- Use D16 VMs (16 vCPU) for all Windows CI runners
- Pin Node.js to 24.3.0 (ABI 137) for duckdb prebuilt + test compat
- Wrap test runner in vs-shell.ps1 so node-gyp has cl.exe
- Revert ccache persistent config (VMs are ephemeral)
- Restore Uninstall-Windows-Defender (reboot clears pending state)
- Add Windows ARM64 to install.ps1, install.sh, bun-release
- Set parallelism to 2 for Windows tests
- Clarify Packer vs CI runner VM sizes in comments
[build images]
- Cygwin: fix path to root/bin, use Machine scope (survives Sysprep)
- Git: add usr/bin to Machine PATH (ships cat, head, tail, echo, etc.)
Previously cygwin used User scope PATH which was wiped by Sysprep,
and the path was wrong (missing root/ prefix).
[build images]
bun.sh/install.ps1 installs to user profile PATH which is lost
after Sysprep generalizes the image. Copy to C:\Windows\System32
like we do for ARM64.
[build images]
Install-Scoop-Package uses SilentlyContinue which never throws,
so the catch block (blob storage mirror) was unreachable. Now we
check if nssm is actually on PATH after the scoop attempt.
[build images]
Move error suppression from Install-7zip to Install-Scoop-Package so
all scoop installs are resilient to non-fatal post_install errors
(7zip 7zr.exe locked, llvm-arm64 missing Uninstall.exe, etc).
[build images]
The Remove-Item error on 7zr.exe happens on ARM64 regardless of user
context (packer user, not just SYSTEM). Temporarily set ErrorActionPreference
to SilentlyContinue and convert all output streams to strings so the
error doesn't propagate to PowerShell's exit code.
[build images]
The ARM64 Remove-Item error only happens under Azure Run Command
SYSTEM context. With Packer's WinRM, scoop runs as the packer user
and the post_install cleanup works fine.
[build images]
Azure Run Command uses x64-emulated PowerShell on ARM64 VMs, which
causes all tools to think they're on x64. Use Sysnative path to
re-launch bootstrap as native ARM64 so PROCESSOR_ARCHITECTURE,
package installs, cmake, and everything else sees the real arch.
[build images]
Azure Run Command runs x64-emulated PowerShell on ARM64 VMs, so
$env:PROCESSOR_ARCHITECTURE reports AMD64. The registry value at
HKLM:\SYSTEM\CurrentControlSet\Control\Session Manager\Environment
always reports the real architecture.
[build images]
RuntimeInformation.OSArchitecture doesn't exist in PowerShell 5.1
(which Azure Run Command uses), so IsARM64 was always false. This
caused ALL ARM64-specific code paths to be skipped.
[build images]
Child process scoop installs break PATH/shims. Only use child process
for 7zip on ARM64 (post_install error). Everything else runs in-process.
nssm falls back to blob storage mirror when nssm.cc is down.
[build images]
- VS installer exit code 3010 (reboot required) is not a real error
- mingw command is gcc, not mingw
- Use powershell -Command for scoop install isolation
[build images]
Scoop's 7zip post_install throws a terminating error on ARM64 that
propagates regardless of ErrorActionPreference. Running scoop in a
child PowerShell process isolates the error completely.
[build images]
7zip is a Scoop dependency of git. On ARM64 SYSTEM, 7zip's post_install
fails trying to delete 7zr.exe from TEMP. Fix by:
1. Moving Install-7zip before Install-Git so it's already installed
2. Using SilentlyContinue in Install-Scoop-Package then verifying
[build images]
The Remove-Item error in Scoop's 7zip post_install is a terminating
error that propagates through try/catch. Use SilentlyContinue to
suppress it, then verify 7z is actually installed.
[build images]
- 7zip: tolerate post_install cleanup error (access denied on TEMP)
- Rust: set CARGO_HOME/RUSTUP_HOME before rustup-init to avoid
SYSTEM profile path issue (Move-Item failure)
- nssm: fall back to blob storage mirror when nssm.cc returns 503
On ARM64 Windows, Chocolatey installs x64 cmake and LLVM. x64 cmake
reads PROCESSOR_ARCHITECTURE=AMD64 (overridden by OS for x64 processes),
causing CMAKE_SYSTEM_PROCESSOR=AMD64 and -march=haswell to be passed to
ARM64 clang-cl.
