Re: [PATCH v4 04/45] x86: asm: instrument usercopy in get_user() and __put_user_size()

[Alexander Potapenko <glider@google.com>]

On Sat, Jul 2, 2022 at 12:46 PM kernel test robot <lkp@intel.com> wrote:
>
> Hi Alexander,
>
> I love your patch! Yet something to improve:
>
> [auto build test ERROR on masahiroy-kbuild/for-next]
> [also build test ERROR on linus/master v5.19-rc4 next-20220701]
> [cannot apply to tip/x86/core tip/x86/mm]
> [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]
>
> url:    https://github.com/intel-lab-lkp/linux/commits/Alexander-Potapenko/Add-KernelMemorySanitizer-infrastructure/20220701-222712
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild.git for-next
> config: i386-randconfig-a011 (https://download.01.org/0day-ci/archive/20220702/202207021844.0J4s1Gjz-lkp@intel.com/config)
> compiler: clang version 15.0.0 (https://github.com/llvm/llvm-project a9119143a2d1f4d0d0bc1fe0d819e5351b4e0deb)
> reproduce (this is a W=1 build):
>         wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
>         chmod +x ~/bin/make.cross
>         # https://github.com/intel-lab-lkp/linux/commit/0ca0e4029535365a65588446ba55a952ca186079
>         git remote add linux-review https://github.com/intel-lab-lkp/linux
>         git fetch --no-tags linux-review Alexander-Potapenko/Add-KernelMemorySanitizer-infrastructure/20220701-222712
>         git checkout 0ca0e4029535365a65588446ba55a952ca186079
>         # save the config file
>         mkdir build_dir && cp config build_dir/.config
>         COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=i386 SHELL=/bin/bash
>
> If you fix the issue, kindly add following tag where applicable
> Reported-by: kernel test robot <lkp@intel.com>

Uh oh, this is a very good catch.

> All errors (new ones prefixed by >>):
>
> >> arch/x86/math-emu/reg_ld_str.c:1043:3: error: address of bit-field requested
>                    FPU_get_user(instruction_address.selector,
>                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>    arch/x86/math-emu/fpu_system.h:127:36: note: expanded from macro 'FPU_get_user'
>    #define FPU_get_user(x,y) do { if (get_user((x),(y))) FPU_abort; } while (0)
>                                   ~~~~^~~~~~~~~~~~~~~~~~
>    arch/x86/include/asm/uaccess.h:131:43: note: expanded from macro 'get_user'
>    #define get_user(x,ptr) ({ might_fault(); do_get_user_call(get_user,x,ptr); })
>                                              ^
>    arch/x86/include/asm/uaccess.h:103:43: note: expanded from macro 'do_get_user_call'
>            instrument_copy_from_user_before((void *)&(x), ptr, sizeof(*(ptr))); \
>                                                     ^
>    include/linux/compiler.h:56:47: note: expanded from macro 'if'
>    #define if(cond, ...) if ( __trace_if_var( !!(cond , ## __VA_ARGS__) ) )
>                               ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~
>    include/linux/compiler.h:58:52: note: expanded from macro '__trace_if_var'
>    #define __trace_if_var(cond) (__builtin_constant_p(cond) ? (cond) : __trace_if_value(cond))
>                                                       ^~~~
> >> arch/x86/math-emu/reg_ld_str.c:1043:3: error: address of bit-field requested
>                    FPU_get_user(instruction_address.selector,
>                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Actually `x` in get_user(x, ptr) and put_user(x, ptr) is not
guaranteed to be addressable (e.g. in this case it is a bitfield).
We can copy the values to a local var, like it's done in __put_user_size:

254         __typeof__(*(ptr)) __pus_val = x;                               \
255         __chk_user_ptr(ptr);                                            \
256         instrument_copy_to_user(ptr, &(__pus_val), size);               \

, but this won't help KASAN or KCSAN detect the bugs accessing x.
Perhaps a better solution would be to declare instrument_get_user()
and instrument_put_user(), which will unpoison/check the values for
KMSAN and do nothing for other tools.


-- 
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Liana Sebastian
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg