[32.x] More Backports #36300

1. 
   fanquake commented at 3:53 PM on September 19, 2026: member

   Includes:

   • #36230
   • #36297
   • #36309
   • #36312
   • #36316
   • #36322
   • #36324
   • #36336 (partial)
   • #36364
   • #36365
   • #36377
2. rpc: Correct OpenRPC default metadata

   Correct six argument metadata entries that produce misleading or invalid
   OpenRPC defaults.
   
   The getdeploymentinfo blockhash fallback and four sighashtype fallbacks
   describe how omitted arguments are resolved. They are not literal values
   accepted by the RPCs, so mark them as DefaultHint values.
   
   The send include_watching option is boolean, but its string default makes
   the generated schema internally inconsistent. Use a boolean value,
   matching the analogous sendall option.
   
   Runtime behavior is unchanged.
   
   Github-Pull: #36297
   Rebased-From: b7f740c594ba1482344a7906b799fb62c533464f

   7a0964a252
3. fanquake added this to the milestone 32.0 on Sep 19, 2026
4. DrahtBot added the label Backport on Sep 19, 2026
5. 
   DrahtBot commented at 3:53 PM on September 19, 2026: contributor

   <!--e57a25ab6845829454e8d69fc972939a-->

   The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

   <!--006a51241073e994b41acfe9ec718e94-->

   Code Coverage & Benchmarks

   For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/36300.

   <!--021abf342d371248e50ceaed478a90ca-->

   Reviews

   See the guideline and AI policy for information on the review process.

   Type	Reviewers
   ACK	marcofleon, willcl-ark
   Approach ACK	ismaelsadeeq

   If your review is incorrectly listed, please copy-paste <code>&lt;!--meta-tag:bot-skip--&gt;</code> into the comment that the bot should ignore.

   <!--5faf32d7da4f0f540f40219e4f7537a3-->

   LLM Linter (✨ experimental)

   Possible places where named args for integral literals may be used (e.g. func(x, /*named_arg=*/0) in C++, and func(x, named_arg=0) in Python):

   • Parse(descriptor_str, keys, error, true) in src/wallet/walletutil.cpp

   Possible places where comparison-specific test macros should replace generic comparisons:

   • [test/functional/interface_http.py] assert count < num_req, f"Server handled the whole batch of {num_req}: nothing was throttled" -> use assert_greater_than(num_req, count) for the comparison instead.

   ^{2026-10-01 15:57:12}

6. wallet: Remove WalletDescriptor's default constructor

   Drop the default constructor and add a factory function to construct a
   WalletDescriptor from stream
   
   Github-Pull: #36230
   Rebased-From: 4cc00f7bc04b4c07827728ba5ce1094b338a76a8

   98b9a497e0
7. wallet: Make WalletDescriptor's descriptor const

   Github-Pull: #36230
   Rebased-From: bb5e8320d76f5175fdf86bdc4c2d24bfc1eed438

   655148d0ac
8. wallet: Compare descriptors by hash of canonical string

   The canonical string comparison was slow because it would compute the
   canonical string for each comparison. This can be sped up by holding the
   canonical string in memory, computed upon construction of
   WalletDescriptor. To reduce memory usage, this string is further hashed
   so that the comparison operates over the hash of the canonical string.
   
   Github-Pull: #36230
   Rebased-From: 64abb3e8d3610449d541bfff2db23b9ace3fa111

   7fdb8584da
9. descriptor: Explicitly handle use_apostrophe cases

   Github-Pull: #36230
   Rebased-From: 405b1d6055e78866fe2d86ba6fb16534f9554a32

   81e580401d
10. descriptor, doc: Clarify miniscript CompatDescriptorHash

    The hardened indicator for Minscript expressions in
    CompatDescriptorHash uses whichever hardened indicator was originally
    given by the user.
    
    Github-Pull: #36230
    Rebased-From: f0f6dce849634e801ac81705176ca8b96abbaa09

    d17455aae8
11. wallet: Document WalletDescriptor::UpdateFrom

    Github-Pull: #36230
    Rebased-From: f6cbcfd98a000deb17521520074ba46965c410b1

    e6c57c0982
12. test: Simplify miniscript descriptor check in wallet back compat

    Github-Pull: #36230
    Rebased-From: ebf2f69ee81925057cdcc881e42e861dbdf39ae5

    c46dacf923
13. test: Add 31.1 to wallet back compat

    Github-Pull: #36230
    Rebased-From: 61edcf92532502e91c2e17b8c2a23b45f5773936

    21a947002e
14. DrahtBot added the label CI failed on Sep 19, 2026
15. DrahtBot removed the label CI failed on Sep 19, 2026
16. doc: scope claims about private broadcast feature

