Re: mm/maccess.c:41:17: sparse: sparse: incorrect type in argument 2 (different address spaces)
From: Palmer Dabbelt
Date: Sat Aug 16 2025 - 12:54:53 EST
On Sat, 16 Aug 2025 09:19:28 PDT (-0700), viro@xxxxxxxxxxxxxxxxxx wrote:
On Sat, Aug 16, 2025 at 05:28:29PM +0800, kernel test robot wrote:
tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
head: dfd4b508c8c6106083698a0dd5e35aecc7c48725
commit: ca1a66cdd685030738cf077e3955fdedfe39fbb9 riscv: uaccess: do not do misaligned accesses in get/put_user()
date: 2 months ago
config: riscv-randconfig-r122-20250816 (https://download.01.org/0day-ci/archive/20250816/202508161713.RWu30Lv1-lkp@xxxxxxxxx/config)
compiler: clang version 22.0.0git (https://github.com/llvm/llvm-project 93d24b6b7b148c47a2fa228a4ef31524fa1d9f3f)
reproduce: (https://download.01.org/0day-ci/archive/20250816/202508161713.RWu30Lv1-lkp@xxxxxxxxx/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@xxxxxxxxx>
| Closes: https://lore.kernel.org/oe-kbuild-all/202508161713.RWu30Lv1-lkp@xxxxxxxxx/
sparse warnings: (new ones prefixed by >>)
WARNING: invalid argument to '-march': '_zacas_zabha'
>> mm/maccess.c:41:17: sparse: sparse: incorrect type in argument 2 (different address spaces) @@ expected void const [noderef] __user *from @@ got unsigned long long [usertype] * @@
mm/maccess.c:41:17: sparse: expected void const [noderef] __user *from
mm/maccess.c:41:17: sparse: got unsigned long long [usertype] *
>> mm/maccess.c:43:17: sparse: sparse: incorrect type in argument 2 (different address spaces) @@ expected void const [noderef] __user *from @@ got unsigned int [usertype] * @@
mm/maccess.c:43:17: sparse: expected void const [noderef] __user *from
mm/maccess.c:43:17: sparse: got unsigned int [usertype] *
>> mm/maccess.c:45:17: sparse: sparse: incorrect type in argument 2 (different address spaces) @@ expected void const [noderef] __user *from @@ got unsigned short [usertype] * @@
mm/maccess.c:45:17: sparse: expected void const [noderef] __user *from
mm/maccess.c:45:17: sparse: got unsigned short [usertype] *
>> mm/maccess.c:46:9: sparse: sparse: incorrect type in argument 2 (different address spaces) @@ expected void const [noderef] __user *from @@ got unsigned char [usertype] * @@
mm/maccess.c:46:9: sparse: expected void const [noderef] __user *from
mm/maccess.c:46:9: sparse: got unsigned char [usertype] *
>> mm/maccess.c:73:17: sparse: sparse: incorrect type in argument 1 (different address spaces) @@ expected void [noderef] __user *to @@ got unsigned long long [usertype] * @@
mm/maccess.c:73:17: sparse: expected void [noderef] __user *to
mm/maccess.c:73:17: sparse: got unsigned long long [usertype] *
>> mm/maccess.c:75:17: sparse: sparse: incorrect type in argument 1 (different address spaces) @@ expected void [noderef] __user *to @@ got unsigned int [usertype] * @@
mm/maccess.c:75:17: sparse: expected void [noderef] __user *to
mm/maccess.c:75:17: sparse: got unsigned int [usertype] *
>> mm/maccess.c:77:17: sparse: sparse: incorrect type in argument 1 (different address spaces) @@ expected void [noderef] __user *to @@ got unsigned short [usertype] * @@
mm/maccess.c:77:17: sparse: expected void [noderef] __user *to
mm/maccess.c:77:17: sparse: got unsigned short [usertype] *
>> mm/maccess.c:78:9: sparse: sparse: incorrect type in argument 1 (different address spaces) @@ expected void [noderef] __user *to @@ got unsigned char [usertype] * @@
mm/maccess.c:78:9: sparse: expected void [noderef] __user *to
mm/maccess.c:78:9: sparse: got unsigned char [usertype] *
mm/maccess.c:98:17: sparse: sparse: incorrect type in argument 2 (different address spaces) @@ expected void const [noderef] __user *from @@ got unsigned char [usertype] * @@
mm/maccess.c:98:17: sparse: expected void const [noderef] __user *from
mm/maccess.c:98:17: sparse: got unsigned char [usertype] *
... and that clearly has fuck-all to do with mm/maccess.c.
Ya, sorry, looks like something's just broken on the RISC-V side of
things. We lost the __user annotations when moving around the
misaligned access handling.
The problem is in
#define __get_kernel_nofault(dst, src, type, err_label) \
__get_user_nocheck(*((type *)(dst)), (type *)(src), err_label)
Make that
__get_user_nocheck(*((type *)(dst)), (__force __user type *)(src), err_label)
and similar in
#define __put_kernel_nofault(dst, src, type, err_label) \
__put_user_nocheck(*((type *)(src)), (type *)(dst), err_label)
(cast also on the second argument) and see how much noise will go away.
Thanks, I'll go clean it up.