Re: [BUG] Use of probe_kernel_address() in task_rcu_dereference() without checking return value

[Russell King - ARM Linux admin]

On Fri, Aug 30, 2019 at 08:30:00AM -0700, Linus Torvalds wrote:
> On Fri, Aug 30, 2019 at 7:08 AM Russell King - ARM Linux admin
> <linux@xxxxxxxxxxxxxxx> wrote:
> >
> > which means that when probe_kernel_address() returns -EFAULT, the
> > destination is left uninitialised.  In the case of
> > task_rcu_dereference(), this means that "siginfo" can be used without
> > having been initialised, resulting in this function returning an
> > indeterminant result (based on the value of an uninitialised variable
> > on the stack.)
> 
> Do you actually see that behavior?

No, it was an observation of the code.

> Because the foillowing lines:
> 
>         smp_rmb();
>         if (unlikely(task != READ_ONCE(*ptask)))
>                 goto retry;
> 
> are what is supposed to protect it - yes, it could have faulted, but
> only if 'task' isn't valid any more, and we just re-checked it.

Ah, ok.  Might be worth some comments - I find the comments in that
function particularly unhelpful (even after Oleg mentions this is
case 2.)

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up