Re: [PATCH v2 1/2] RAS: Introduce AMD Address Translation Library

[Yazen Ghannam <yazen.ghannam@amd.com>]

On 10/16/2023 8:28 AM, kernel test robot wrote:
> Hi Yazen,
> 
> kernel test robot noticed the following build errors:
> 
> [auto build test ERROR on ras/edac-for-next]
> [also build test ERROR on linus/master v6.6-rc6 next-20231016]
> [If your patch is applied to the wrong git tree, kindly drop us a note.
> And when submitting patch, we suggest to use '--base' as documented in
> https://git-scm.com/docs/git-format-patch#_base_tree_information]
> 
> url:    https://github.com/intel-lab-lkp/linux/commits/Yazen-Ghannam/RAS-Introduce-AMD-Address-Translation-Library/20231006-014458
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/ras/ras.git edac-for-next
> patch link:    https://lore.kernel.org/r/20231005173526.42831-2-yazen.ghannam%40amd.com
> patch subject: [PATCH v2 1/2] RAS: Introduce AMD Address Translation Library
> config: i386-buildonly-randconfig-002-20231016 (https://download.01.org/0day-ci/archive/20231016/202310162015.fuitGuiJ-lkp@intel.com/config)
> compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20231016/202310162015.fuitGuiJ-lkp@intel.com/reproduce)
> 
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <lkp@intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202310162015.fuitGuiJ-lkp@intel.com/
> 
> All errors (new ones prefixed by >>):
> 
>     ld: drivers/ras/amd/atl/denormalize.o: in function `denormalize_address':
>>> denormalize.c:(.text+0x84f): undefined reference to `__umoddi3'
> 

This is the code of interest:
	addr_mod = temp_addr_b % 3;

"temp_addr_b" is a u64 value.

__umoddi3() takes "unsigned long" values as input. This works on x86_64 
where "unsigned long" is 64-bits, but will fail on i386 where "unsigned 
long" is 32-bits. I *think* __umodti3() should be used which takes 
"unsigned long long" values as input. This would work on both i386 and 
x86_64, since "unsigned long long" will be 64-bits.

I was able to reproduce the issue with GCC 9 and 10, and the issue is 
not present with GCC 11 and 12.

I'm assuming a compiler/linker issue at the moment. In particular, I 
wonder if gcc and ld need to be matched. I'm using a newer ld version 
(2.38), and this works with newer gcc versions (11 and 12) but fails 
with the older versions (9 and 10).

Any thoughts?

Thanks,
Yazen