llvm dev - Jun 2016 - lld/x86_64 linux elf invalid link_map

I'm having a curious issue with LLD/x86_64 linux/elf (Ubuntu 14.04); 
Where the l_addr of the link_map is invalid when linked with lld, but is 
fine with gnu ld.

I'm using the libgc (boehm) code which when initializing reads the 
DYNAMIC/DEBUG link_map data, and crashes because the l_addr field has 
value out of readable memory. The strange this is that it happens only 
on some linux systems even though they're the same ubuntu version.

My question is, what could possibly cause this, where can I start looking?

-- 
Carlo Kok

On 7 June 2016 at 03:18, Carlo Kok via llvm-dev <llvm-dev at
lists.llvm.org> wrote:
> I'm having a curious issue with LLD/x86_64 linux/elf (Ubuntu 14.04);
Where
> the l_addr of the link_map is invalid when linked with lld, but is fine
with
> gnu ld.
>
> I'm using the libgc (boehm) code which when initializing reads the
> DYNAMIC/DEBUG link_map data, and crashes because the l_addr field has value
> out of readable memory. The strange this is that it happens only on some
> linux systems even though they're the same ubuntu version.
>
> My question is, what could possibly cause this, where can I start looking?
Hard to tell just from the description. If you can shore the testcase,
just pass --reproduce foo to ld.lld and it will create a cpio file
that you can put somewhere and refer from a bugzilla

Cheers,
Rafael

On 2016-06-07 20:28, Rafael Espíndola wrote:
> On 7 June 2016 at 03:18, Carlo Kok via llvm-dev <llvm-dev at
lists.llvm.org> wrote:
>> I'm having a curious issue with LLD/x86_64 linux/elf (Ubuntu
14.04); Where
>> the l_addr of the link_map is invalid when linked with lld, but is fine
with
>> gnu ld.
>>
>> I'm using the libgc (boehm) code which when initializing reads the
>> DYNAMIC/DEBUG link_map data, and crashes because the l_addr field has
value
>> out of readable memory. The strange this is that it happens only on
some
>> linux systems even though they're the same ubuntu version.
>>
>> My question is, what could possibly cause this, where can I start
looking?
>
> Hard to tell just from the description. If you can shore the testcase,
> just pass --reproduce foo to ld.lld and it will create a cpio file
> that you can put somewhere and refer from a bugzilla
>
logged as https://llvm.org/bugs/show_bug.cgi?id=28040 (couldn't attach 
the cpio file, it's 5mb and it accepts up to 1mb). is there anything 
else i can provide?


-- 
Carlo Kok
RemObjects Software