On Fri, Oct 12, 2018 at 10:54 AM Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx> wrote: > > On 10/04/2018 07:05 AM, Sebastian Andrzej Siewior wrote: > > The PKRU value is not set for kernel threads because they do not have > > the ->initialized value set. As a result the kernel thread has a random > > PKRU value set which it inherits from the previous task. > > It has been suggested by Paolo Bonzini to set it for kernel threads, too > > because it might be a fix. > > I *think* this is not required because the kernel threads don't copy > > data to/from userland and don't have access to any userspace mm in > > general. > > However there is this use_mm(). If we gain a mm by use_mm() we don't > > have a matching PKRU value because those are per thread. It has been > > suggested to use 0 as the PKRU value but this would bypass PKRU. > > > > Set the initial (default) PKRU value for kernel threads. > > We might want to do this for cleanliness reasons... Maybe. > > But this *should* have no practical effects. Kernel threads have no > real 'mm' and no user pages. They should not have do access to user > mappings. Protection keys *only* apply to user mappings. Thus, > logically, they should never be affected by PKRU values. > > So I'm kinda missing the point of the patch. use_mm().