- Add ARM64 detection to bootstrap.ps1
- Install cmake ARM64 MSI directly from GitHub on ARM64
- Install LLVM ARM64 (woa64) installer directly from GitHub on ARM64
- Skip nasm (x86 assembler) and mingw on ARM64
- Warn about bun ARM64 not being available via Chocolatey
## Summary
- Update LLVM version references across build scripts, Dockerfiles, CI,
Nix configs, and documentation
- Fix LLVM 21 `-Wcharacter-conversion` errors in WebKit bindings:
- `EncodingTables.h`: pragma for intentional char32_t/char16_t
comparisons
- `TextCodecCJK.cpp`: widen `gb18030AsymmetricEncode` param to char32_t
- `URLPatternParser`: widen `isValidNameCodepoint` param to char32_t,
cast for `startsWith`
- Fix `__libcpp_verbose_abort` noexcept mismatch (LLVM 21 uses
`_NOEXCEPT`)
- Fix dangling pointer in `BunJSCModule.h` (`toCString` temporary
lifetime)
- Remove `useMathSumPreciseMethod` (removed upstream in JSC)
**Before merging:** Merge https://github.com/oven-sh/WebKit/pull/153
first, then update `WEBKIT_VERSION` in `cmake/tools/SetupWebKit.cmake`
to point to the merged commit.
## Test plan
- [ ] Build bun debug on macOS with LLVM 21
- [ ] Build bun on Linux (glibc)
- [ ] Build bun on Linux (musl)
- [ ] Build bun on Windows
- [ ] Run test suite
Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
## Summary
- Remove sccache support entirely, use ccache only
- Missing ccache no longer fails the build (just skips caching)
- Remove S3 distributed cache support
## Changes
- Remove `cmake/tools/SetupSccache.cmake` and S3 distributed cache
support
- Simplify `CMakeLists.txt` to only use ccache
- Update `SetupCcache.cmake` to not fail when ccache is missing
- Replace sccache with ccache in bootstrap scripts (sh, ps1)
- Update `.buildkite/Dockerfile` to install ccache instead of sccache
- Update `flake.nix` and `shell.nix` to use ccache
- Update documentation (CONTRIBUTING.md, contributing.mdx,
building-windows.mdx)
- Remove `scripts/build-cache/` directory (was only for sccache S3
access)
## Test plan
- [x] Build completes successfully with `bun bd`
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude Bot <claude-bot@bun.sh>
Co-authored-by: Claude <noreply@anthropic.com>
### What does this PR do?
Closes#13012
On Linux, when any Bun process spawned by `runner.node.mjs` crashes, we
run GDB in batch mode to print a backtrace from the core file.
And on all platforms, we run a mini `bun.report` server which collects
crashes reported by any Bun process executed during the tests, and after
each test `runner.node.mjs` fetches and prints any new crashes from the
server.
