| Summary: | Decode DWARF5 in eu-readelf | ||
|---|---|---|---|
| Product: | elfutils | Reporter: | Ben Woodard <woodard> |
| Component: | tools | Assignee: | Not yet assigned to anyone <unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | elfutils-devel, mark |
| Priority: | P2 | ||
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Host: | Target: | ||
| Build: | Last reconfirmed: | ||
| Attachments: | clang compiled object that includes the example above | ||
Although it isn't full DWARF5 support all DWARF5 tags and attributes are recognized in 0.170. From the NEWS file:
Version 0.170
libdw: Added new DWARF5 attribute, tag, character encoding, language code,
calling convention, defaulted member function and macro constants
to dwarf.h.
New functions dwarf_default_lower_bound and dwarf_line_file.
dwarf_peel_type now handles DWARF5 immutable, packed and shared tags.
dwarf_getmacros now handles DWARF5 .debug_macro sections.
And the above example now shows:
[ 7e1] namespace
name (strp) "__cxx11"
decl_file (data1) 1
decl_line (data2) 2214
export_symbols (flag_present) yes
|
Created attachment 10274 [details] clang compiled object that includes the example above We are starting to use Clang 4.0.0 at LLNL and it is emitting some DWARF5 opcodes. Please add DWARF5 decoding to eu-readelf. For example: #include <string> #include <vector> std::vector<std::string> var; and tangentially I saw something in the DWARF which neither readelf or eu-readelf understands. [ 7e1] namespace name (strp) "__cxx11" decl_file (data1) 1 decl_line (data2) 2214 ??? (0x89) (flag_present) yes Since clang is already emitting some DWARF5 we should at least be able to decode what it is emitting reliably.