On Tue, Mar 02, 2021 at 03:46:31PM -0800, Yu, Yu-cheng wrote:
> There is a problem of doing that: pmd_write() is defined after this
> function. Maybe we can declare it first, or leave this as-is?
Or a third option: you lift those two functions in a prepatch and then
in this patch use them directl
On 3/1/2021 7:52 AM, Borislav Petkov wrote:
On Wed, Feb 17, 2021 at 02:27:12PM -0800, Yu-cheng Yu wrote:
[...]
static inline pmd_t pmd_mkdirty(pmd_t pmd)
{
- return pmd_set_flags(pmd, _PAGE_DIRTY | _PAGE_SOFT_DIRTY);
+ pmdval_t dirty = _PAGE_DIRTY;
+
+ /* Avoid creating
On 3/1/2021 7:52 AM, Borislav Petkov wrote:
On Wed, Feb 17, 2021 at 02:27:12PM -0800, Yu-cheng Yu wrote:
@@ -182,7 +206,7 @@ static inline int pud_young(pud_t pud)
static inline int pte_write(pte_t pte)
{
- return pte_flags(pte) & _PAGE_RW;
Put here a comment along the lines of:
On Wed, Feb 17, 2021 at 02:27:12PM -0800, Yu-cheng Yu wrote:
> @@ -182,7 +206,7 @@ static inline int pud_young(pud_t pud)
>
> static inline int pte_write(pte_t pte)
> {
> - return pte_flags(pte) & _PAGE_RW;
Put here a comment along the lines of:
/*
* Shadow stack pages ar
There is essentially no room left in the x86 hardware PTEs on some OSes
(not Linux). That left the hardware architects looking for a way to
represent a new memory type (shadow stack) within the existing bits.
They chose to repurpose a lightly-used state: Write=0, Dirty=1.
The reason it's lightly
5 matches
Mail list logo
