Re: pipe/page fault oddness.

["Kirill A. Shutemov" <kirill@shutemov.name>]

On Thu, Oct 02, 2014 at 09:01:38AM -0700, Linus Torvalds wrote:
> On Thu, Oct 2, 2014 at 7:25 AM, Kirill A. Shutemov <kirill@shutemov.name> wrote:
> >
> > I don't see what prevents the code to make zero page writable here.
> > We need at least pmd = pmd_wrprotect(pmd) before set_pmd_at();
> 
> Do we? If it's the zero page, it had better be an anonymous mapping,
> and vm_page_prot had better not be writable.
> 
> Anonymous pages don't _start_ out writable, we explicitly make them so
> with code like
> 
>         if (vma->vm_flags & VM_WRITE)
>                 entry = pte_mkwrite(pte_mkdirty(entry));
> 
> so it should be fine to just use "pmd_modify(pmd, vma->vm_page_prot);" directly.
> 
> But hey, this is the kind of thing that maybe I'm missing something on..

You're right.
It means we have redundant pmd_wrprotect() in set_huge_zero_page().

==========================================================================

Subject: [PATCH] thp: do not mark zero-page pmd write-protected explicitly

Zero pages can be used only in anonymous mappings, which never have
writable vma->vm_page_prot: see protection_map in mm/mmap.c and __PX1X
definitions.

Let's drop redundant pmd_wrprotect() in set_huge_zero_page().

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
---
 mm/huge_memory.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index d9a21d06b862..2c17d184b56d 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -784,7 +784,6 @@ static bool set_huge_zero_page(pgtable_t pgtable, struct mm_struct *mm,
 	if (!pmd_none(*pmd))
 		return false;
 	entry = mk_pmd(zero_page, vma->vm_page_prot);
-	entry = pmd_wrprotect(entry);
 	entry = pmd_mkhuge(entry);
 	pgtable_trans_huge_deposit(mm, pmd, pgtable);
 	set_pmd_at(mm, haddr, pmd, entry);
-- 
 Kirill A. Shutemov