Re: [kernel-hardening] [RFC V2 2/6] lib: vsprintf: whitelist stack traces

[Will Deacon]

On Wed, Oct 04, 2017 at 10:56:57AM +0200, Greg KH wrote:
> On Mon, Oct 02, 2017 at 11:42:05AM +0100, Will Deacon wrote:
> > On Sun, Oct 01, 2017 at 11:06:46AM +1100, Tobin C. Harding wrote:
> > > Use the %pP functionality to explicitly allow kernel
> > > pointers to be logged for stack traces.
> > > 
> > > Signed-off-by: Tobin C. Harding <me@xxxxxxxx>
> > > ---
> > >  arch/arm64/kernel/traps.c | 4 ++--
> > >  kernel/printk/printk.c    | 2 +-
> > >  2 files changed, 3 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c
> > > index 5ea4b85..fe09660 100644
> > > --- a/arch/arm64/kernel/traps.c
> > > +++ b/arch/arm64/kernel/traps.c
> > > @@ -147,7 +147,7 @@ void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
> > >  	struct stackframe frame;
> > >  	int skip;
> > >  
> > > -	pr_debug("%s(regs = %p tsk = %p)\n", __func__, regs, tsk);
> > > +	pr_debug("%s(regs = %pP tsk = %pP)\n", __func__, regs, tsk);
> > 
> > Why do we care for pr_debug?
> 
> Because you really want the real value?  Seems to make sense to me...

Just seems like anybody debugging the kernel using pr_debug can probably
change /proc/sys/kernel/kptr_restrict...

Will