fix(runtime): unify the three pthread stack-bounds extern declarations by proggeramlug · Pull Request #9776 · PerryTS/perry

proggeramlug · 2026-09-05T04:42:03Z

The break

main cannot compile perry-runtime under -D warnings, so the warnings job on
https://github.com/PerryTS/perry/actions/runs/33926006467 (main at 12efed12220e) fails
with build failed, waiting for other jobs to finish:

error: `pthread_getattr_np` redeclared with a different signature
   --> crates/perry-runtime/src/gc/roots.rs:761:9
761 |         fn pthread_getattr_np(thread: usize, attr: *mut [u64; 8]) -> i32;
    |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ this signature doesn't match the previous declaration
    ::: crates/perry-runtime/src/error_stack_frames.rs:200:13
200 |             fn pthread_getattr_np(thread: usize, attr: *mut u8) -> i32;
    |             ------------------------------------------------------- `pthread_getattr_np` previously declared here
    = note: expected `unsafe extern "C" fn(usize, *mut u8) -> i32`
               found `unsafe extern "C" fn(usize, *mut [u64; 8]) -> i32`
    = note: `-D clashing-extern-declarations` implied by `-D warnings`
error: could not compile `perry-runtime` (lib) due to 3 previous errors

Same for pthread_attr_getstack and pthread_attr_destroy.

Root cause

perry-runtime declares these three libc functions in three separate extern "C"
blocks, in two different spellings:

site	attr spelling	cfg
`gc/roots.rs:757` `get_stack_bottom`	`[u64; 8]`	`target_os = "linux"`
`gc/roots/stack_maps.rs:1788` `fp_chain::stack_top`	`*mut u8`	`target_os = "linux"` and `target_arch = "aarch64"`
`error_stack_frames.rs:197` `stack_top_uncached`	`*mut u8`	`target_os = "linux"` (no arch gate)

The *mut u8 spelling is not new — but the only site that used it was inside
mod fp_chain, which is gated to target_arch = "aarch64". On CI's linux-x86_64
host that module is cfg'd out, so the two spellings never met and the lint never fired.

Culprit: 1ebc65e87 — "feat(runtime): real function names in Error stacks — frame-pointer
walk + the existing name registry (#9486) (#9521)" (2026-09-02). It added a third copy of
the block in error_stack_frames.rs gated only on target_os = "linux". With no arch gate
it collides with gc/roots.rs on x86_64, and the crate stopped compiling under -D warnings.

This is macOS-invisible: every declaration involved is behind cfg(target_os = "linux").

The fix

All three blocks now use the [u64; 8] spelling that gc::roots::get_stack_bottom — the
oldest and most-shipped of the three — has always used. This also clears the latent
aarch64-linux collision between fp_chain and gc/roots.rs, which would have broken that
target the moment anyone built it with -D warnings.

[u64; 8] is 64 bytes: at least the size of pthread_attr_t on every supported glibc/musl
target (56 bytes on both), and correctly aligned for it — which the [u8; 128] buffer it
replaces, with alignment 1, was not. No behavioural change otherwise; all three functions
compute the same stack-high address they did before.

Verification

Minimised standalone repro: the two spellings in one crate produce exactly the two
redeclared with a different signature errors under -D warnings; unified, rustc
exits 0.
RUSTFLAGS="-D warnings" cargo check -p perry-runtime --all-targets on a linux-x86_64
host reproduces the 3 errors at 12efed12220e and passes with this commit.

Not fixed here

main is red for four further, independent reasons — none of them this compile error, and
none of them fixed by this PR. Detailed separately:

check — API-docs drift: Bun.connect/Bun.listen were added to the API manifest by
868787448 ("feat(bun): add TCP socket facades") without re-running scripts/regen_api_docs.sh.
ext-link (red since 2026-09-04) — js_bun_tcp_listen calls perry_ffi::run_pending,
and crates/perry-ext-http/src/test_async_shims.rs has no perry_ffi_run_pending stub.
Same culprit commit 868787448.
cargo-test — commands::compile::build_cache::tests::codegen_env_vars_are_build_cache_inputs.
gc-stress matrix (4/4) / gap-suite (2) — output mismatches in
test_gap_repsel_gc_stress, test_gap_perfhooks_3088_3008_3010_3011 and
test_gap_prop_plan_cache_invalidation.

gc-stress and main-gate are aggregator jobs (14s / 3s) and go green once their
dependencies do.

https://claude.ai/code/session_01YPfnmWZmSpSWpmnoXvH8z2

Summary by CodeRabbit

Bug Fixes
- Fixed Linux builds with warnings treated as errors.
- Improved compatibility across supported Linux architectures by aligning system-level stack handling declarations.
- Corrected buffer alignment and sizing used during thread stack inspection.

