------- Additional Comments From andrejoh at gmail dot com 2008-04-10 09:08
-------
Nick, I'm not sure reverting the cleanup is correct. At least in binutils
2.18, the function concat_filename uses bfd_malloc, which again uses malloc.
As far as I can see, it does the same in current CVS.
I get leak reports like this when running under Valgrind:
==28969== 120,701 bytes in 1,422 blocks are definitely lost in loss record 2
of 2
==28969== at 0x4904A06: malloc (vg_replace_malloc.c:149)
==28969== by 0x407305: bfd_malloc (libbfd.c:173)
==28969== by 0x46BC09: concat_filename (dwarf2.c:1064)
==28969== by 0x46CBFC: scan_unit_for_symbols (dwarf2.c:1936)
==28969== by 0x46CFFE: comp_unit_find_nearest_line (dwarf2.c:2262)
==28969== by 0x46DD9D: find_line (dwarf2.c:3093)
==28969== by 0x46E4A6: _bfd_dwarf2_find_nearest_line (dwarf2.c:3127)
==28969== by 0x41F3DA: _bfd_elf_find_nearest_line (elf.c:6889)
Running with a change similar to the one in comment #14 plus freeing
caller_file removes the leaks I see when doing call-stack mapping.
(CentOS 4.5/x86_64. using GCC 4 gcc4-4.1.1-53.EL4 and BFD from binutils 2.18.)
--
What |Removed |Added
----------------------------------------------------------------------------
CC| |andrejoh at gmail dot com
Status|RESOLVED |REOPENED
Resolution|FIXED |
http://sourceware.org/bugzilla/show_bug.cgi?id=868
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
_______________________________________________
bug-binutils mailing list
bug-binutils@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-binutils
