mm: fix huge zero page accounting in smaps report
authorKirill A. Shutemov <kirill@shutemov.name>
Wed, 10 Dec 2014 23:44:36 +0000 (15:44 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 11 Dec 2014 01:41:08 +0000 (17:41 -0800)
commitc164e038eee805147e95789dddb88ae3b3aca11c
tree6e3c5c99920142d46a9acb3d334e58c78d005467
parent2314b42db67be30b747122d65c6cd2c85da34538
mm: fix huge zero page accounting in smaps report

As a small zero page, huge zero page should not be accounted in smaps
report as normal page.

For small pages we rely on vm_normal_page() to filter out zero page, but
vm_normal_page() is not designed to handle pmds.  We only get here due
hackish cast pmd to pte in smaps_pte_range() -- pte and pmd format is not
necessary compatible on each and every architecture.

Let's add separate codepath to handle pmds.  follow_trans_huge_pmd() will
detect huge zero page for us.

We would need pmd_dirty() helper to do this properly.  The patch adds it
to THP-enabled architectures which don't yet have one.

[akpm@linux-foundation.org: use do_div to fix 32-bit build]
Signed-off-by: "Kirill A. Shutemov" <kirill@shutemov.name>
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Tested-by: Fengwei Yin <yfw.kernel@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/arm64/include/asm/pgtable.h
arch/powerpc/include/asm/pgtable-ppc64.h
arch/sparc/include/asm/pgtable_64.h
arch/x86/include/asm/pgtable.h
fs/proc/task_mmu.c