Re: [RFC][PATCH 03/10] x86/mm: introduce "default" kernel PTE mask

[Tom Lendacky]

On 2/22/2018 4:26 PM, Dave Hansen wrote:
> On 02/22/2018 02:21 PM, Nadav Amit wrote:
>> Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx> wrote:
>>> From: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
>>> The __PAGE_KERNEL_* page permissions are "raw".  They contain bits
>>> that may or may not be supported on the current processor.  They
>>> need to be filtered by a mask (currently __supported_pte_mask) to
>>> turn them into a value that we can actually set in a PTE.
>>>
>>> These __PAGE_KERNEL_* values all contain _PAGE_GLOBAL.  But, with
>>> PTI, we want to be able to support _PAGE_GLOBAL (have the bit set
>>> in __supported_pte_mask) but not have it appear in any of these
>>> masks by default.
>>
>> There might be a potential issue with this approach. __supported_pte_mask is
>> exported, so out-of-tree modules might use it. They therefore can
>> unknowingly use this value to set PTEs with _PAGE_GLOBAL set.
> 
> I don't think we can help out-of-tree modules getting this wrong.
> They're OK if they use PAGE_KERNEL*, btw.

You will probably need to change __default_kernel_pte_mask to be an
EXPORT_SYMBOL instead of an EXPORT_SYMBOL_GPL for (some) out-of-tree
modules.

Thanks,
Tom

>