rnk added a comment. I'd be happy to take this patch, address the comments, and land it, if you don't want to deal with all the nits.
================ Comment at: test/CodeGen/debug-info-no-location.cpp:1 +// RUN: %clang -cc1 -emit-llvm %s -gcodeview -debug-info-kind=limited -o - | FileCheck %s +// CHECK: call void @"??1?$c@UB@@@@QEAA@XZ"(%struct.c* @"?f@?1??d@@YAPEAU?$c@UB@@@@XZ@4U2@A"), !dbg !51 ---------------- I think the `%clang_cc1` substitution is preferred, it sets the resource directory. ================ Comment at: test/CodeGen/debug-info-no-location.cpp:4 +// CHECK-NEXT: ret void, !dbg !51 +// CHECK: !48 = distinct !DISubprogram(name: "`dynamic atexit destructor for 'f'", scope: !3, file: !3, line: 16, type: !49, scopeLine: 16, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !27, retainedNodes: !29) +// CHECK: !51 = !DILocation(line: 16, scope: !48) ---------------- aprantl wrote: > Please don't hardcode metadata numbers in the tests, this is bound to break > almost immediately. > > It seems like this function is compiler-generated and thus should be marked > as artificial? I think using an artificial location (line 0) is consistent with other calls to StartFunction in CGDeclCXX. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D66328/new/ https://reviews.llvm.org/D66328 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
