deps(mimalloc): set MI_OVERRIDE=OFF on Windows by dylan-conway · Pull Request #29467 · oven-sh/bun

dylan-conway · 2026-04-18T20:59:22Z

Summary

Windows debug builds fail to link since the dev3 mimalloc bump (#29420 / #29435):

lld-link: error: duplicate symbol: _expand
>>> defined at mimalloc-debug.lib(alloc.c.obj)
>>> defined at libucrtd.lib(expand.obj)

Root cause: mimalloc.ts never set MI_OVERRIDE for Windows — only a comment claiming the upstream default was "no override". The actual default is ON. Pre-dev3 this was harmless because alloc-override.c's _MSC_VER block was an empty comment ("cannot override malloc unless using a dll"). Upstream microsoft/mimalloc#1259 / #1263 filled it with real CRT symbol definitions (_expand, _msize, _msize_base, _free_base, free), so the static lib now exports them and collides with the debug CRT.

Fix: explicit MI_OVERRIDE=OFF on Windows. Bun links the static CRT and calls mi_* directly; nothing routes through CRT malloc, so override has no benefit there. This restores the effective pre-dev3 behavior.

Not stale cache: dep_configure already uses cmake --fresh, so the cache was correctly regenerated — it got ON because that's the real default.

Why CI didn't catch it: all ci-* profiles use buildType: "Release" (/MT → libucrt.lib). The duplicate only fires under /MTd because libucrtd.lib's expand.obj is pulled in for its debug-heap symbols. CI never builds Windows debug.

Test plan

bunx tsc --noEmit -p scripts/build/tsconfig.json
bun scripts/build.ts --configure-only
Windows debug: bun bd --version → 1.3.13-debug (previously failed at link)
Verified build/debug/deps/mimalloc/CMakeCache.txt shows MI_OVERRIDE:BOOL=OFF after reconfigure
CI (release Windows — also flips ON→OFF; expected no-op since pre-dev3 override did nothing on Windows static)

mimalloc's CMake defaults MI_OVERRIDE to ON. The Windows branch in mimalloc.ts was a comment ("use mimalloc's defaults — no override"), which was wrong about the default and set nothing. This was harmless before dev3 because alloc-override.c's _MSC_VER block was a no-op ("cannot override malloc unless using a dll"). Upstream microsoft/mimalloc#1259/#1263 filled it in with real CRT symbol definitions (_expand/_msize/free), so the static lib now exports them and lld-link fails Windows debug builds with: duplicate symbol: _expand defined in mimalloc-debug.lib(alloc.c.obj) and libucrtd.lib(expand.obj) Bun links the static CRT and calls mi_* directly on Windows; nothing routes through CRT malloc, so override has no benefit. Setting OFF restores the effective pre-dev3 behavior. CI didn't catch this because all ci-* profiles use buildType: Release, which links /MT (libucrt) — its expand.obj isn't pulled into the link. Only /MTd (libucrtd) drags expand.obj in for its debug-heap symbols.

robobun · 2026-04-18T20:59:33Z

^{Updated 7:56 PM PT - Apr 18th, 2026}

❌ @dylan-conway, your commit 1432ab3 has 3 failures in Build #46361 (All Failures):

test/js/bun/webview/webview-chrome.test.ts - code 1 on 🍎 13 aarch64
test/js/bun/shell/leak.test.ts - code 1 on 🐧 3.23 x64-baseline
test/js/bun/util/v8-heap-snapshot.test.ts - SIGKILL on 🐧 25.04 x64-baseline

🧪 To try this PR locally:

bunx bun-pr 29467

That installs a local version of the PR into your bun-29467 executable, so you can run:

bun-29467 --bun

coderabbitai · 2026-04-18T21:00:56Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 2f6bc007-161e-47e4-8695-8f79cecb6041

📥 Commits

Reviewing files that changed from the base of the PR and between 684dd94 and 1432ab3.

📒 Files selected for processing (1)

scripts/build/deps/mimalloc.ts

Walkthrough

Added an explicit Windows branch in the mimalloc build script that sets MI_OVERRIDE = "OFF" for Windows targets; other existing cfg.asan, cfg.darwin, and cfg.linux branches remain unchanged and subsequent debug/valgrind flag logic is unaffected.

Changes

Cohort / File(s)	Summary
Windows CMake configuration `scripts/build/deps/mimalloc.ts`	Introduced an explicit `cfg.windows` branch that sets `args.MI_OVERRIDE = "OFF"` for Windows builds. Existing `cfg.asan`, `cfg.darwin`, and `cfg.linux` logic unchanged; debug/valgrind/flag handling also unchanged.

🚥 Pre-merge checks | ✅ 2

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title directly and concisely summarizes the main change: setting MI_OVERRIDE=OFF on Windows to fix debug build linking failures.
Description check	✅ Passed	The description comprehensively explains what the PR does and how it was verified, exceeding the template requirements with detailed context on root cause, fix rationale, and testing.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

claude

LGTM — simple, targeted build fix with clear root-cause analysis; the only finding is a minor doc nit in the comment matrix.

Extended reasoning...

Overview

Single-file change to scripts/build/deps/mimalloc.ts: adds an explicit else if (cfg.windows) branch setting MI_OVERRIDE=OFF, and removes a stale/incorrect comment that claimed the Windows default was already "no override". Touches only the build-script layer, not runtime code.

Security risks

None. This is a CMake flag passed during dependency configuration. It controls whether mimalloc replaces the system malloc/free on Windows — explicitly disabling it (which matches the pre-dev3 effective behavior) has no security implications.

