Hi,
I think we are almost ready to release 0.159. I updated the release
related files on master and merge master to the portable branch. This is
the current NEWS entry:
Version 0.159
stack: New option -d, --debugname to lookup DWARF debuginfo name for frame.
New option -i, --inlines to show inlined frames using DWARF debuginfo.
libdwelf: New libdwelf.h header for libdw.so DWARF ELF Low-level Functions.
New function dwelf_elf_gnu_debuglink, dwelf_dwarf_gnu_debugaltlink,
and dwelf_elf_gnu_build_id.
libdw: Support for DWZ multifile forms DW_FORM_GNU_ref_alt and
DW_FORM_GNU_strp_alt is now enabled by default and no longer
experimental. Added new functions dwarf_getalt and dwarf_setalt
to get or set the alternative debug file used for the alt FORMs.
The dwfl_linux_proc_find_elf callback will now find ELF from
process memory for (deleted) files if the Dwfl has process state
attached.
libdwfl: The dwfl_build_id_find_debuginfo and dwfl_standard_find_debuginfo
functions will now try to resolve and set the alternative debug file.
backends: Add CFI unwinding for arm. Relies on .debug_frame.
Add arm process initial register state compatible mode to AARCH64.
Add aarch64 native and core unwind support.
other: All separate elfutils-robustify patches have been merged.
CVE-2014-0172 Check overflow before calling malloc to uncompress data.
Tested against fedora rawhide on aarch64, i686, x86_64, s390, s390x and
ppc64 without any FAILS. And on armv7hl with one FAIL.
We added elflint checks against .o ET_REL files which produce on arm:
section [ 8] '.rel.ARM.exidx': invalid destination section type
I'll see if I can figure out what that should be, but I don't think it
is a showstopper because it isn't really a regression.
Tested against RHEL6 on ppc, ppc64, i686, s390, s390x and x86_64 without
any FAILS.
portable branch tested against RHEL5 on ppc64, ppc, ia64, x86_64, i386,
s390 and s390x. All had one failure in the backtrace-demangle test. The
issue is that the test uses local/static functions and apparently the
mangling for that changed (the abi only defines mangling for public
functions). I have a fix for that, which I'll post and integrate before
release.
On RHEL5 ia64 elflint also failed with two issues:
section [13] '.rela.IA_64.unwind': invalid destination section type
Again in an ET_REL file, so it looks like the armv7hl failure above.
section [ 2] '.dynsym': symbol 1: symbol in dynamic symbol table with
non-default visibility
I haven't investigated this one yet.
I'll see if I can figure them out and fix them, but I don't think the
ia64 elflint issues are regressions, so they should not be showstoppers.
Please let me know if there are any regressions you want to fix before
the 0.159 release.
Thanks,
Mark
