zequanwu added a comment.
In D138326#3938128 <https://reviews.llvm.org/D138326#3938128>, @rnk wrote:
> Since Clang creates the member list and only adds types for the benefit of
> codeview, should we change clang instead?
Yes, that's something clang does only for codeview, changed in clang side. It
will still emit type info for unnamed records but not for anonymous records.
================
Comment at: llvm/test/DebugInfo/COFF/nested-types.ll:214
+; CHECK-NEXT: }
; CHECK-NEXT: NestedType {
; CHECK-NEXT: TypeLeafKind: LF_NESTTYPE (0x1510)
----------------
rnk wrote:
> So the test shows that we don't have NestedType records for unnamed unions &
> structs, right?
Updated. We want NestedType records for unnamed unions & structs, but not for
anonymous unions & structs. That's what MSVC does.
```
0x1004 | LF_FIELDLIST [size = 168]
- LF_MEMBER [name = `i1`, Type = 0x0074 (int), offset = 0, attrs =
public]
- LF_NESTTYPE [name = `<unnamed-type-unnamed_union>`, parent =
0x1002]
- LF_MEMBER [name = `unnamed_union`, Type = 0x1002, offset = 4,
attrs = public]
- LF_MEMBER [name = `i3`, Type = 0x0074 (int), offset = 8, attrs =
public]
- LF_NESTTYPE [name = `<unnamed-type-unnamed_struct>`, parent =
0x1003]
- LF_MEMBER [name = `unnamed_struct`, Type = 0x1003, offset = 12,
attrs = public]
```
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D138326/new/
https://reviews.llvm.org/D138326
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
