No idea what I'm doing or if this is the right fix, a good test, or if
this metadata should fail the verifier. I've had this crashing
testcase sitting around for months so I'm not entirely sure where it
came from, but I think I was manually reducing the metadata.
Details
Details
- Reviewers
- None
Diff Detail
Diff Detail
Unit Tests
Unit Tests
Event Timeline
Comment Actions
@aprantl - validator's more your wheelhouse than mine, but I'd say if we don't have a specific use case for this we should default to making it invalid/fail the verifier for now - can widen that if/when needed.
Comment Actions
It surprises me that the assembler doesn't seem to require a type either. I think I agree with David that it would be better to let the Verifier fail here and require a type. For reference, this is how clang produces void():
!9 = distinct !DISubprogram(name: "f", scope: !10, file: !10, line: 1, type: !11, scopeLine: 1, spFlags: DISPFlagDefinition, unit: !5, retainedNodes: !7) !10 = !DIFile(filename: "/tmp/t.c", directory: "") !11 = !DISubroutineType(types: !12) !12 = !{null}
clang-format: please reformat the code