pass.
Note for the integration tree: patch 5 changes the prop_handler
-extract() prototype; the stripe_alloc topic's two property handlers
-need the new prototype when both are merged.
+extract() prototype. The stripe_alloc topic's two property handlers
+are not inheritable and, since 2026-11-04 (ck1104), set no extract hook
+at all, so the two series merge without a resolution.
Filipe Manana (1):
btrfs: dont propagate nodatacow mount option on files with inherited
rc 0. Regression smoke (rb1028-regress.sh minus qu-p1) on #314 (bhive)
and #64 (dhive): same as rb1103 (all PASS, retire-wait-abort SKIP 77).
-**Integration conflict, not promoted:** patch 5 (koraynilay 4/6) changes
+**Integration conflict, not promoted (resolved by ck1104, below):** patch 5 (koraynilay 4/6) changes
the prop_handler extract() prototype; the stripe-alloc topic adds two
property handlers (stripe_alloc, stripe_alloc_allow_rmw) with the old
prototype, so the testing octopus does not build without a resolution.
exactly those 4 lines). Zygo's 10-29 rule puts interacting changes in
topic commits, not in the octopus, so promotion waits for his call.
+## ck1104 (2026-11-04): stripe-alloc drops its extract hooks; promoted
+
+Zygo, 2026-11-04, option (a) of the ck1103 integration conflict: a
+stripe-alloc commit drops `.extract` and the two extract functions from the
+stripe_alloc and stripe_alloc_allow_rmw handlers (not inheritable, so
+btrfs_inode_inherit_props() never called them). Dropping them alone still
+conflicted: the stripe_alloc functions sat directly above
+prop_compression_extract(), whose prototype line patch 5 changes. The same
+commit therefore moves prop_compression_extract() back next to
+prop_compression_ignore(). The testing octopus now merges with **no
+resolution** on both lanes. Proof (raid56-wh-harness
+`forensics/ck1104-{618,mn}/oct-*.txt`): props.c from the lane base, merged
+with plain `git merge-file` in either order (compress-fixes, stripe-alloc),
+is clean and byte-identical to testing's.
+
+compress-fixes content is the ck1103 candidate's. misc-next: the same
+commit, 464a17c7110f. 6.18: koraynilay 3/6 gets `Assisted-by:
+Claude:claude-opus-5-5` after its trailers (Zygo 11-04: an edited backport
+gets the tag; message-only, every tree identical), so it is db2c3773cf30.
+No other compress-fixes commit qualifies: the zstd backport and e8a0095c7df17
+match upstream line for line, and Filipe's patch and koraynilay 1/6, 2/6,
+4/6-6/6 match the misc-next commits (1/6 through a clean 3-way apply).
+Ours carry the tag already where they changed code.
+
+Promoted 2026-11-04: misc-next testing 2365f2f67131, zb64 1ca72f265f4e (#66);
+6.18 compress-fixes db2c3773cf30, testing d339564c74f1, zb64 7af5ce2893df
+(#316). Backups: `refs/backup/ck1104-promote-20261104/<lane>/`.
+
+| test | kernel | result |
+|---|---|---|
+| compress-props, all phases | 6.18 #316 (bhive, ahive), misc-next #66 (dhive, ehive); KASAN inline + lockdep + BTRFS_ASSERT | PASS 66/66 on all four; X1 `zlib` 1245184 = zlib:3, X2 `zstd` 1024000 = zstd:15 (= ck1103) |
+| rb1028-regress minus qu-p1 | #316 (bhive), #66 (dhive) | per-test verdicts identical to ck1103 (all PASS, retire-wait-abort SKIP 77) |
+
## Not covered / open
- (Obsolete since rb1103: the patch that handled a parent directory
recording BTRFS_ORDERED_IOERR. In 4/5, v2 turns v1's ASSERT(0, ...)
into DEBUG_WARN(...). 2/5 changed only its commit message and tags.
