[30.x] Backports

[fanquake]

Backports:

• multiprocess: Don't require bitcoin -m argument when IPC options are used #33229
• ci: fix buildx gha cache authentication on forks #33508
• multiprocess: Fix high overhead from message logging #33517
• ci: Use native platform for win-cross task #33558
• miner: fix empty mempool case for waitNext() #33566
• depends: Use $(package)_file_name when downloading from the fallback #33580
• ci: Properly include $FILE_ENV in DEPENDS_HASH #33581
• test: change log rate limit version gate #33612
• doc: correct topology requirements in submitpackage helptext #33630
• interfaces: enable cancelling running waitNext calls #33676
• ci: Fix lint runner selection (and docker cache) #33744
• scripted-diff: Remove obsolete comment #33826
• doc: Correct pkgin command usage on NetBSD #33827
• depends: Add patch for Windows11Style plugin #33906
• Modernize custom filtering bitcoin-core/gui#899
• Add createwallet, createwalletdescriptor, and migratewallet to history filter bitcoin-core/gui#901

[DrahtBot]

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

Code Coverage & Benchmarks

For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/33609.

Reviews

See the guideline for information on the review process.

Type	Reviewers
ACK	willcl-ark, marcofleon

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

LLM Linter (✨ experimental)

Possible typos and grammar issues:

+Use the calculation from Commonstyle and add the increased padding and
+horizontally layouted buttons to the horizontal size hint.

• Commonstyle -> CommonStyle [proper name/camel-case; "Commonstyle" is likely a typo and may confuse readers]
• layouted -> laid-out [incorrect past participle; "laid-out" (or "laid out") is the correct phrasing]

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

• std::accumulate(..., CAmount{0}) in src/node/miner.cpp
• std::accumulate(..., CAmount{0}) in src/node/miner.cpp

^2025-12-01

[ryanofsky]

It could make sense to backport #33229 too, I think. It does have the Needs backport (30.x) label, but doesn't appear by default because it is closed. (Sorry if this is the wrong place to discuss)

[waketraindev]

Please add bitcoin-core/gui#910 , it's test coverage for bitcoin-core/gui#901

[plebhash]

please add #33676, it would be highly appreciated as a way to expedite stratum-mining/sv2-apps#81

[plebhash]

reporting here some testing methodology and results

---

originally I reported #33554, which I can still reliably reproduce with branch 2025-10-06-abort-proxy-io of https://github.com/plebhash/sv2-bitcoin-core against Bitcoin Core:

• v30.0rc3
• v30.0

basically, the rust code runs for a little bit, and then Bitcoin Core crashes with:

Assertion failed: (m_loop), function operator->, file proxy.h, line 60.
[1]    8022 abort      ./build/bin/bitcoin-node -signet -ipcbind=unix  -debug=ipc

(plus all the backtraces reported on #33554)

---

on the same issue #33554 @ryanofsky said:

Just to be clear about next steps, the node aborts should be prevented by bitcoin-core/libmultiprocess#214 which is set to be backported in #33519.

this backporting was already merged... so running the same rust client from 2025-10-06-abort-proxy-io against 30.x now triggers some new behavior, where Bitcoin Core does NOT crash, and instead I see this on the rust client logs:

2025-11-29T16:09:03.869139Z  INFO logger: Sent new CoinbaseOutputConstraints
2025-11-29T16:09:03.872778Z ERROR bitcoin_core_sv2: Failed to get response: Failed: remote exception: std::exception: thread busy

(cc @ismaelsadeeq this confirms what we discussed over lunch today: if a client naively "forgets" to call interruptWait when it's supposed to, Bitcoin Core no longer crashes)

---

on #33554 @ryanofsky also said:

Regardless of this, the rust client code needs to be changed to either not make other IPC calls at the same time as there is an active waitNext() IPC call, or to just create more server threads and request the waitNext call to run on a different thread than the other IPC calls.

which is my segway to #33676, which is being backported here

the rust client of 2025-11-29-abort-proxy-io-interrupt-wait branch is essentially the same code of 2025-10-06-abort-proxy-io, but adapted to call interruptWait at the appropriate time, thus avoiding two concurrent waitNext requests against the same server thread

running this rust client against fanquake:30_1_backports, Bitcoin Core does NOT crash, and also we do NOT get the thread busy error on the rust client side

---

TLDR: #33676 successfully tested on the branch of this PR, and I'm confident we have a way forward on stratum-mining/sv2-apps#81

🙏 thanks a lot @ryanofsky @ismaelsadeeq @fanquake @Sjors

[willcl-ark]

Backports all look good to me.

We are missing a release note for #33229 currently, I think it's one worth adding too.

[willcl-ark]

ACK b2cb203

[marcofleon]

Lgtm, ACK b2cb203