARGO_BUILD_FINGERPRINT=content
+  # Errexit off across the call, so the subshell can arm its own — see the
+  # header. `probe_rc` is read before it is restored.
+  probe_rc=0
+  set +e
+  checksum_freshness_probe
+  probe_rc=$?
+  set -e
+  case "$probe_rc" in
+    0)
+      CARGO_BIN=(cargo +nightly)
+      CHECKSUM_MODE="on"
+      CHECKSUM_REASON=""
+      ;;
+    3)
+      unset CARGO_UNSTABLE_CHECKSUM_FRESHNESS CARGO_BUILD_FINGERPRINT
+      CHECKSUM_REASON="this nightly accepts -Z checksum-freshness and build.fingerprint=content but still resolves freshness by mtime"
+      ;;
+    *)
+      unset CARGO_UNSTABLE_CHECKSUM_FRESHNESS CARGO_BUILD_FINGERPRINT
+      CHECKSUM_MODE="unmeasured"
+      CHECKSUM_REASON="the probe exited ${probe_rc}, which is not one of its answer codes, so this was NOT MEASURED — this toolchain may or may not resolve freshness by content"
+      # Loud, because the cost is silently lost coverage on a machine that
+      # might have had it. The suite continues: everything else it asserts is
+      # independent of freshness mode.
+      echo "::warning::hardlink-clone-selftest: could not measure whether this toolchain resolves freshness by content — the probe exited ${probe_rc}. This is a failure to measure, not a finding about Cargo."
+      tail -n 15 "$scratch/freshness-probe-warm.log" "$scratch/freshness-probe.log" 2>/dev/null | sed 's/^/    /' >&2 || true
+      ;;
+  esac
+fi
+cd "$crate_dir"
+echo "=== checksum-freshness mode: ${CHECKSUM_MODE}${CHECKSUM_REASON:+ — ${CHECKSUM_REASON}} ==="
+
 build_base() {
   local dir="$1"
   printf '%s\n' "$CONTENT_A" > src/lib.rs
@@ -112,11 +484,35 @@ fi
 ok "raw cp -al clone mutates the source ($(printf '%s\n' "$ctl_mutated" | wc -l) paths)"
 printf '%s\n' "$ctl_mutated" | sed 's/^/       /'
 
+# Reported, not asserted, and the distinction is the point. The control's job
+# is to prove the hazard exists at all, which the non-empty set above already
+# does; this line records WHICH families a given Cargo exhibits.
+#
+# The dep-info file is the worst of them — it carries the per-source
+# checksums, so mutating it through a shared inode turns a hardlink clone into
+# silent stale-artifact reuse rather than a slow build. Failing on its absence
+# would mean this suite goes red whenever upstream stops doing something we
+# never wanted it to do — and it would go red in the CONTROL, where a failure
+# reads as "the hazard is gone" rather than "upstream changed". Nothing is lost
+# by reporting it: the fix scenario below asserts the source is byte-identical
+# after a full rebuild in the clone, which covers every family this Cargo has,
+# named or not.
+#
+# THE PATTERN MUST MATCH BOTH LAYOUTS, and that is not a detail. Cargo's
+# build-dir layout v2 moved the file from `<profile>/.fingerprint/<unit>/dep-*`
+# to `<profile>/build/<pkg>/<hash>/fingerprint/dep-*` (stabilised by cargo PR
+# #17354, cargo 1.100.0, stable 2026-11-12; nightly default since 1.99). An
+# earlier cut of this line looked for the v1 path only, so on 2026-08-26,
+# against 1.100.0-nightly with content freshness genuinely on, it printed
+# "does NOT rewrite ... in place" directly beneath a control listing that
+# showed the rewrite. A reporting line that can contradict the data three
+# lines above it is worse than no line at all. `fingerprint/.*dep-` matches
+# either layout and neither `.d` family.
 if [ "$CHECKSUM_MODE" = "on" ]; then
-  if printf '%s' "$ctl_mutated" | grep -q '\.fingerprint/.*/dep-'; then
+  if printf '%s' "$ctl_mutated" | grep -q 'fingerprint/.*dep-'; then
-    ok "control confirms the checksum-freshness dep-info file is among the mutated set"
+    echo "  note: this cargo DOES rewrite its dep-info fingerprint file in place under content freshness"
   else
-    fail "expected .fingerprint/*/dep-* in the control's mutated set under checksum freshness"
+    echo "  note: this cargo does NOT rewrite its dep-info fingerprint file in place; only the build/ and *.d families appear above"
   fi
 fi
 
