A dev build splits a large program into LLVM shards, compiles them in
parallel and merges the objects with ld -r. On a native Linux host the
merge then ran objcopy --keep-global-symbols to turn the shards' promoted
symbols back into locals. A sanitized build of three.js's node materials
merges into a 660 MB object with 278k sections, and objcopy spent 425 s
of a 616 s build on it. localizeElfObject, which cross-ELF builds already
use, does the same demotion in about 2 s.
- Program-shard merges for executables, library archives and packed
libraries use localizeElfObject after ld -r on native Linux. Library
runtime localization keeps objcopy.
- localizeElfObject reads and writes extended section numbering: with
SHN_LORESERVE or more sections the counts go in section 0, and symbol
section indices from SHN_LORESERVE up go in .symtab_shndx. Before, it
truncated e_shnum and read such indices as reserved values.
- The Linux shard-merge cache identity drops objcopy.
A value that contradicts a type an npm package declares used to throw
"expected number at $, got undefined" and nothing else. Such a check now
names the declaration and the type patch that admits the value:
profilekit@3.1.0 declares Profile.email as string (index.d.ts:7)
if that declaration is wrong, patch it in package.json: ...
Origins are found for values a package function returns under a written
return type, for properties of records typed by a package-declared type
(record projections into such a type are keyed by it), and for
declaration-boundary casts, whose origin the projection records from the
declaration-world checker. Generic declarations get none: their types
come from the program's own instantiation.
Checked conversions carry dynCheck.origin, an index into
IrModule.declOrigins; the site's existing cold exception block calls
scr_type_origin_note (new runtime unit behind the typeOrigins feature),
which appends the note to the pending TypeError and derives the selector
suffix and the admitted kind from the failure. Union extractions get the
note per arm at compile time. The hot path is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Emitted functions report an exception by returning with the thread's
pending flag set, and every caller tests the flag after each call that
may throw. On Linux executables (glibc and musl), functions reached only
by direct calls now propagate exceptions by unwinding the native stack
instead:
- unwindAbiFunctions selects the may-throw functions that are not async
or generators, constructors, virtually called or overridden methods,
closure targets, the entry, library exports, module initializers or
callees of coroutine bodies.
- A native unwind costs about 2 µs a raise and 0.1 µs a frame, so functions that
exceptions are likely to cross keep the flag: those with a `throw` of
their own, those up to three calls above one, and recursive functions,
counting calls through methods, closures and callbacks.
- Their propagating exits call scr_unwind_raise (scr_unwind.c), a forced
unwind through _Unwind_ForcedUnwind with the exception still in the
active cell.
- Direct calls to them are `invoke`s without a pending check. The
landing pad runs the frame's usual unwind path: it releases the frame's
values and then enters the enclosing try handler or resumes the unwind.
- Remaining plain calls to them (wrappers, adapters, runtime-facing entry
points) become invokes whose landing pad continues on the pending flag,
or resumes when the wrapper is itself invoked. Every other call is
marked `nounwind`, so inlining a raising function at an invoke keeps its
calls plain.
- When emitted text breaks these rules (a raising function's address is
taken, or an invoked function returns with the flag pending), the
module is emitted again with those functions on the pending flag.
Emitted code does not declare scr_unwind_raise `noreturn`, so sanitized
builds do not run the sanitizer's no-return hook, which needs stack
bounds that fiber stacks do not report. Instead, scr_unwind_raise clears
the shadow of the frames it unwinds.
Executables link a static unwinder: libgcc_eh with the host toolchain and
the glibc runtime packs, and Zig's libunwind with Zig and the musl packs.
macOS, Windows, Android, WASI, libraries and debug builds keep the
pending flag everywhere and emit the same code as before.
The throw-chains runtime workload measures exceptions caught 2, 5 and 20
recursive calls above the throw, through direct calls and a callback,
and a parser that backtracks by throwing.
In JavaScript dependencies, a createRequire require call with a computed
specifier, and require.resolve called with a computed request or read as
a value, now compile. The new runtime unit scr_require.c resolves the
specifier with Node 24's CommonJS algorithm (builtins, relative and
absolute paths with Node's extension search, package "main", "exports"
and self-reference, node_modules and NODE_PATH lookup) when the call runs:
- require.resolve returns the id or real path Node returns, and throws
Node's MODULE_NOT_FOUND error when nothing resolves.
- require throws the same MODULE_NOT_FOUND error when nothing resolves,
so packages that probe for optional addons, such as pi-tui, take their
absent branch as in Node. When the specifier resolves to a native addon,
a builtin or another file, Node would load that module: the program
prints an SC2020 error naming it and exits with status 1, without
running the catch clause around the call.
Coverage lists these calls as deferred to runtime. TypeScript sources keep
the compile-time refusal.
Refs #1777.
A codec stored in untyped storage (hono's StreamingApi keeps
`this.encoder = new TextEncoder()` in a JavaScript class field) had no
dynamic representation and threw SC1101 at run time. The codec records
now cross 'unknown' slots as handles of two new tags that hold the record
itself, so identity and a decoder's streaming state survive the round
trip, and checked casts unbox the same record.
The handles answer encode, encodeInto and decode (with options objects),
the encoding, fatal and ignoreBOM properties, method values with Node's
ERR_INVALID_THIS, instanceof, String() tags and JSON's {}. The unit
scr_text_codec_dyn.c links only into programs that convert a codec to or
from an untyped value (the textCodecDyn runtime feature, found with the
dgram carrier walk, now shared).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's FinalizationRegistry took handle tag 72, so the Node island
handle moves to 73. Island handles keep their instanceof path in the
new scr_bytes_instanceof_at.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The kept LLVM text and IR are named after the entry (`<stem>.ll`,
`<stem>.ir.json`) in the output directory, so two programs whose entries
share a name and are built into one directory share those files. The
executable cache published the translation unit by copying the shared
`.ll` after the native build, so when the other program wrote it in
between, the stamp recorded the other program's text, and a later cache
hit relinked the wrong program without any error.
A build now writes, compiles and caches private copies beside the shared
names and renames them into place when it ends. A cache hit that relinks
compiles a private copy of the cached translation unit, and leaves a
shared file that already holds those bytes as it is. The native build
reads the private copy and keeps the shared name as the spelling clang
sees, so native cache keys and debug names do not change.
Malte Ubl