On 5/29/22 03:32, Hyeonggon Yoo wrote:
> On Mon, May 16, 2022 at 07:04:32AM -0700, Dave Hansen wrote:
> Writing code I'm a bit confused:
> commit d1440b23c922d8 ("x86/mm: Factor out pageattr
> _PAGE_GLOBAL setting") says:
>
> "This unconditional setting of _PAGE_GLOBAL is a problem when we have
> PTI and non-PTI and we want some areas to have _PAGE_GLOBAL and some
> not."
>
> Is this this sentence not valid anymore in PTI,
> and just unconditionally setting _PAGE_GLOBAL would be okay in kernel
> side regardless of PTI?
I believe it's still valid.
IIRC, there are three cases:
1. No KPTI. All kernel mappings are _PAGE_GLOBAL. Basically, for
present mappings, if _PAGE_USER is clear, _PAGE_GLOBAL is set.
2. KPTI with PCID hardware support (or in a few other cases): The kernel
image is mostly non-global. Anything mapped into userspace *is*
marked global, like entry text.
3. KPTI without PCIDs: Basically case #2, but with more of the kernel
image left global.
So, not only are there different KPTI modes, there a different pars of
the kernel that require different _PAGE_GLOBAL behavior.
pti_kernel_image_global_ok() in arch/x86/mm/pti.c explains it pretty well.
