On Thu, Sep 16, 2021 at 8:22 AM Ondrej Mosnacek <omosnace(a)redhat.com> wrote:
On Thu, Sep 16, 2021 at 2:08 PM kernel test robot
<lkp(a)intel.com> wrote:
> tree:
https://github.com/SELinuxProject/selinux-kernel stable-5.15
> head: c491f0a471580712a4254adece400c3ebb3d8e44
> commit: c491f0a471580712a4254adece400c3ebb3d8e44 [1/1] lockdown,selinux: fix wrong
subject in some SELinux lockdown checks
> config: um-x86_64_defconfig (attached as .config)
> compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
> reproduce (this is a W=1 build):
> #
https://github.com/SELinuxProject/selinux-kernel/commit/c491f0a471580712a...
> git remote add selinuxproject-selinux
https://github.com/SELinuxProject/selinux-kernel
> git fetch --no-tags selinuxproject-selinux stable-5.15
> git checkout c491f0a471580712a4254adece400c3ebb3d8e44
> # save the attached .config to linux build tree
> make W=1 ARCH=um SUBARCH=x86_64
>
> If you fix the issue, kindly add following tag as appropriate
> Reported-by: kernel test robot <lkp(a)intel.com>
>
> All warnings (new ones prefixed by >>):
>
> In file included from include/linux/rbtree.h:24,
> from include/linux/mm_types.h:10,
> from include/linux/mmzone.h:21,
> from include/linux/gfp.h:6,
> from include/linux/mm.h:10,
> from drivers/char/mem.c:12:
> drivers/char/mem.c: In function 'open_port':
> >> include/linux/rcupdate.h:395:2: warning: passing argument 1 of
'security_locked_down' discards 'const' qualifier from pointer target type
[-Wdiscarded-qualifiers]
> 395 | ({ \
> | ~^~~
> 396 | RCU_LOCKDEP_WARN(!(c), "suspicious rcu_dereference_protected()
usage"); \
> |
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 397 | rcu_check_sparse(p, space); \
> | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 398 | ((typeof(*p) __force __kernel *)(p)); \
> | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 399 | })
> | ~~
> include/linux/rcupdate.h:587:2: note: in expansion of macro
'__rcu_dereference_protected'
> 587 | __rcu_dereference_protected((p), (c), __rcu)
> | ^~~~~~~~~~~~~~~~~~~~~~~~~~~
> include/linux/cred.h:299:2: note: in expansion of macro
'rcu_dereference_protected'
> 299 | rcu_dereference_protected(current->cred, 1)
> | ^~~~~~~~~~~~~~~~~~~~~~~~~
> drivers/char/mem.c:620:28: note: in expansion of macro 'current_cred'
> 620 | rc = security_locked_down(current_cred(), LOCKDOWN_DEV_MEM);
> | ^~~~~~~~~~~~
> In file included from include/linux/fs_context.h:14,
> from include/linux/fs_parser.h:11,
> from include/linux/shmem_fs.h:11,
> from drivers/char/mem.c:25:
> include/linux/security.h:1347:53: note: expected 'struct cred *' but
argument is of type 'const struct cred *'
> 1347 | static inline int security_locked_down(struct cred *cred, enum
lockdown_reason what)
> | ~~~~~~~~~~~~~^~~~
Ah, I forgot to add the const qualifier to the function definition in
the CONFIG_SECURITY=n branch... Paul, will you amend the commit or
should I send an updated patch?
This patch is cursed. I had to hack up a fixup patch to test this in
my Rawhide test automation last night too; evidently Rawhide carries
an out-of-tree lockdown patch which results in some merge rejects.
Sigh.
Yes, the quickest path is for me to just fix up the dummy function and
do a force-push (grrrrrr) back on top of selinux/stable-5.15.
--
paul moore