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

On Wed, 29 Aug 2018, Nadav Amit wrote:

at 8:47 AM, Andy Lutomirski [off-list ref] wrote:

quoted

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@vger.kernel.org
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Nadav Amit <redacted>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---

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:

quoted

+
+#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.

Thanks,

	tglx