Re: [net] a6211caa63: dmesg.UBSAN:signed-integer-overflow_in_arch/x86/include/asm/atomic.h

[Peter Zijlstra]

On Fri, Jun 05, 2020 at 06:17:51AM -0700, Eric Dumazet wrote:
> On Fri, Jun 5, 2020 at 1:10 AM kernel test robot <rong.a.chen@xxxxxxxxx> wrote:

> There you go.
> 
> We decided this was a bogus report, and that UBSAN requires sane compilers.
> 
> Please read the fine comment that was added in this commit and update
> your compiler or do not mess with compiler flags.
> 
> +       /* If UBSAN reports an error there, please make sure your compiler
> +        * supports -fno-strict-overflow before reporting it that was a bug
> +        * in UBSAN, and it has been fixed in GCC-8.
> +        */
> +       return atomic_add_return(segs + delta, p_id) - segs;

> >         make HOSTCC=gcc-4.9 CC=gcc-4.9 ARCH=i386 olddefconfig prepare modules_prepare bzImage

Rong, can you make sure to exclude gcc<8 for UBSAN reports ?