Skip to content

Commit

Permalink
KVM: MMU: Store nx bit for large page shadows
Browse files Browse the repository at this point in the history
We need to distinguish between large page shadows which have the nx bit set
and those which don't.  The problem shows up when booting a newer smp Linux
kernel, where the trampoline page (which is in real mode, which uses the
same shadow pages as large pages) is using the same mapping as a kernel data
page, which is mapped using nx, causing kvm to spin on that page.

Signed-off-by: Avi Kivity <avi@qumranet.com>
  • Loading branch information
Avi Kivity committed Jul 20, 2007
1 parent 2cb7e71 commit d55e2cb
Show file tree
Hide file tree
Showing 2 changed files with 4 additions and 2 deletions.
4 changes: 2 additions & 2 deletions drivers/kvm/kvm.h
Original file line number Diff line number Diff line change
Expand Up @@ -121,7 +121,7 @@ struct kvm_pte_chain {
* bits 4:7 - page table level for this shadow (1-4)
* bits 8:9 - page table quadrant for 2-level guests
* bit 16 - "metaphysical" - gfn is not a real page (huge page/real mode)
* bits 17:18 - "access" - the user and writable bits of a huge page pde
* bits 17:19 - "access" - the user, writable, and nx bits of a huge page pde
*/
union kvm_mmu_page_role {
unsigned word;
Expand All @@ -131,7 +131,7 @@ union kvm_mmu_page_role {
unsigned quadrant : 2;
unsigned pad_for_nice_hex_output : 6;
unsigned metaphysical : 1;
unsigned hugepage_access : 2;
unsigned hugepage_access : 3;
};
};

Expand Down
2 changes: 2 additions & 0 deletions drivers/kvm/paging_tmpl.h
Original file line number Diff line number Diff line change
Expand Up @@ -366,6 +366,8 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
metaphysical = 1;
hugepage_access = *guest_ent;
hugepage_access &= PT_USER_MASK | PT_WRITABLE_MASK;
if (*guest_ent & PT64_NX_MASK)
hugepage_access |= (1 << 2);
hugepage_access >>= PT_WRITABLE_SHIFT;
table_gfn = (*guest_ent & PT_BASE_ADDR_MASK)
>> PAGE_SHIFT;
Expand Down

0 comments on commit d55e2cb

Please sign in to comment.