Re: [PATCH v5 0/2] provide check for ro_after_init memory sections

[Kees Cook]

On Fri, Apr 7, 2017 at 3:12 PM, Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> On Fri, 7 Apr 2017 14:53:23 -0700 Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
>> > Eddie Kovsky (2):
>> >   module: verify address is read-only
>> >   extable: verify address is read-only
>> >
>> >  include/linux/kernel.h |  2 ++
>> >  include/linux/module.h | 12 ++++++++++++
>> >  kernel/extable.c       | 29 +++++++++++++++++++++++++++
>> >  kernel/module.c        | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++
>> >  4 files changed, 96 insertions(+)
>>
>> Andrew, do you have these in your mailbox (it went to lkml), or should
>> I resend them directly to you? Since they depend on the
>> __start_ro_after_init naming fixes in -mm, it seemed like it'd be best
>> to carry these two patches there. If so, please consider them both:
>>
>> Acked-by: Kees Cook <keescook@xxxxxxxxxxxx>
>>
>> (And, from the thread on the module patch, Jessica has Acked that one too.)
>
> Well I grabbed them, but the patches don't actually do anything - they
> add interfaces with no users.  What's the plan here?

I'd like to have a way for interfaces (especially the various
*_register()) to be able to check that a structure is either const or
__ro_after_init. My expectation is to add those and similar
sanity-checks now that we can do so.

-Kees

-- 
Kees Cook
Pixel Security