@@ -127,27 +523,44 @@ build_base "$base_fix"
 before=$(snapshot_tree "$base_fix")
 hardlink_clone_into "$base_fix" "$clone_fix" "selftest" || fail "hardlink_clone_into reported the destination already existed"
 
-# The clone's contract, asserted before anything builds in it: artifacts
+# The clone's contract, asserted before anything builds in it and asserted in
-# share inodes (that is what makes the clone near-free), and every file Cargo
+# BOTH directions: every file Cargo rewrites in place is privately owned (that
-# rewrites in place does not (that is what makes it sound). Checking after a
+# is what makes the clone sound), and every artifact still shares its inode
-# rebuild would prove nothing — the rebuild replaces those files anyway.
+# (that is what makes it near-free). Checking after a rebuild would prove
-shared=0; unshared=0
+# nothing — the rebuild replaces those files anyway.
+#
+# The mutable-family patterns cover both layouts: `.fingerprint/` is v1's,
+# `fingerprint/` and `run/` are v2's, and `gen.txt` is this crate's build
+# script's OUT_DIR product, which under v2 sits in a directory called `out`
+# beside sibling units whose `out` holds artifacts.
+shared=0; unshared=0; shared_bytes=0; copied_bytes=0
 while IFS= read -r f; do
-  rel="${f#$base_fix/}"
+  rel="${f#"$base_fix"/}"
   [ -e "$clone_fix/$rel" ] || continue
+  sz=$(stat -c '%s' "$f")
   if [ "$(stat -c '%i' "$f")" = "$(stat -c '%i' "$clone_fix/$rel")" ]; then
     case "$rel" in
-      */.fingerprint/*|*/build/*|*.d|.rustc_info.json)
+      */.fingerprint/*|*/fingerprint/*|*/run/*|*/out/gen.txt|*/output|*/root-output|*/stderr|*/invoked.timestamp|*.d|.rustc_info.json)
         fail "mutable path still shares an inode with the source: $rel" ;;
     esac
-    shared=$((shared + 1))
+    shared=$((shared + 1)); shared_bytes=$((shared_bytes + sz))
   else
-    unshared=$((unshared + 1))
+    case "$rel" in
+      *.rlib|*.rmeta)
+        fail "artifact was real-copied rather than shared: $rel" ;;
+    esac
+    unshared=$((unshared + 1)); copied_bytes=$((copied_bytes + sz))
   fi
 done < <(find "$base_fix" -type f)
 [ "$shared" -gt 0 ] || fail "nothing is shared — the clone degenerated into a full copy"
 [ "$unshared" -gt 0 ] || fail "nothing was unshared — unshare_mutable_paths did not run"
+total_bytes=$((shared_bytes + copied_bytes))
 ok "fresh clone shares ${shared} artifact files and privately owns ${unshared} mutable ones"
+# Reported, not asserted. This crate has no dependencies, so almost all of its
+# bytes are the two executables — a ratio that says nothing about a real tree.
+# The fixtures above are where the cost model is gated, because there the
+# composition is fixed.
+echo "  note: this clone real-copies $((copied_bytes * 100 / total_bytes))% of ${total_bytes} B"
 
 printf '%s\n' "$CONTENT_B" > src/lib.rs
 CARGO_TARGET_DIR="$clone_fix" "${CARGO_BIN[@]}" build -q
@@ -161,10 +574,105 @@ fi
 ok "no file in the source changed after a full rebuild in the clone"
 