+On both lanes the topic also ends, after the 85 patches numbered here,
+with "btrfs: stripe_alloc: drop the extract hooks of the stripe_alloc
+properties" (2026-11-04). The stripe_alloc and stripe_alloc_allow_rmw
+handlers are not inheritable, and btrfs_inode_inherit_props(), the only
+caller of a handler's extract hook, skips those before calling it, so
+both hooks were dead code returning NULL. Dropping them keeps the
+handlers independent of the extract() prototype, which koraynilay's
+per-inode compression level series changes. The patch also moves
+prop_compression_extract() back next to prop_compression_ignore(), so
+the stripe_alloc functions no longer sit between the compression
+hooks. No functional change. At the next export it folds into patches
+4 and 40, which add the two handlers.
+
The write hole
==============
under lockdep with no report, which covers the 6.18 port's lock class.
Evidence: raid56-wh-harness `forensics/rb1103-618/`.
+## ck1104 (2026-11-04): drop the extract hooks of the stripe_alloc properties
+
+New tip commit on both lanes (6.18 5387a3c4275b, misc-next 92ed6cc158af):
+"btrfs: stripe_alloc: drop the extract hooks of the stripe_alloc
+properties". It removes prop_stripe_alloc_extract() and
+prop_stripe_rmw_extract() and their `.extract` initializers. Both
+handlers are `.inheritable = 0`, and the only `->extract` caller,
+btrfs_inode_inherit_props(), skips non-inheritable handlers before the
+call. The commit also moves prop_compression_extract() back next to
+prop_compression_ignore(). No functional change. It keeps the handlers
+independent of the extract() prototype that koraynilay's level series
+(compress-fixes patch 5) changes, so the testing octopus merges without a
+resolution. Caller audit: every `extract` / `prop_handler` /
+`inheritable` use in fs/btrfs on the stripe-alloc, low-space and test-knobs
+tips and on the ck1103 candidate testing/zb64 (raid56-wh-harness
+`forensics/ck1104-{618,mn}/audit.txt`). Nothing outside props.c names the
+functions or reads `->extract`. low-space and test-knobs are replayed on
+top: per-commit content, authors, committers, dates and messages
+identical (`authorship.txt`).
+
+Builds: per-commit buildcheck of the new commit and of the 7 / 3 low-space
+and 2 test-knobs commits (6.18 / misc-next): failed=0, new warnings 0, tip
+links rc 0. zb64 KASAN inline + lockdep + BTRFS_ASSERT: 6.18 #316 (md5
+8d584efd61bc2288caca92ecd28b7950), misc-next #66 (md5
+92da00ba1da7e2b61637a02d0af0c6db), 0 fs/btrfs warnings.
+
+| test | kernel | result |
+|---|---|---|
+| mount-state-test (new case 4b, harness 30d25186: btrfs.stripe_alloc and btrfs.stripe_alloc_allow_rmw on the top-level root are not copied to new files, directories or a subvolume, live and after a mount cycle; btrfs.compression inherited as the positive control; setting either outside the root is EINVAL) | #316 bhive + ahive, #66 dhive + ehive | PASS 8/8 on all four |
+| same, before the change | 6.18 #313 (ahive), misc-next #63 (ehive) | PASS 8/8 (non-inheritance was already the behaviour; the commit is dead-code removal) |
+| property-remount-test | all four | PASS (18 OK) |
+| stripe-meta-toggle-test (raid5 metadata, runtime enable/disable) | all four | PASS (7 OK) |
+| rb1028-regress minus qu-p1 | #316 bhive, #66 dhive | per-test rc identical to ck1103/rb1103: all PASS, retire-wait-abort SKIP 77; enospc arms 0 acked loss |
+| compress-props, all phases | all four | PASS 66/66 |
+
+Other tests also set the properties, through common.sh's wh_sa_enable /
+wh_mopts_apply: every stripe_alloc test, the rb1028 set included. They
+exercise validate/apply, which the commit does not touch. Only
+mount-state-test 4b reads the properties on inodes other than the root.
+Kernel logs: no KASAN/WARN/BUG/lockdep/hung task/assertion; debug_locks 1,
+taint 0. At the next export the commit folds into patches 4 and 40.
+
## Not covered / open
Patches without a test of their own (covered at most incidentally by the