If we get a vmalloc fault while current->active_mm->pgd doesn't
match CR3, we'll crash without this change.  I've seen this failure
mode on heavily instrumented kernels with virtually mapped stacks.

Signed-off-by: Andy Lutomirski <l...@kernel.org>
---
 arch/x86/mm/fault.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 7d1fa7cd2374..ca44e2e7fd00 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -439,7 +439,7 @@ static noinline int vmalloc_fault(unsigned long address)
         * happen within a race in page table update. In the later
         * case just flush:
         */
-       pgd = pgd_offset(current->active_mm, address);
+       pgd = (pgd_t *)__va(read_cr3()) + pgd_index(address);
        pgd_ref = pgd_offset_k(address);
        if (pgd_none(*pgd_ref))
                return -1;
-- 
2.7.4

Reply via email to