There was some duplication here, and also a potential
memory management issue; it appears that we were
converting a temporary std::string to a StringRef
when setting the section name of a global.
Ensure they get emitted at the end of the job by the dispatcher, and
also use a proper mangling and shared linkage for these symbols so
that if multiple threads emit the same descriptor it gets merged.
The new tests attempt to exercise these scenarios.
Fixes <rdar://problem/27906876>.
Rather than collection nominal type and extension decls and emit
reflection metadata records in one go, we can emit them as they
are encountered and instead collection builtin types referenced
by those at the end.
This test was mostly to make sure -strip-reflection-metadata
and -strip-reflection-names does what it says, so I'm temporarily
removing the check for the full constant reflection struct.
- Implement emission of type references for nominal type field
reflection, using a small custom encoder resulting in packed
structs, not strings. This will let us embed 7-bit encoded
32-bit relative offsets directly in the structure (not yet
hooked in).
- Use the AST Mangler for encoding type references
Archetypes and internal references were complicating this before, so we
can take the opportunity to reuse this machinery and avoid unique code
and new ABI.
Next up: Tests for reading the reflection sections and converting the
demangle tree into a tree of type references.
Todo: For concrete types, serialize the types for associated types of
their conformances to bootstrap the typeref substitution process.
rdar://problem/15617914