 echo
+echo "=== a linked TEST binary, which nothing uplifts and nothing replaces ==="
+# The one artifact family that is NOT safe to share, and the reason
+# `unshare_mutable_paths` privately owns every executable. rustc writes an
+# rlib to a temporary and renames it in; the LINKER writes an executable
+# through whatever inode is already at the path.
+#
+# THE CRATE SHAPE IS LOAD-BEARING AND WAS WRONG ONCE. An earlier cut of this
+# scenario reused the lib+bin probe crate above, whose test binaries relink to
+# a FRESH inode — a shape gitdan-actions#17 records as measured safe. Both
+# halves then passed green against the unfixed selection, on the strength of
+# dep-info mutations the previous scenario already covers, and the scenario
+# pinned nothing. A bin-only crate with a unit test does exhibit the rewrite,
+# on cargo 1.93.1 stable and on 1.98.0-nightly and 1.100.0-nightly, so that is
+# what this builds. The shape is chosen by measurement rather than derived:
+# what separates a rewritten executable from an intact one is not established,
+# so the only crate shape this scenario may rest on is one observed to exhibit
+# the rewrite.
+mkbincrate() {
+  local dir="$1" marker="$2"
+  mkdir -p "$dir/src"
+  cat > "$dir/Cargo.toml" <<'TOML'
+[package]
+name = "binprobe"
+version = "0.1.0"
+edition = "2021"
+[workspace]
+TOML
+  cat > "$dir/src/main.rs" <<RS
+fn main() { println!("${marker}"); }
+#[cfg(test)]
+mod t { #[test] fn a() { assert_eq!("${marker}".len() > 0, true); } }
+RS
+}
+
+bin_dir="$scratch/binprobe"
+mkbincrate "$bin_dir" MARKER_AAAA
+cd "$bin_dir"
+base_exe="$scratch/base-exe"; clone_exe_ctl="$scratch/clone-exe-ctl"; clone_exe="$scratch/clone-exe"
+
+# Only the executables are read here. The families the other scenarios cover
+# would satisfy a "something changed" assertion on their own, which is exactly
+# how the earlier cut of this passed while pinning nothing.
+source_exe_digest() {
+  (cd "$1" && find . -type f -executable -print0 | sort -z | xargs -0 -r sha1sum) 2>/dev/null
+}
+
+CARGO_TARGET_DIR="$base_exe" "${CARGO_BIN[@]}" test --no-run -q > /dev/null 2>&1 \
+  || fail "the bin-only probe crate failed to build"
+before=$(source_exe_digest "$base_exe")
+cp -al "$base_exe" "$clone_exe_ctl"
+strip_cargo_locks "$clone_exe_ctl"
+mkbincrate "$bin_dir" MARKER_BBBB
+CARGO_TARGET_DIR="$clone_exe_ctl" "${CARGO_BIN[@]}" test --no-run -q > /dev/null 2>&1
+exe_ctl_mutated=$(mutated_paths "$before" "$(source_exe_digest "$base_exe")")
+
+# THREE OUTCOMES, as the freshness probe above has, and for the same reason: a
+# scenario that cannot tell "the fix works" from "the hazard never fired" is
+# not a gate. If this Cargo does not rewrite the source's test binary, the
+# assertion below would pass for a toolchain reason rather than a code one, so
+# it is skipped LOUDLY instead of passing quietly.
+if [ -z "$exe_ctl_mutated" ]; then
+  echo "::warning::hardlink-clone-selftest: this toolchain did not rewrite the source's test binary through a raw cp -al clone, so the linked-output scenario proves nothing here and was SKIPPED. That is a statement about this Cargo, not about unshare_mutable_paths."
+else
+  ok "control: a raw cp -al clone rewrites the source's own linked test binary"
+  printf '%s\n' "$exe_ctl_mutated" | sed 's/^/       /'
+
+  # Rebuild the base from the original marker so it is warm and consistent
+  # again, then do the same thing through the real clone.
+  mkbincrate "$bin_dir" MARKER_AAAA
+  CARGO_TARGET_DIR="$base_exe" "${CARGO_BIN[@]}" test --no-run -q > /dev/null 2>&1
+  before=$(snapshot_tree "$base_exe")
+  hardlink_clone_into "$base_exe" "$clone_exe" "selftest-exe" \
+    || fail "hardlink_clone_into refused the destination"
+  mkbincrate "$bin_dir" MARKER_BBBB
+  CARGO_TARGET_DIR="$clone_exe" "${CARGO_BIN[@]}" test --no-run -q > /dev/null 2>&1
+  exe_mutated=$(mutated_paths "$before" "$(snapshot_tree "$base_exe")")
+  if [ -n "$exe_mutated" ]; then
+    printf '%s\n' "$exe_mutated" | sed 's/^/       /' >&2
+    fail "a test build in the clone mutated the source through a shared inode"
+  fi
+  ok "no file in the source changed after a full test build in the clone"
+fi
+cd "$crate_dir"
+
+echo
+if [ "$CHECKSUM_MODE" = "on" ]; then
   echo "=== the whole point: the source's next build is still correct ==="
-# The source's cache holds artifacts built from CONTENT_A. Advance the source
+  # The source's cache holds artifacts built from CONTENT_A. Advance the
-# to CONTENT_B (as a merge would) and rebuild in it. If the clone had
+  # source to CONTENT_B (as a merge would) and rebuild in it. If the clone had
   # corrupted its dep-info, Cargo would report Fresh and keep the stale rlib.
+  #
+  # CHECKSUM-FRESHNESS ONLY, and the backdated mtime is why. Under checksum
+  # freshness the dep-info file's per-source checksums decide, so a 2001
+  # timestamp on changed content must still rebuild — the assertion below.
+  # Under Cargo's ordinary MTIME freshness the same timestamp means the source
+  # is older than the artifact, and reporting Fresh is the correct answer;
+  # asserting otherwise asserts a bug. This scenario was written against a
+  # machine with a nightly installed and, run without one, failed on that
+  # correct answer.
   printf '%s\n' "$CONTENT_B" > src/lib.rs
   touch -d '@1000000000' src/lib.rs
   log="$scratch/rebuild.log"
@@ -173,6 +681,10 @@ if grep -qE '^\s+Fresh probe' "$log"; then
     fail "source declared its own crate Fresh against sources it has never built — stale-artifact reuse"
   fi
   ok "source correctly rebuilt its crate after advancing to the clone's content"
+else
+  echo "=== skipped: the source's-next-build scenario needs content-based freshness ==="
+  echo "  reason: ${CHECKSUM_REASON}"
+fi
 
