[Bug dynamic-link/20105] bad version variable in elf_dynamic_do_Rel() elf/do-rel.h (2.23) causes coredump in dl-machine.h elf_machine_rela()

["jason.vas.dias at gmail dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=20105

--- Comment #17 from Jason Vas Dias <jason.vas.dias at gmail dot com> ---
Yes, but please define "non corrupt system" - I cannot progress farther in
diagnosing which library is corrupt in this case without your link map -
but I think it is OOO's copy of libxml2.so, as in my previous post:
" /usr/fw/ooo-4.1.2/opt/openoffice4/program/libxml2.so.2: no version
information available (required by
/mnt/oel7/usr/lib64/libgnomevfs-2.so.0)
".
The only "corruption" is a library shipped by OOO which does not have
version information (its old version of libxml2.so ).
If glibc GIT master can handle loading this library when in the presence of a
later version of libxml2 in its system ld path ( which is /usr/lib64 (
+/usr/lib32 for 32-bit) ) only on my system ) without core-dumping ,
I'd be happy to build the
master & give it a try - is this really the case ?
 When is glibc-2.24 being released ?


On 18/05/2016, carlos at redhat dot com
<sourceware-bugzilla@sourceware.org> wrote:
> https://sourceware.org/bugzilla/show_bug.cgi?id=20105
>
> --- Comment #16 from Carlos O'Donell <carlos at redhat dot com> ---
> (In reply to Jason Vas Dias from comment #15)
>> I really think the authors of the glibc dynamic linker should rethink
>> handling
>> of map->l_versions being null fully in both do-rel.h and all the
>> functions
>> that
>> receive its potentially null version pointer - otherwise I'll have to &
>> will
>> post patches back here.
>
> It is false economy to make changes to the fast path of the dynamic loader
> in
> order to diagnose a broken binary. There are other tools that can do this
> and
> better. You must have a working non-corrupt system or the dynamic loader
> will
> fail. There is simply too much in the ELF format that has to be "correct"
> otherwise you crash.
>
> --
> You are receiving this mail because:
> You reported the bug.

-- 
You are receiving this mail because:
You are on the CC list for the bug.