    Reframe in terms of risk reduction rather than absolutes
    
    Github-Pull: #36309
    Rebased-From: 2630d8e6c9d60579372c019e4b2cc7e73163a5d8

    bfc87b96ef
17. private broadcast: mark feature as experimental

    Github-Pull: #36309
    Rebased-From: f41372998377ab8310476f0c03294e69cc3222e8

    fa11293ddc
18. ci: raise BuildKit step log limit

    The Windows cross-toolchain build can exceed BuildKit's default 2 MiB
    per-step log limit, which hides useful output.
    
    Raise the limit to 5 MiB for both CI builder configurations. This
    retains substantially more diagnostics without enabling unlimited logs.
    
    Github-Pull: #36316
    Rebased-From: 96d487c11e8ff55f5df3975272916a6d13c70162

    c65060e6d3
19. ci: use POSIX threads in Nix Windows cross shell

    Guix release binaries use the POSIX GCC thread model, but the pinned
    Nixpkgs cross compiler defaults to MCF. Select POSIX through the
    Nixpkgs threads hook so GCC and its target libraries agree.
    
    Leave threads.package null because making winpthreads a target
    dependency exposes its headers to native GCC build tools. Supply its
    headers and libraries through EXTRA_*_FOR_TARGET for target libraries
    instead.
    
    Add winpthreads to the compiler wrapper so builds in the shell get
    its include and library paths without manual NIX_CFLAGS_COMPILE or
    NIX_LDFLAGS exports.
    
    Github-Pull: #36316
    Rebased-From: 1d59d5ad2550b2ff6aaea2a630d1c6afa877ea2b

    6854ead11c
20. ci: print Windows cross compiler thread model

    BuildKit can truncate the compiler build log, hiding its configure
    flags. Print the compiler version and thread model when CI enters the
    Nix test shell so a mismatch is visible in the job output.
    
    Github-Pull: #36316
    Rebased-From: 9fd18aedb088b5ca6a7c424aab2a4a1ec3ef4eeb

    d06f51300e
21. net: don't discourage private broadcast peers

    Keep private broadcast connections outside normal discouragement handling.
    Misbehaving private broadcast peers are still disconnected.
    
    Github-Pull: #36312
    Rebased-From: eab6630addc385d013db5fe6779679228e445d09

    8ec0fd46f4
22. test: Fixup MAX_BODY_SIZE http throttling test

    Github-Pull: #36324
    Rebased-From: fa2c62036b1d1a1f5858e7ea96cbc719503a942d

    782ca5a16f
23. test: characterize fee callback loss at shutdown

    Shutdown saves and unregisters the fee estimator before draining queued validation callbacks, so a pending block update can be missing from the saved estimates.
    The mempool estimator also ignores its saved file on restart when the saved tip no longer matches the active tip.
    
    Queue a block update after the scheduler stops and check that the block-policy height in the saved estimates does not reflect it.
    
    Co-authored-by: Rob Hamilton <6456095+Rob1Ham@users.noreply.github.com>
    
    Github-Pull: #36322
    Rebased-From: 08729e6ef9c9c5ec75834633a39f51f768d789c7

    651670dec1
24. init: drain callbacks before saving fee estimates

    Save and unregister the fee estimator after the existing callback drain so queued block updates are reflected in the saved estimates.
    
    The estimator comment now says its shutdown actions follow the callback drain.
    The old paragraph below the indexes claimed the drain only let wallets catch up, although wallets are already unloaded.
    The new comment beside the final chainstate flush keeps the accurate point that callbacks queued there can be dropped.
    
    Github-Pull: #36322
    Rebased-From: 8c27c38526419de2eaca1d4857df2172ad1da1ab

    8b93cadccc
25. DrahtBot added the label CI failed on Sep 25, 2026
26. DrahtBot removed the label CI failed on Sep 25, 2026
27. 
    hebasto commented at 12:18 PM on September 28, 2026: member

    Can we have #36233 here?

28. tools: Call SHA256AutoDetect in bitcoin-util, bitcoin-tx and bitcoin-wallet

    Only kernel::Context users selected a hardware SHA256 implementation,
    so these tools fell back to the generic one. On an M2 Max, grind is
    ~5x faster and signing 2000 P2PKH inputs with bitcoin-tx ~2.3x.
    