`perry-runtime` declared `pthread_getattr_np`, `pthread_attr_getstack` and `pthread_attr_destroy` in three separate `extern "C"` blocks using two different argument spellings. Declaring one symbol twice in a crate with different signatures is `clashing_extern_declarations`, which `-D warnings` denies, so the `warnings` job could not compile the crate at all: error: `pthread_getattr_np` redeclared with a different signature --> crates/perry-runtime/src/gc/roots.rs:761:9 ::: crates/perry-runtime/src/error_stack_frames.rs:200:13 = note: expected `unsafe extern "C" fn(usize, *mut u8) -> i32` found `unsafe extern "C" fn(usize, *mut [u64; 8]) -> i32` = note: `-D clashing-extern-declarations` implied by `-D warnings` error: could not compile `perry-runtime` (lib) due to 3 previous errors `gc::roots::get_stack_bottom` has spelled the attr buffer `[u64; 8]` since long before the break. The competing `*mut u8` spelling also predates it, in `gc::roots::stack_maps::fp_chain`, but that module is gated to `target_arch = "aarch64"`, so on CI's linux-x86_64 host the two never met. 1ebc65e (PerryTS#9486/PerryTS#9521) added a third copy in `error_stack_frames.rs` gated only on `target_os = "linux"`; with no arch gate it collides with `gc/roots.rs` on x86_64, and the crate stopped compiling under `-D warnings`. All three blocks now use the `[u64; 8]` spelling, which additionally clears the latent aarch64-linux collision between `fp_chain` and `gc/roots.rs`. `[u64; 8]` is 64 bytes — at least the size of `pthread_attr_t` on every supported glibc/musl target, and correctly aligned for it, which the `[u8; 128]` buffer it replaces was not. Unbreaks the `warnings` job on https://github.com/PerryTS/perry/actions/runs/33926006467 Claude-Session: https://claude.ai/code/session_01YPfnmWZmSpSWpmnoXvH8z2

Claude-Session: https://claude.ai/code/session_01YPfnmWZmSpSWpmnoXvH8z2

coderabbitai · 2026-09-05T04:44:33Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Team

Run ID: 1e259a73-684d-44d8-96b6-b67ef1a4f609

📥 Commits

Reviewing files that changed from the base of the PR and between 12efed1 and 7ce6e9f.

📒 Files selected for processing (3)

changelog.d/9776-pthread-extern-signature-clash.md
crates/perry-runtime/src/error_stack_frames.rs
crates/perry-runtime/src/gc/roots/stack_maps.rs

Included review availability: Your plan provides up to 8 included reviews per hour; 3 remain after this review.

📝 Walkthrough

Walkthrough

The Linux pthread declarations now use a consistent [u64; 8] attribute representation and typed stack-address pointers. The changes remove conflicting extern signatures, provide correct alignment, and document the fix.

Changes

Linux pthread stack binding updates

Layer / File(s)	Summary
Align pthread stack bindings `crates/perry-runtime/src/error_stack_frames.rs`, `crates/perry-runtime/src/gc/roots/stack_maps.rs`, `changelog.d/9776-pthread-extern-signature-clash.md`	The stack discovery paths now use matching `[u64; 8]` pthread attribute buffers and typed `u8` stack-address pointers. Existing cleanup and error handling remain unchanged. The changelog records the `clashing_extern_declarations` fix.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: ⚪ Minimal · up to 7ce6e

This change unifies Linux pthread stack-discovery bindings, restoring warning-free compilation and removing inconsistent extern declarations. No merge-blocking production risk is identified.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely describes the main change: unifying the three pthread stack-bounds extern declarations.
Description check	✅ Passed	The description is detailed and on-topic. It explains the failure, root cause, fix, verification, and unrelated CI failures. It does not use the template headings and omits the Related issue and Check…
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 2 files. (1 skipped: 1 …
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

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

proggeramlug · 2026-09-05T07:32:05Z

Conflicts with merge train #9798, which just landed 19 PRs (including #9750's rework of gc_runtime_root_holders.py and a regex.rs split under the 2000-line cap). Could you rebase onto current main? I'd rather you resolve it than have me hand-merge — several of these touch GC root scanning or regex internals where the two changes are independent rewrites of the same code, and that's exactly where a mechanical merge goes quietly wrong. Everything that picked clean is in the next train; I'll pick these up as soon as they rebase.

proggeramlug · 2026-09-05T17:47:55Z

Superseded — closing. This PR unifies the three pthread stack-bounds extern declarations, but main already does that via 3bfbe40e9 (#9752, landed in train #9798), which introduced the shared native_stack::stack_top() helper. Only two files declare pthread_getattr_np now (native_stack.rs itself and gc/roots/stack_bottom.rs), down from three, and callers go through the helper rather than repeating the extern block per file — which is the stronger form of what this PR proposed. Nothing here is lost.

Ralph Küpper added 2 commits September 5, 2026 06:40

docs(changelog): add PR 9776 fragment

7ce6e9f

Claude-Session: https://claude.ai/code/session_01YPfnmWZmSpSWpmnoXvH8z2

This was referenced Sep 5, 2026

fix(compile): key the build cache on PERRY_CONCAT_SITE_CACHE #9777

Closed

fix: complete 868787448's Bun TCP landing (API docs drift + ext-link shim) #9780

Open

fix(gc): route the census thread-locals through hot TLS #9781

Closed

This was referenced Sep 5, 2026

perf(shapes): append a freshly allocated descriptor id without the family membership scan #9768

Closed

Merge train: #9762, #9768–#9797 (11 PRs) + gate follow-ups #9804

Merged

perf(gc): young-entry logs for the side-table root scanners #9755

Closed

proggeramlug closed this Sep 5, 2026

Navigation Menu

Uh oh!

Uh oh!

fix(runtime): unify the three pthread stack-bounds extern declarations - #9776

fix(runtime): unify the three pthread stack-bounds extern declarations#9776
proggeramlug wants to merge 2 commits into
PerryTS:mainfrom
proggeramlug:fix/main-green-9730

proggeramlug commented Sep 5, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented Sep 5, 2026 •

edited

Loading

Walkthrough

Changes

Uh oh!

proggeramlug commented Sep 5, 2026

Uh oh!

proggeramlug commented Sep 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

proggeramlug commented Sep 5, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

The break

Root cause

The fix

Verification

Not fixed here

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Sep 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

proggeramlug commented Sep 5, 2026

Uh oh!

proggeramlug commented Sep 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

proggeramlug commented Sep 5, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Sep 5, 2026 •

edited

Loading