Re: [PATCH] [for mmotm-1113] mm: Simplify try_to_unmap_one()

[KOSAKI Motohiro]

> On Tue, 17 Nov 2009 17:39:27 +0900 (JST)
> KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx> wrote:
> 
> > +out_mlock:
> > +	pte_unmap_unlock(pte, ptl);
> > +
> > +	if (down_read_trylock(&vma->vm_mm->mmap_sem)) {
> > +		if (vma->vm_flags & VM_LOCKED) {
> > +			mlock_vma_page(page);
> > +			ret = SWAP_MLOCK;
> >  		}
> > +		up_read(&vma->vm_mm->mmap_sem);
> 
> It's somewhat unobvious why we're using a trylock here.  Ranking versus
> lock_page(), perhaps?
> 
> In general I think a trylock should have an associated comment which explains
> 
> a) why it is being used at this site and
> 
> b) what happens when the trylock fails - why this isn't a
>    bug, how the kernel recovers from the inconsistency, what its
>    overall effect is, etc.
> 
> <wonders why we need to take mmap_sem here at all>

This mmap_sem is needed certainenaly. Following comment is sufficient?


---
 mm/rmap.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/mm/rmap.c b/mm/rmap.c
index 70dec01..b1c9342 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -860,6 +860,14 @@ out:
 out_mlock:
 	pte_unmap_unlock(pte, ptl);
 
+
+	/*
+	 * We need mmap_sem locking, Otherwise VM_LOCKED check makes
+	 * unstable result and race. Plus, We can't wait here because
+	 * we now hold anon_vma->lock or mapping->i_mmap_lock.
+	 * If trylock failed, The page remain evictable lru and
+	 * retry to more unevictable lru by later vmscan.
+	 */
 	if (down_read_trylock(&vma->vm_mm->mmap_sem)) {
 		if (vma->vm_flags & VM_LOCKED) {
 			mlock_vma_page(page);
-- 
1.6.2.5



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/