On Mon, 27 Feb 2017 16:01:34 +0300
Andrey Ryabinin <aryabi...@virtuozzo.com> wrote:
> On 02/27/2017 04:03 AM, kernel test robot wrote:
> >
> > FYI, we noticed the following commit:
> >
> > commit: 243b72aae28ca1032284028323bb81c9235b15c9 ("x86/mm/ptdump: Optimize
> > check for W+X mappings for CONFIG_KASAN=y")
> > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
>
> ...
> >
> > caused below changes (please refer to attached dmesg/kmsg for entire
> > log/backtrace):
> >
> >
> > +-----------------------------------------------------+------------+------------+
> > | | 5b1ad68f9b |
> > 243b72aae2 |
> > +-----------------------------------------------------+------------+------------+
> > | boot_successes | 0 | 0
> > |
> > | boot_failures | 8 | 6
> > |
> > | BUG:KASAN:slab-out-of-bounds | 8 | 6
> > |
> > | WARNING:at_arch/x86/mm/dump_pagetables.c:#note_page | 0 | 6
> > |
> > +-----------------------------------------------------+------------+------------+
>
> Ok, I reproduced this, but it's definitely caused *not* by 243b72aae28.
> This WARN is also reproducible on the parent commit 5b1ad68f9b.
> The only difference here is that on parent one needs dozens of seconds to
> reach this WARNING.
> It seems that this time difference somehow confused the robot.
>
> As for the warning itself, it caused by kprobes. krpobe code use
> module_alloc() which
> creates these RWX mappings.
> I'm not sure how to fix this as AFAIK kprobes actually need RWX mapping.
Thanks! I also noticed this issue. I actually have a plan to make kprobes with
R+X mappings (as ftrace does).
I'll queue this on my TODO list.
--
Masami Hiramatsu <mhira...@kernel.org>
