TPPPP72 wrote: > Rather than fixing the symptom, this PR needs to fix the source. > > The test case in the original actually has an error, and often an error > results in an invalid data. So I looked at the code you're working with, and > the real cause of this crash is that we're trying to do this kind of analysis > on an invalid object. > > I think the correct fix will be something like checking for an invalid decl > before trying to process documentation comments, and possibly emitting a > warning or error if you think it helps.
My modification is based on shafik's solution in the original issue, but from this perspective, a null pointer shouldn't exist here in the first place. Is it feasible to trace back to getParamVars? https://github.com/llvm/llvm-project/pull/183274 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