    Github-Pull: #36364
    Rebased-From: 4048909a53422bf86b207a9b5c8650e8edd25737

    02a7f11038
29. 
    ismaelsadeeq commented at 6:37 AM on September 30, 2026: member

    Approach ACK, please include #36365 and #36336.

30. doc: add 461 (Deterministic ECDSA signatures with low-R grinding) to bips.md

    Github-Pull: #36377
    Rebased-From: d57346d682f17e6cf09f8acc5e7893ee16eec193

    26f3c91025
31. log: move CreateNewBlock() log line behind a new mining category

    Github-Pull: #36336
    Rebased-From: 54b8795121d3081019befd8a2bc70adcd06fcb65

    18765f6734
32. fanquake referenced this in commit 5cf7f5ed46 on Sep 30, 2026
33. fanquake referenced this in commit 635745539f on Sep 30, 2026
34. fanquake referenced this in commit 7f0ffc9b10 on Sep 30, 2026
35. DrahtBot added the label CI failed on Sep 30, 2026
36. DrahtBot removed the label CI failed on Sep 30, 2026
37. 
    willcl-ark commented at 12:23 PM on September 30, 2026: member

    ACK 7f0ffc9b10cfd855ab8f6fcd0fe29e555336b459

    All nice clean backports, all appear functionally correct too.

    Many missing release notes, but IIUC this is fine here as we are going to update the wiki directly (should remember to do this though!)

38. DrahtBot requested review from ismaelsadeeq on Sep 30, 2026
39. fametrano referenced this in commit f0eaefcd8a on Sep 30, 2026
40. fametrano referenced this in commit 10f299afd2 on Sep 30, 2026
41. fametrano referenced this in commit 4e6d832a25 on Sep 30, 2026
42. fametrano referenced this in commit 213193b6bd on Sep 30, 2026
43. fanquake marked this as ready for review on Oct 1, 2026
44. fanquake requested review from stickies-v on Oct 1, 2026
45. fanquake requested review from marcofleon on Oct 1, 2026
46. fees: rename the default fee_rate_estimator value to "auto"

    "none" read as though fee estimation was disabled, so rename the public
    value (and the FeeRateEstimatorType::NONE enumerator) to "auto", which
    describes what it does.
    
    Also reject unknown fee_rate_estimator values instead of silently
    coercing them to the default, so a mistyped value surfaces an error. The
    previous "none" spelling was never released, so it is rejected too.
    
    Co-authored-by: w0xlt <94266259+w0xlt@users.noreply.github.com>
    
    Github-Pull: #36365
    Rebased-From: 4056908f0feae2a1caf9b53cfbe77bb8091246c9

    d7de90cc1e
47. fees: fall back to block_policy when the mempool estimator can't estimate

    When the mempool policy estimator cannot produce an estimate, GetFeeRateEstimate()
    now returns the block policy estimate rather than an error. The combined estimate
    starts from the block policy estimate and lowers it with the mempool estimate only
    when present, so the fallback path shares the log that reports the selected fee
    rate. An error is returned only when the block policy estimate itself is unavailable.
    
    Github-Pull: #36365
    Rebased-From: c0b7ca3dbdbaecdd045c8a36d9758470358880a0

    162c3b2187
48. fees: clear mined-block stats when the mempool load fails

    When the mempool fails to load at startup (persistence disabled, or a
    missing or corrupt mempool.dat), init notifies the fee_rate_estimator_man,
    which clears its tracked mined-block window. Otherwise the estimator
    would keep a window describing a mempool the node no longer has and,
    once the mempool refilled, serve an estimate built on it.
    
    Clearing makes the estimator report insufficient data until the window refills from the
    current tip, so the combined estimate falls back to the block policy
    estimate meanwhile.
    
    Github-Pull: #36365
    Rebased-From: 5b77288673e2b74de1512180a8911f435311ca7a

    532b496bc1
49. fanquake force-pushed on Oct 1, 2026
50. marcofleon approved
51. 
    marcofleon commented at 3:59 PM on October 1, 2026: contributor

    ACK 532b496bc1194f36d28e333ab4eb2b7b8dbdaf59

    Backports are clean.

52. DrahtBot requested review from willcl-ark on Oct 1, 2026
53. willcl-ark approved
54. 
    willcl-ark commented at 4:04 PM on October 1, 2026: member

    ACK 532b496bc11

    metadata update in git range-diff 7f0ffc9b10c...532b496bc11 only

55. fanquake merged this on Oct 1, 2026
56. fanquake closed this on Oct 1, 2026

    ---

Review Requested
ismaelsadeeqstickies-v

Milestone
32.0