hctim added a comment. In D123534#3509667 <https://reviews.llvm.org/D123534#3509667>, @rnk wrote:
> That's fine, a Windows machine shouldn't be necessary. This patch has to > change the AsmPrinter for DWARF, so it makes sense that it should do the same > for CodeView. I think it should be sufficient to continue this loop when a > global variable display name is empty: > https://github.com/llvm/llvm-project/blob/main/llvm/lib/CodeGen/AsmPrinter/CodeViewDebug.cpp#L3175 > > It will need a test, and the tests at llvm/test/DebugInfo/COFF/global* should > be good templates. Basically, have a file with two globals, one named and > unnamed, and use FileCheck to confirm that there is only one GDATA record. Thanks for the pointer to the test. That was very helpful in figuring out how to invoke clang to produce CodeView/COFF. Funnily enough, it found a bug where I didn't realise I had to relax the invariant that a `DIGlobalVariable` needs a name in the LL parser. I wonder whether it's feasible to re-use `FunctionLineTable` to encode `line:col` for data symbols. A hack for sure, and left as an exercise for the reader. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D123534/new/ https://reviews.llvm.org/D123534 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