 echo
 echo "hardlink-clone-selftest: ${pass_count} assertions passed"

scripts/prune-cache-selftest.sh

@@ -65,10 +65,10 @@ origin="$scratch/origin.git"; git init -q --bare "$origin"
 work="$scratch/work"; git init -q "$work"
 (
   cd "$work"
-  git -c user.email=t@t -c user.name=t commit -q --allow-empty -m init
+  git -c user.email=t@t -c user.name=t -c commit.gpgsign=false commit -q --allow-empty -m init
   git branch -M main
-  git checkout -q -b dev; git -c user.email=t@t -c user.name=t commit -q --allow-empty -m dev
+  git checkout -q -b dev; git -c user.email=t@t -c user.name=t -c commit.gpgsign=false commit -q --allow-empty -m dev
-  git checkout -q -b feat/live; git -c user.email=t@t -c user.name=t commit -q --allow-empty -m live
+  git checkout -q -b feat/live; git -c user.email=t@t -c user.name=t -c commit.gpgsign=false commit -q --allow-empty -m live
   git remote add origin "$origin"
   git push -q origin main dev feat/live
 )

scripts/publish-snapshot-selftest.sh

@@ -53,9 +53,11 @@ pass_count=0
 fail() { echo "ASSERTION FAILED: $*" >&2; exit 1; }
 ok() { pass_count=$((pass_count + 1)); echo "PASS: $*"; }
 
-# Cargo and rustc REPLACE an artifact (write elsewhere, rename over the path)
+# rustc REPLACES an `.rlib`/`.rmeta` (writes elsewhere, renames over the path)
 # rather than truncating it in place, which is exactly why a snapshot may
-# share artifact inodes with the live target dir it was cloned from. The
+# share those inodes with the live target dir it was cloned from. Linker
+# outputs are the exception and are real-copied instead — see
+# `unshare_mutable_paths` in cache-lib.sh. The
 # fixtures here have to model that faithfully — a plain `>` redirect truncates
 # in place and would write straight through the shared inode into the
 # snapshot and every consumer, which is a property of the test fixture, not of

scripts/restore-mtimes-selftest.sh

@@ -75,6 +75,15 @@
 #      Asserts BOTH jobs correctly recompile the dependency and succeed.
 set -euo pipefail
 
+# Every assertion below reads cargo's own words out of a build log
+# (`Compiling libdep`, `Fresh probe`). A CI image that forces colour splices an
+# ANSI reset between the status word and the crate name, at which point every
+# one of those greps silently stops matching and the suite reports the
+# opposite of what happened — observed on gitdan-ci's runner image, where
+# scenario 2 failed while the log it printed plainly showed `Compiling libdep`.
+# Pin the format the assertions are written against.
+export CARGO_TERM_COLOR=never
+
 script_dir=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 restore_mtimes="$script_dir/restore-mtimes.sh"
 