Level of scrutiny

Low. The change is mechanical: one new else if branch, one removed stale comment, well-explained rationale. The author verified locally (CMakeCache shows MI_OVERRIDE:BOOL=OFF) and the PR description traces the root cause in detail. The fix is consistent with the existing pattern for macOS/ASAN.

Other factors

The sole bug report is a documentation nit (decision matrix comment omits the new Windows row) — non-blocking and does not affect correctness. No prior reviews on this PR.

claude

LGTM — my prior inline comment was a false positive; the Windows row is already present in the decision matrix. Straightforward, well-documented fix.

Extended reasoning...

Overview

Single-file change to scripts/build/deps/mimalloc.ts: adds an explicit MI_OVERRIDE=OFF CMake argument for Windows builds and adds a Windows row to the override decision matrix comment. The old code had an incorrect comment claiming Windows used "no override" via mimalloc defaults, when the actual default is ON.

Security risks

None. This is a build configuration change affecting a memory allocator compile-time flag. It has no effect on runtime security boundaries, authentication, or data handling.

Level of scrutiny

Low. The change is mechanical: one new else if (cfg.windows) branch setting a single CMake boolean, plus a documentation update to keep the matrix in sync. The root cause (upstream mimalloc dev3 filling in alloc-override.c with real CRT symbols) is well-researched and documented. The fix (explicit OFF) is the obvious correct response.

Other factors

My previous inline comment (flagging a missing Windows row in the decision matrix) was a false positive — the PR already adds that row. No bugs were found by the bug hunting system. The test plan is thorough: TypeScript type-check, configure-only dry run, and a verified CMakeCache.txt showing MI_OVERRIDE:BOOL=OFF after reconfigure. CI coverage gap (release-only Windows CI) is acknowledged and explained.

## Summary Windows debug builds fail to link since the dev3 mimalloc bump (oven-sh#29420 / oven-sh#29435): ``` lld-link: error: duplicate symbol: _expand >>> defined at mimalloc-debug.lib(alloc.c.obj) >>> defined at libucrtd.lib(expand.obj) ``` **Root cause:** `mimalloc.ts` never set `MI_OVERRIDE` for Windows — only a comment claiming the upstream default was "no override". The actual default is `ON`. Pre-dev3 this was harmless because `alloc-override.c`'s `_MSC_VER` block was an empty comment ("cannot override malloc unless using a dll"). Upstream [microsoft/mimalloc#1259](microsoft/mimalloc#1259) / oven-sh#1263 filled it with real CRT symbol definitions (`_expand`, `_msize`, `_msize_base`, `_free_base`, `free`), so the static lib now exports them and collides with the debug CRT. **Fix:** explicit `MI_OVERRIDE=OFF` on Windows. Bun links the static CRT and calls `mi_*` directly; nothing routes through CRT malloc, so override has no benefit there. This restores the effective pre-dev3 behavior. **Not stale cache:** `dep_configure` already uses `cmake --fresh`, so the cache was correctly regenerated — it got `ON` because that's the real default. **Why CI didn't catch it:** all `ci-*` profiles use `buildType: "Release"` (`/MT` → `libucrt.lib`). The duplicate only fires under `/MTd` because `libucrtd.lib`'s `expand.obj` is pulled in for its debug-heap symbols. CI never builds Windows debug. ## Test plan - [x] `bunx tsc --noEmit -p scripts/build/tsconfig.json` - [x] `bun scripts/build.ts --configure-only` - [x] Windows debug: `bun bd --version` → `1.3.13-debug` (previously failed at link) - [x] Verified `build/debug/deps/mimalloc/CMakeCache.txt` shows `MI_OVERRIDE:BOOL=OFF` after reconfigure - [ ] CI (release Windows — also flips ON→OFF; expected no-op since pre-dev3 override did nothing on Windows static)

claude Bot reviewed Apr 18, 2026

View reviewed changes

Comment thread scripts/build/deps/mimalloc.ts

deps(mimalloc): add Windows row to override decision matrix

1432ab3

claude Bot reviewed Apr 18, 2026

View reviewed changes

dylan-conway merged commit 40ffda1 into main Apr 18, 2026
58 of 61 checks passed

dylan-conway deleted the claude/interesting-mahavira-c5f1c5 branch April 18, 2026 22:06

robobun mentioned this pull request Apr 20, 2026

test: sync Node.js test suite with v24.14.1 (d89bb1b482) #29359

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

deps(mimalloc): set MI_OVERRIDE=OFF on Windows#29467

deps(mimalloc): set MI_OVERRIDE=OFF on Windows#29467
dylan-conway merged 2 commits into
mainfrom
claude/interesting-mahavira-c5f1c5

dylan-conway commented Apr 18, 2026

Uh oh!

robobun commented Apr 18, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Apr 18, 2026 •

edited

Loading

Uh oh!

claude Bot left a comment

Uh oh!

Uh oh!

claude Bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

dylan-conway commented Apr 18, 2026

Summary

Test plan

Uh oh!

robobun commented Apr 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot commented Apr 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

claude Bot left a comment

Choose a reason for hiding this comment

Overview

Security risks

Level of scrutiny

Other factors

Uh oh!

Uh oh!

claude Bot left a comment

Choose a reason for hiding this comment

Overview

Security risks

Level of scrutiny

Other factors

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

robobun commented Apr 18, 2026 •

edited

Loading

coderabbitai Bot commented Apr 18, 2026 •

edited

Loading