zturner added inline comments.
================
Comment at: source/Plugins/ObjectFile/PECOFF/ObjectFilePECOFF.cpp:183
@@ +182,3 @@
+ }
+ return lldb::eSymbolTypeInvalid;
+}
----------------
amccarth wrote:
> zturner wrote:
> > Instead of returning `eSymbolTypeInvalid` here, how about
> > `eSymbolTypeData`? If you look in `llvm/Support/COFF.h` all of the non
> > function types are data.
> I'm not sure about that. Some of the symbols are clearly sections (.text,
> etc.). Those have 0 for the COFF type, which COFF.h says means "No type
> information or unknown base type." If .text has a "valid" symbol type, then
> it will be found (instead of, say, "_main"), and I'm not sure if the
> unwinding/stepping would work right.
How about:
if (coff_symbol_type == 0)
return lldb::eSymbolTypeInvalid;
if (coff_symbol_type >> llvm::COFF::SCT_COMPLEX_TYPE_SHIFT) ==
llvm::COFF::IMAGE_SYM_DTYPE_FUNCTION)
return lldb::eSymbolTypeCode;
return lldb::eSymbolTypeData;
I feel like we can at least do a little better than always returning
`eSymbolTypeInvalid` if it's not a function without too much work.
http://reviews.llvm.org/D16563
_______________________________________________
lldb-commits mailing list
[email protected]
http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
