You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 37a74f6
Browse filesBrowse the repository at this point in the historyBrowse files
fix: a payload that starts through its home's loader runs on the home's runtime, read from its programs; a compiler that did not answer says so
Which toolchain payloads run on the C runtime payload was a list (gcc,
llvm). The emsdk payload's clang names the home's loader link
(<home>/abi) as well; restored from a store cached without that link it
did not start, and nothing pinned the runtime again (mtest 0.8.4's Web
row). runtime_payload_kind now reads a placed payload's programs: one of
bin/ whose PT_INTERP lies inside its own provider home, as the provider's
layout gives it. No name and no system loader is written down; a Mach-O or
PE program names no loader.
A probe that fails with its errors states its last line of output, and
clang's std probe keeps why it went unanswered (stdModuleUnanswered), which
the refusal of `import std` gives.
built_artifact.sh reads the artifact events with grep and sed -E: the
hermetic container has no python.
Design record 2026-11-14 §5.5 (I-L); docs/91 (en, zh); unit cases
RuntimePayload/APayloadStartsThroughItsHome and
RuntimePayload/APayloadOnTheSystemsLoaderDoesNot.
Copy file name to clipboardExpand all lines: docs/91-toolchain-internals.md
+15-3Lines changed: 15 additions & 3 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -327,11 +327,22 @@ from an upstream home runs on that home's runtime; what it produces is the same,
327
327
because the artifact's loader, run paths and sysroot come from the build's own
328
328
command lines (§5).
329
329
330
+
The loader a payload names may be the home's link to it (`<home>/abi/<abi>/`)
331
+
rather than the runtime payload itself. The links are state of the home, made
332
+
by the pin; a store restored without them (a CI cache of `data/xpkgs` alone)
333
+
holds payloads that do not start. So whenever a payload that runs on the
334
+
runtime is taken -- installed, or found already placed -- mcpp asks the
335
+
provider to keep the pin (`ensure_declared_runtime`), which makes the links
336
+
again when they are missing. Which payloads run on the runtime is read from the
337
+
payload's programs, not from its name (2026.11.13.1): one of `bin/` whose
338
+
`PT_INTERP` lies in its own provider home. A Mach-O or PE program names no
339
+
loader, so the question has the same form on every system.
340
+
330
341
**What mcpp still does here** (`post_install.cppm`, all read-only):
331
342
332
343
| function | answer |
333
344
|---|---|
334
-
|`runtime_payload_kind`| whether a toolchain runs on a C runtime payload: `gcc`, `llvm` on Linux; nothing for musl-gcc, a system compiler or a PE toolchain |
345
+
|`runtime_payload_kind`| whether a toolchain runs on a C runtime payload: `gcc`, `llvm` on Linux by their recipes; once placed, any payload whose programs start through its home's loader (`starts_through_its_home`, the emsdk payload among them); nothing for musl-gcc, a system compiler or a PE toolchain |
335
346
|`select_glibc_payload_lib`| the library directory of exactly `glibc@<version>` (never another installed version) |
336
347
|`detect_baked_loader`, `baked_runtime_binding`| the loader and runtime a payload was bound to, read from its ELF headers |
337
348
|`write_clean_link_specs`| the per-build `-specs=` file in the build directory (§5) |
@@ -502,8 +513,9 @@ loader coupling, so a host binary that stumbles onto them is at worst confused.
0 commit comments