Re: [PATCH v2] x86/nmi: Fix some races in NMI uaccess

[Thomas Gleixner]

On Thu, 30 Aug 2018, Andy Lutomirski wrote:
> > On Aug 30, 2018, at 6:36 AM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
> > 
> >> On Wed, 29 Aug 2018, Nadav Amit wrote:
> >> at 8:47 AM, Andy Lutomirski <luto@xxxxxxxxxx> wrote:
> >> 
> >>> In NMI context, we might be in the middle of context switching or in
> >>> the middle of switch_mm_irqs_off().  In either case, CR3 might not
> >>> match current->mm, which could cause copy_from_user_nmi() and
> >>> friends to read the wrong memory.
> >>> 
> >>> Fix it by adding a new nmi_uaccess_okay() helper and checking it in
> >>> copy_from_user_nmi() and in __copy_from_user_nmi()'s callers.
> >>> 
> >>> Cc: stable@xxxxxxxxxxxxxxx
> >>> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> >>> Cc: Nadav Amit <nadav.amit@xxxxxxxxx>
> >>> Signed-off-by: Andy Lutomirski <luto@xxxxxxxxxx>
> >>> ---
> >>> 
> >>> Nadav, this is intended for your series.  Want to add it right
> >>> before the use_temporary_mm() stuff?
> >> 
> >> Sure. Thanks! I will apply the following small fix:
> >> 
> >>> +
> >>> +#ifdef CONFIG_DEBUG_VM
> >>> +    WARN_ON_ONCE(!loaded_mm);
> >>> +#endif
> >> 
> >> Will be changed to VM_WARN_ON_ONCE() in the two instances.
> > 
> > Unless I'm completely lost, this can just be applied to tip right
> > away. It's not depending on anything else.
> > 
> 
> Fine with me. Do you want to do the VM_WARN_ON cleanup yourself or should
> I send a v3?

I think, I'll manage