Details
- Reviewers
shchenz jhenderson DiggerLin MaskRay - Group Reviewers
Restricted Project - Commits
- rGd29e986ed5c1: [XCOFF] write the real source file name in C_FILE symbol.
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
llvm/lib/MC/XCOFFObjectWriter.cpp | ||
---|---|---|
502 | What if we are compiling a .s file and there are more than one .file in the file? |
llvm/lib/MC/XCOFFObjectWriter.cpp | ||
---|---|---|
987 | llvm coding standard uses pre-increments. |
llvm/test/MC/PowerPC/aix-symtbl-order.s | ||
---|---|---|
4 ↗ | (On Diff #435420) | Do you mean "when support is added to llvm-mc"? "when llvm-mc is supported" implies the application isn't yet supported, but you are actively using it in this test. |
No further comments from me at this time, but I'm not knowledgeable enough about this area to be able to LGTM it.
The AIX assembler doesn't really support multiple .file statements and its output symbol table starts with all the .file symbols.
This update drops the symbol order logic from the previous version in favor of a behavior consistent with the AIX assembly.
llvm/test/MC/PowerPC/aix-symtbl-order.s | ||
---|---|---|
7 ↗ | (On Diff #437443) | We are over design here? Like what we discussed offline, AIX assembler does not have the functionality to guarantee that the local/global symbols are associated with their precedent .file directive. I think it is better for us to do as AIX assembler does, just generate the C_FILE symbols for each .file and don't do any symbol scope related handling. |
8 ↗ | (On Diff #437443) | Nit: for function entry, we need to make it like .foo. |
What if we are compiling a .s file and there are more than one .file in the file?