@@ -122,7 +131,7 @@ assert_log_lacks() {
 # every run — simulate that before each restore-mtimes.sh pass, exactly as
 # CI would see it, so this test exercises the script the same way CI does.
 stamp_checkout_now() {
-  find . -path ./.git -prune -o -type f -print | xargs touch
+  find . -path ./.git -prune -o -type f -print0 | xargs -0 touch
 }
 
 echo "=== building scratch workspace ==="
@@ -131,6 +140,11 @@ cd "$repo"
 git init -q
 git config user.email test@example.com
 git config user.name "restore-mtimes-selftest"
+# Local to this mktemp'd throwaway repo. Without it the eight commits below
+# inherit the developer's GLOBAL commit.gpgsign, which makes whether this gate
+# passes depend on their gpg agent — observed as a red run caused by a full
+# disk breaking gpg, in a suite that has nothing to say about either.
+git config commit.gpgsign false
 
 cat > Cargo.toml <<'EOF'
 [workspace]

scripts/restore-mtimes.sh

@@ -64,7 +64,8 @@
 # two refs' fingerprints ever share a directory and this script never has to
 # arbitrate freshness across refs — only within one ref's own history, which
 # is exactly what it is built to do soundly. On a nightly toolchain,
-# CARGO_UNSTABLE_CHECKSUM_FRESHNESS is a complementary, stronger guarantee
+# CARGO_UNSTABLE_CHECKSUM_FRESHNESS plus CARGO_BUILD_FINGERPRINT=content (both,
+# since cargo PR #17382 on 2026-08-22) is a complementary, stronger guarantee
 # (content-addressed rather than mtime-based freshness); this script is not
 # made redundant by it, because directory-form `rerun-if-changed` build-script
 # watches are not covered by it and stable historical mtimes stay cheap

scripts/seed-target-dir-selftest.sh

@@ -74,6 +74,17 @@
 #      onto one set of fingerprints. That is the silent stale-reuse bug the
 #      whole scheme exists to prevent, so the failure has to abort the clone
 #      rather than be swallowed.
+#  11. THE CONSUMER'S MARKER ACTUALLY STOPS THE PUBLISHER — reader_lock_acquire
+#      is exercised (not faked, unlike publish-snapshot-selftest.sh's scenario
+#      6) by a real hardlink_clone_into racing a real, concurrent
+#      publish-snapshot.sh republish of the exact snapshot being cloned. Pins
+#      that the publisher OBSERVABLY WAITS on this consumer's marker — its own
+#      log reports entering the drain wait — rather than only that the run
+#      succeeds, which stayed green with the marker call deleted (issue #10).
+#      Kept as its own scenario, not folded into 8a, because 8a already pins
+#      exactly one property (the identity check) for exactly one mutant, and
+#      the suite's one-scenario-one-mutant diagonal across 8a to 8d and 10 is
+#      deliberate.
 set -euo pipefail
 script_dir=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 . "$script_dir/cache-lib.sh"
@@ -635,5 +646,84 @@ leftovers=$(find "$root" -maxdepth 1 \( -name '.stage-*' -o -name '.reading-*' \
 ok "no staging or reader-marker scratch left behind"
 rm -f "$scratch/bin/cp"
 
