This is the mail archive of the
mailing list for the GDB project.
Re: [patch] Fix ELF stale reference
- From: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- To: gdb-patches at sourceware dot org
- Cc: Tom Tromey <tromey at redhat dot com>, Doug Evans <dje at google dot com>
- Date: Thu, 14 Oct 2010 18:06:53 +0200
- Subject: Re: [patch] Fix ELF stale reference
- References: <20100908185837.GA24606@host1.dyn.jankratochvil.net> <email@example.com> <20100909090511.GA937@host1.dyn.jankratochvil.net> <AANLkTikm4c_dPTiqA8hsw8AkNbpKbyQNxC3qgRJi98Au@mail.gmail.com> <20100909145615.GA5771@host1.dyn.jankratochvil.net>
I was debugging https://bugzilla.redhat.com/show_bug.cgi?id=642879 and got to
this fix from a different side. It is in fact a very common GDB crash - due
to CTRL-C hit (to get GDB prompt) in the moment an ELF file is being read in.
Original thread: http://sourceware.org/ml/gdb-patches/2010-09/msg00192.html
On Thu, 09 Sep 2010 16:56:15 +0200, Jan Kratochvil wrote:
> OTOH this patch is not completely clean, it can needlessly allocate
> bfd-associated memory and the right fix would probably span into bfd/ IMO.
While the memory could use for example register_objfile_data_with_cleanup
instead of bfd_alloc so that if errors/CTRL-Cs happen the dynamic symbol table
pointers memory is not allocated twice. Still I would not find it correct as
such memory would be objfile-bound instead of abfd-bound - while being
referenced by abfd.
OK to check-in? Or some bfd/ API improvement should be made?
2010-09-09 Jan Kratochvil <firstname.lastname@example.org>
Fix stale memory references.
* elfread.c: Include libbfd.h.
(elf_symfile_read): Replace xmalloc by bfd_alloc, drop xfree, new
@@ -37,6 +37,7 @@
extern void _initialize_elfread (void);
@@ -792,8 +793,14 @@ elf_symfile_read (struct objfile *objfile, int symfile_flags)
if (storage_needed > 0)
- dyn_symbol_table = (asymbol **) xmalloc (storage_needed);
- make_cleanup (xfree, dyn_symbol_table);
+ /* Memory gets permanently referenced from ABFD after
+ bfd_get_synthetic_symtab so it must not get freed before ABFD gets.
+ It happens only in the case when elf_slurp_reloc_table sees
+ asection->relocation NULL. Determining which section is asection is
+ done by _bfd_elf_get_synthetic_symtab which is all a bfd
+ implementation detail, though. */
+ dyn_symbol_table = bfd_alloc (abfd, storage_needed);
dynsymcount = bfd_canonicalize_dynamic_symtab (objfile->obfd,