not tested. Not shown on a kernel without the series: without hints the
device takes chunks, so the control is the device-less filesystem instead.
+## Fixed-width mirror coverage and negative control (2026-11-06, ah1106)
+
+New arms of `alloc-hints-test.sh` (harness 03d60649) for patch 8: e3 (e2
+with raid1c3 data), m3 (kakra/linux#40's raid1c3 METADATA case: a small
+PREFERRED_METADATA pair + PREFERRED_DATA devices of three sizes, 3 rounds of
+`balance -m`: every new metadata/system chunk on exactly {1,2,most-free}),
+m4 (raid1c4 metadata over METADATA_ONLY, PREFERRED_METADATA and the best two
+of three PREFERRED_DATA, a large PREFERRED_NONE unused), nf (a nearly full
+preferred device, data and metadata: sizes reported; no ENOSPC/abort, the
+fallback once it is full). Control: 6.18 zb64 876b62099832 minus the two
+trim lines in `btrfs_select_hint_tiers()` (refs/cand/ah1106/618-ctl
+b4c90a2c65da, #319's config, 0 lines differ).
+
+| arm | 6.18 zb64 876b62099832 (bhive #319) | control (bhive #1) | misc-next zb64 dcf19c743b68 (dhive #69) |
+|---|---|---|---|
+| a b c d e1 s | PASS | PASS | PASS |
+| e2 | PASS | **FAIL** (2/2 new data chunks off devid 1) | PASS |
+| e3 | PASS | **FAIL** (2/2) | PASS |
+| m3 | PASS (6/6 on {1,2,5}) | **FAIL** (6/6 on {3,4,5}) | PASS |
+| m4 | PASS (6/6 on {1,2,4,5}) | **FAIL** (6/6 lose a preferred device) | PASS |
+| nf1 | PASS (960 + 319 MiB chunks on devid 1, then {2,3}) | **FAIL** (devid 1 never used) | PASS (same chunks) |
+| nf2 | PASS (31 MiB system chunks on the nearly full pair) | PASS (vacuous) | PASS (same chunks) |
+
+KASAN + PROVE_LOCKING + BTRFS_ASSERT on all three, lockdep live at the end,
+no WARN/abort/forced read-only. The control FAILs are placement only (no
+ENOSPC, every file verifies). No arm is blind. Evidence: harness
+`forensics/ah1106/` (README, full-319.log, full-ctl.log, full-mn69.log).
+This closes "e2 (patch 8) negative control: not built" below.
+
## Not covered / open
- **KERNEL BUG (found 2026-10-18): space estimates ignore the hints.** See
layout, so the loop did not reproduce it. The deterministic
`alloc-hints-space-test.sh` reproduces what is very likely the same
defect.
-- e2 (patch 8) negative control: needs a kernel without patch 8; not built.
+- e2 (patch 8) negative control: done 2026-11-06 (ah1106, above).
- `make headers_install` on the series (patch 9): not run this round.
- Striped profiles (raid0/10/5/6) and stripe_alloc data under hints: not
tested.