[Bug ld/31954] [ld] [lto] [clang] using ld and lto, crash while dynamic compile executable

[zhenhangwang at huawei dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=31954

Austin <zhenhangwang at huawei dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zhenhangwang at huawei dot com

--- Comment #5 from Austin <zhenhangwang at huawei dot com> ---
(In reply to Nick Clifton from comment #4)
> Hi Hanwei,
> 
>   Thanks for uploading the object files.  As I suspected however, on their
> own they are not enough to trigger the bug.  For example a I tried test,
> based upon the command line you provided, but with the system files and
> libraries removed, and it worked:
> 
>   $ ld -pie --be8 -EB -z relro --hash-style=gnu --eh-frame-hdr -m
> armelfb_linux_eabi -o g++-dg-lto-200811125.exe -plugin
> /usr/lib64/LLVMgold.so -plugin-opt=mcpu=generic cp_lto_20081125_0.o
> cp_lto_20081125_1.o -e 0
> 
>   $ echo $?
>   0
> 
> I suspect that the problem is related to the fact that you are targeting a
> big-endian ARM system.  (The little-endian version tends to get more testing
> than the big-endian version).
> 
> Given that you do have the full linker command line available, are you able
> to run the linker under GDB and find out where the seg-fault is happening ? 
> And what is causing it ? A null pointer ?  Accessing freed memory ? 
> Something else ?
> 
> Cheers
>   Nick

I also find this problem,too. When it links and processes other libraries, the
requested memory overlaps with the address allocated during plug-in
processing,As a result, the local_iplt[i]->root address of
elf32_arm_output_plt_map_1 in the elf32_arm_output_arch_local_syms function is
invalid (little-endian).

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