<details>
<summary>example 1</summary>
```
#0 crash_handler.crash () at crash_handler.zig:1513
#1 0x0000000002cf4020 in crash_handler.crashHandler (reason=..., error_return_trace=0x0, begin_addr=...) at crash_handler.zig:479
#2 0x0000000002cefe25 in crash_handler.handleSegfaultPosix (sig=<optimized out>, info=<optimized out>) at crash_handler.zig:800
#3 0x00000000045a1124 in WTF::jscSignalHandler (sig=11, info=0x7ffe044e30b0, ucontext=0x0) at vendor/WebKit/Source/WTF/wtf/threads/Signals.cpp:548
#4 <signal handler called>
#5 JSC::JSCell::type (this=0x0) at vendor/WebKit/Source/JavaScriptCore/runtime/JSCellInlines.h:137
#6 JSC::JSObject::getOwnNonIndexPropertySlot (this=0x150bc914fe18, vm=..., structure=0x150a0102de50, propertyName=..., slot=...) at vendor/WebKit/Source/JavaScriptCore/runtime/JSObject.h:1348
#7 JSC::JSObject::getPropertySlot<false> (this=0x150bc914fe18, globalObject=0x150b864e0088, propertyName=..., slot=...) at vendor/WebKit/Source/JavaScriptCore/runtime/JSObject.h:1433
#8 JSC::JSValue::getPropertySlot (this=0x7ffe044e4880, globalObject=0x150b864e0088, propertyName=..., slot=...) at vendor/WebKit/Source/JavaScriptCore/runtime/JSCJSValueInlines.h:1108
#9 JSC::JSValue::get (this=0x7ffe044e4880, globalObject=0x150b864e0088, propertyName=..., slot=...) at vendor/WebKit/Source/JavaScriptCore/runtime/JSCJSValueInlines.h:1065
#10 JSC::LLInt::performLLIntGetByID (bytecodeIndex=..., codeBlock=0x150b861e7740, globalObject=0x150b864e0088, baseValue=..., ident=..., metadata=...) at vendor/WebKit/Source/JavaScriptCore/llint/LLIntSlowPaths.cpp:878
#11 0x0000000004d7b055 in llint_slow_path_get_by_id (callFrame=0x7ffe044e4ab0, pc=0x150bc92ea0e7) at vendor/WebKit/Source/JavaScriptCore/llint/LLIntSlowPaths.cpp:946
#12 0x0000000003dd6042 in llint_op_get_by_id ()
#13 0x0000000000000000 in ?? ()
```
</details>
<details>
<summary>example 2</summary>
```
#0 crash_handler.crash () at crash_handler.zig:1513
#1 0x0000000002c5db80 in crash_handler.crashHandler (reason=..., error_return_trace=0x0, begin_addr=...) at crash_handler.zig:479
#2 0x0000000002c59f60 in crash_handler.handleSegfaultPosix (sig=<optimized out>, info=<optimized out>) at crash_handler.zig:800
#3 0x00000000042ecc88 in WTF::jscSignalHandler (sig=11, info=0xfffff60141b0, ucontext=0xfffff6014230) at vendor/WebKit/Source/WTF/wtf/threads/Signals.cpp:548
#4 <signal handler called>
#5 bun.js.api.FFIObject.Reader.u8 (globalObject=0x4000554e0088) at /var/lib/buildkite-agent/builds/ip-172-31-75-92/bun/bun/src/bun.js/api/FFIObject.zig:65
#6 bun.js.jsc.host_fn.toJSHostCall__anon_1711576 (globalThis=0x4000554e0088, args=...) at /var/lib/buildkite-agent/builds/ip-172-31-75-92/bun/bun/src/bun.js/jsc/host_fn.zig:97
#7 bun.js.jsc.host_fn.DOMCall("Reader"[0..6],bun.js.api.FFIObject.Reader,"u8"[0..2],.{ .reads = .{ ... }, .writes = .{ ... } }).slowpath (globalObject=0x4000554e0088, thisValue=70370172175040, arguments_ptr=0xfffff6015460, arguments_len=1) at /var/lib/buildkite-agent/builds/ip-172-31-75-92/bun/bun/src/bun.js/jsc/host_fn.zig:490
#8 0x000040003419003c in ?? ()
#9 0x0000400055173440 in ?? ()
```
</details>
I used GDB instead of LLDB (as the branch name suggests) because it
seems to produce more useful stack traces with musl libc.
- [x] on linux, use gdb to print from core dump of main bun process
crashed
- [x] on linux, use gdb to print from all new core dumps (so including
bun subprocesses spawned by the test that crashed)
- [x] on all platforms, use a mini bun.report server to print a
self-reported trace (depends on oven-sh/bun.report#15; for now our
package.json points to a commit on the branch of that repo)
- [x] fix trying to fetch stack traces too early on windows
- [x] use output groups so the traces show up alongside the log for the
specific test instead of having to find it in the logs from the entire
run
- [x] get oven-sh/bun.report#15 merged, and point to a bun.report commit
on the main branch instead of the PR branch in package.json
### How did you verify your code works?
Manually, and in CI with a crashing test.
---------
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
