On 07/22/2015 05:17 PM, Catalin Marinas wrote:

On Wed, Jul 22, 2015 at 01:30:33PM +0300, Andrey Ryabinin wrote:

quoted

diff --git a/arch/x86/mm/kasan_init_64.c b/arch/x86/mm/kasan_init_64.c
index e1840f3..2390dba 100644
--- a/arch/x86/mm/kasan_init_64.c
+++ b/arch/x86/mm/kasan_init_64.c

@@ -12,9 +12,9 @@
 extern pgd_t early_level4_pgt[PTRS_PER_PGD];
 extern struct range pfn_mapped[E820_X_MAX];
 
-static pud_t kasan_zero_pud[PTRS_PER_PUD] __page_aligned_bss;
-static pmd_t kasan_zero_pmd[PTRS_PER_PMD] __page_aligned_bss;
-static pte_t kasan_zero_pte[PTRS_PER_PTE] __page_aligned_bss;
+pud_t kasan_zero_pud[PTRS_PER_PUD] __page_aligned_bss;
+pmd_t kasan_zero_pmd[PTRS_PER_PMD] __page_aligned_bss;
+pte_t kasan_zero_pte[PTRS_PER_PTE] __page_aligned_bss;
 
 /*
  * This page used as early shadow. We don't use empty_zero_page

@@ -24,7 +24,7 @@ static pte_t kasan_zero_pte[PTRS_PER_PTE] __page_aligned_bss;
  * that allowed to access, but not instrumented by kasan
  * (vmalloc/vmemmap ...).
  */
-static unsigned char kasan_zero_page[PAGE_SIZE] __page_aligned_bss;
+unsigned char kasan_zero_page[PAGE_SIZE] __page_aligned_bss;

Did you lose part of the patch when rebasing? I can see you copied
kasan_populate_zero_shadow() to the mm code but it's still present in
the x86 one and the above changes to remove static seem meaningless.

Or you plan to submit the rest of the x86 code separately?

Yes, I was going to send x86 patch later.
Static has to be removed because this conflicts with kasan_zero_p* declarations in include/linux/kasan.h.

BTW, you could even move kasan_zero_p[tme]d arrays to mm/.

Makes sense.