fix(runtime): keepalive anchors as #[used(compiler)] — stop pinning the whole runtime into every binary (−15% typical, −21.7% auto-optimized) - #10382
Closed
proggeramlug wants to merge 1 commit into
Conversation
Plain #[used] on Mach-O sets N_NO_DEAD_STRIP, making every KEEP_* anchor an unconditional ld64 root. With codegen-units = 1 perry-runtime is a single archive member, so any runtime reference pulls all anchors and keeps the whole runtime surface alive. #[used(compiler)] keeps the anchor (and its target) in the archive for rustc/LTO without making it a linker root. Left as linker roots: mimalloc __mod_init_func, OHOS .init_array, build stamp. Local experiment branch; not for merge as-is.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Advanced Run ID: 📒 Files selected for processing (113)
Included review availability: Your plan provides up to 8 included reviews per hour; 5 remain after this review. 📝 WalkthroughWalkthroughThe pull request enables ChangesKeepalive anchor migration
Priority: ➖ Normal Estimated code review effort: 3 (Moderate) | ~25 minutes Change: Bug fix Merge Risk: ⚪ Minimal · up to The reported missing feature gates are already present, so no actionable regression remains from this change. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
This was referenced Sep 16, 2026
proggeramlug
pushed a commit
that referenced
this pull request
Sep 17, 2026
Contributor
Author
|
Landed via merge train #10401 (v0.5.1587). All source commits preserve authorship; merged main matches the validated train exactly. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
The ~490 keepalive anchor statics are declared with plain
#[used]. On Mach-O that setsN_NO_DEAD_STRIP, and on ELF it puts the static in a section flaggedSHF_GNU_RETAIN— i.e. every anchor is an unconditional linker root on both platforms.[profile.release]setscodegen-units = 1, soperry-runtimecompiles to one 17.4 MB archive member holding all 504KEEP_*statics. Because an.amember is pulled in whole, referencing any runtime symbol (e.g.console.log) drags in all of them, and each one roots whatever it points at.Cargo.tomlandoptimized_libs/freshness.rsboth argue the cost is bounded:That holds for a multi-object archive. Under
codegen-units = 1there is only one object, so the bound degrades to "the entire runtime surface". Measured withld64 -why_live, aconsole.log("yeah")program retainsfs,child_process,dgram,tls,node_vm,bun_ffi, yoga/taffy and more, via:Fix
Declare the anchors
#[used(compiler)]instead. This keeps the static (and its target) through rustc/LLVM — which is the reason the anchors exist, per #6917's revert — while leaving it strippable by the linker, so the program's own undefined references decide what survives.Deliberately left as plain
#[used]:PERRY_RUNTIME_BUILD_STAMP_EMBEDDED, mimalloc's__DATA,__mod_init_funcentry, and OHOS's.init_arrayentry.perry-ext-http'sFORCE_LINK_HTTP_SERVER(#1652) is untouched — it is intentional linker retention.The archive stays complete
This is the failure #6917 was reverted for, so it is checked directly rather than argued:
libperry_runtime.aKEEP_*statics present in the object[no dead strip]rootsPrograms that call the symbols named in those comments link and run:
js_box_release,js_bool_box_release,js_i32_box_release,js_closure_set_box_capture_ptr,js_link_path_module_parent(CJSrequire),js_module_dynamic_import_deferred(computedawait import(spec)).Measurements
13 programs on macOS/arm64, 12 on Linux/x86_64; stdout, stderr and exit code byte-identical to baseline in every case, provenance checked from each binary's embedded build stamp.
console.log, classes, closures, async, errors, generators, ESMevents/http/net/zlib)import()/ CJSrequirerich.tswith auto-optimize (the shipping path)perrycompiler binary itself (Linux)Linux is within a few tenths of a percent of macOS on every row.
Tests
perry --bin perry: 1129 passed, 0 failed — includesauto_optimize_always_includes_keepalive_anchors,auto_optimize_keepalive_anchors_not_bitcode_only,ext_crates_bundle_a_full_featured_perry_runtime.perry-runtime --lib/perry-stdlib --lib, single-threaded on both arms: identical results — 3952 passed and the same single pre-existing failure,gc::tests::heap_generation::a_free_or_move_outside_every_scope_is_caught_in_debug_builds, which fails the same way on unmodifiedfcd108bfb0. The extra failures seen in parallel runs are flaky and differ in membership between arms (baseline failscached_reads_preserve_identityandbuffered_stdin_is_delivered, which this branch passes).assert_lto_keepalive_anchormatches the literal text#[used], so it is widened to accept#[used(compiler)].Risks / not covered
#![feature(used_with_arg)]toperry-runtimeandperry-stdlib(both crates already require the pinned nightly).#[no_mangle]function in its staticlib, which suggests the anchors may no longer be needed at all. Not tested against perry itself, so this PR makes the conservative change.Summary by CodeRabbit