#36679 ci: enable IPC in the i686 job

Enables IPC in the 32-bit x86 (i686) continuous integration job. It renames the CI environment setup script and drops the configuration flag that disabled IPC for that job.

Problem: The i686 CI job was arbitrarily chosen to test building with `-DENABLE_IPC=OFF`, leaving IPC functionality untested on 32-bit x86 where compiler ABI differences can cause failures.

Evidence of importance:

• Commit message: "Exercise IPC on 32-bit x86, where compiler ABI differences have caused failures."
• Description points to bitcoin-core/libmultiprocess#383 as an example of 32-bit ABI issues requiring coverage.

Risk of a mistake: low (tests, tooling). A mistake would only cause failures or misconfigurations in the i686 CI job.

Changed since the previous assessment: Diff is now available; the implementation renames `ci/test/00_setup_env_i686_no_ipc.sh` to `00_setup_env_i686.sh` and updates the workflow job definition.

Ready

Nothing known that would invalidate a review now.

Waiting on (from the thread read):

• reviewer: input from ryanofsky as requested by fanquake 2026-10-28
• reviewer: feedback from ryanofsky as requested by fanquake 2026-10-28
• author: response to l0rinc's remark regarding 32-bit architecture support 2026-10-28

Strong

substantive support, no open objection (maflcko)

Thread notes: Initial record for PR #36679. maflcko provided a substantive ACK after discussing -DENABLE_IPC=OFF coverage, while l0rinc noted skepticism regarding 32-bit support without blocking. / Initial record. maflcko provided an ACK after confirming CI history. l0rinc questioned spending effort on 32-bit support, and fanquake pinged ryanofsky for thoughts.

Objections, suggestions, and questions:

Support:

• maflcko (member) (ACK, 2026-10-28) [read]: The changes here should be fine since 0136e17c0a963e82bf47c223d41f4d5c009520c0

Participants:

• fanquake (member, neutral, 2 statements 2026-10-28 to 2026-10-28): pointed out recent musl and 32-bit issues and pinged ryanofsky
• l0rinc (member, neutral, 1 statements 2026-10-28 to 2026-10-28): questioned whether supporting 32-bit architectures is the best use of time, but stated they won't block
• maflcko (member, support, 4 statements 2026-10-28 to 2026-10-28): ACKed after clarifying whether -DENABLE_IPC=OFF coverage ever caught CI failures

7 lines under test/bench/ci.

• ci/test/{00_setup_env_i686_no_ipc.sh => 00_setup_env_i686.sh} +4/-3
• .github/workflows/ci.yml +2/-2

The short text the ranking pass compares across PRs.

Enables IPC testing in the 32-bit x86 (i686) continuous integration job. The i686 job previously explicitly disabled IPC to provide coverage for -DENABLE_IPC=OFF, leaving 32-bit IPC configurations untested in CI. Compiler ABI differences on 32-bit platforms have previously caused failures in libmultiprocess, making CI coverage desirable to catch regressions. The change affects only CI configuration.

P3 · test coverage (ranking pass: Moved up from 41: adds 32-bit ABI coverage for IPC where failures have previously occurred)

Enables IPC coverage on 32-bit x86 CI to catch compiler ABI issues. The commit message notes this exercises IPC on 32-bit x86 'where compiler ABI differences have caused failures', providing worthwhile platform test coverage without addressing an active release-blocking build failure.

Membership: Modifies `.github/workflows/ci.yml` and `ci/test/` scripts to enable IPC in the i686 CI job.

Factors: security/stability 0, bug 0, performance 0, user value 0, leverage 1

P3 · test coverage

Extends existing test suite execution to cover IPC under 32-bit x86. While primarily a CI configuration change, it closes a test coverage gap where 32-bit ABI differences previously went unexercised.

Membership: Carries the Tests label and expands test matrix execution to cover IPC on 32-bit systems.

Factors: security/stability 0, bug 0, performance 0, user value 0, leverage 1