+echo
+echo "=== 11: the consumer's marker actually stops the publisher ==="
+# hardlink_clone_into's reader_lock_acquire call is exercised here, not
+# faked. publish-snapshot-selftest.sh's scenario 6 stands a hand-written
+# marker file in for "a consumer whose clone outlasts the grace period" — a
+# deliberate simplification that does not need this consumer's clone code at
+# all, so it cannot tell reader_lock_acquire apart from no marker existing.
+# This scenario forces the two real scripts to race on the same snapshot: a
+# consumer hardlink-cloning it, and a publisher republishing it out from
+# under that clone, exactly as 8a does to force the identity check — but what
+# is pinned here is not the consumer's response to the rotation (8a's
+# property), it is that the PUBLISHER, reading the SAME marker this
+# consumer's clone wrote, is observed to have entered its drain wait. Deleting
+# reader_lock_acquire (issue #10) leaves the marker never written: the
+# publisher's live_reader_count sees zero readers on its first check and
+# proceeds straight to reclaiming the rotated generation, silently — the run
+# still succeeds, and nothing about its own outcome says so, only the absence
+# of a line in the publisher's log.
+INTLK=$(cache_key feat/interlock-consumer)
+INTLK_BASE=$(cache_key release/5)
+TAG_INTLK=jobIntlk
+rm -rf "$root/snapshot-$INTLK_BASE" "$root/target-$INTLK_BASE"
+make_tree "$root/snapshot-$INTLK_BASE" intlk-gen1
+make_tree "$root/target-$INTLK_BASE" intlk-gen2
+intlk_gen1_inode=$(stat -c '%i' "$root/snapshot-$INTLK_BASE")
+
+cat > "$scratch/bin/cp" <<EOF
+#!/usr/bin/env bash
+# Fires once, only on the consumer's own top-level hardlink clone — identified
+# by its destination, the seed's private staging path. Every other cp in the
+# process tree (the unshare copies, and the publisher's own staging clone)
+# falls through to the real one.
+if [ "\${@: -1}" = "$root/.stage-$TAG_INTLK" ] && [ ! -e "$scratch/firedIntlk" ]; then
+  : > "$scratch/firedIntlk"
+  rc=0; "$real_cp" "\$@" || rc=\$?
+  # Concurrently: a real, second publish of the exact snapshot this consumer
+  # is cloning — the republish that, without the marker this consumer's clone
+  # holds, would reclaim the generation out from under it.
+  ( CACHE_READ_GRACE_SECONDS=10 bash "$script_dir/publish-snapshot.sh" "$INTLK_BASE" "$root" pubIntlk \
+      > "$scratch/logPubIntlk" 2>&1
+    echo \$? > "$scratch/rcPubIntlk" ) &
+  # Hand control back only once the swap is on disk, so the identity read
+  # immediately after this cp is guaranteed to resolve to the new generation
+  # — the same technique 8a uses to force the interleaving rather than hope
+  # for it. reader_lock_release does not run until AFTER this script exits,
+  # so the marker stays live for the publisher's whole swap-and-scan.
+  deadline=\$(( \$(date +%s) + 60 ))
+  while [ "\$(stat -c '%i' "$root/snapshot-$INTLK_BASE" 2>/dev/null)" = "$intlk_gen1_inode" ]; do
+    [ "\$(date +%s)" -lt "\$deadline" ] || { echo "stub cp: the publisher never swapped the snapshot" >&2; exit 92; }
+    sleep 0.05
+  done
+  exit \$rc
+fi
+exec "$real_cp" "\$@"
+EOF
+chmod +x "$scratch/bin/cp"
+
+rcIntlk=0
+seed_with_stub "$INTLK" "$INTLK_BASE" "$root" "$TAG_INTLK" > "$scratch/logIntlk" 2>&1 || rcIntlk=$?
+[ -e "$scratch/firedIntlk" ] \
+  || fail "the stubbed cp never fired: the consumer never raced the publisher, so this scenario proves nothing"
+ok "the consumer's clone raced a real, concurrent republish of its own source"
+wait_for_file "$scratch/rcPubIntlk" "the publisher never finished"
+[ "$(cat "$scratch/rcPubIntlk")" = "0" ] || { tail -40 "$scratch/logPubIntlk"; fail "publish-snapshot.sh exited non-zero"; }
+[ "$rcIntlk" = "0" ] || { tail -40 "$scratch/logIntlk"; fail "the seed exited non-zero"; }
+ok "both sides of the race completed"
+
+# The property under test: not that the run succeeded, but that the publisher
+# itself reports having found a live reader and waited on it. This is silent
+# in the consumer's own log and in the run's exit status alike — only the
+# publisher's log carries it.
+grep -q "readers: waiting for 1 in-flight clone(s) of snapshot-${INTLK_BASE}" "$scratch/logPubIntlk" \
+  || { tail -40 "$scratch/logPubIntlk"; fail "the publisher never reported waiting on the consumer's reader marker — the interlock did not observably engage"; }
+ok "the publisher observably waited on the consumer's own reader marker before reclaiming the rotated snapshot generation"
+leftovers=$(find "$root" -maxdepth 1 \( -name '.stage-*' -o -name '.reading-*' -o -name '.publish-*' \) -print)
+[ -z "$leftovers" ] || fail "scratch left behind: ${leftovers}"
+ok "no staging, reader-marker or deferred-generation scratch left behind"
+rm -f "$scratch/bin/cp"
+
 echo
 echo "seed-target-dir-selftest: ${pass_count} assertions passed"

scripts/selftest.sh

@@ -13,7 +13,7 @@ script_dir=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 FAST=0
 [ "${1:-}" = "--fast" ] && FAST=1
 
-FIXTURE_TESTS=(seed-target-dir-selftest.sh publish-snapshot-selftest.sh prune-cache-selftest.sh)
+FIXTURE_TESTS=(cache-root-selftest.sh seed-target-dir-selftest.sh publish-snapshot-selftest.sh prune-cache-selftest.sh)
 CARGO_TESTS=(hardlink-clone-selftest.sh restore-mtimes-selftest.sh)
 
 TESTS=("${FIXTURE_TESTS[@]}")