On Fri, 2 Aug 2013, Sha Zhengju wrote:
> On Fri, Aug 2, 2013 at 2:27 AM, Sage Weil <sage@xxxxxxxxxxx> wrote:
> > On Thu, 1 Aug 2013, Yan, Zheng wrote:
> >> On Thu, Aug 1, 2013 at 7:51 PM, Sha Zhengju <handai.szj@xxxxxxxxx> wrote:
> >> > From: Sha Zhengju <handai.szj@xxxxxxxxxx>
> >> >
> >> > Following we will begin to add memcg dirty page accounting around
> >> __set_page_dirty_
> >> > {buffers,nobuffers} in vfs layer, so we'd better use vfs interface to
> >> avoid exporting
> >> > those details to filesystems.
> >> >
> >> > Signed-off-by: Sha Zhengju <handai.szj@xxxxxxxxxx>
> >> > ---
> >> > fs/ceph/addr.c | 13 +------------
> >> > 1 file changed, 1 insertion(+), 12 deletions(-)
> >> >
> >> > diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
> >> > index 3e68ac1..1445bf1 100644
> >> > --- a/fs/ceph/addr.c
> >> > +++ b/fs/ceph/addr.c
> >> > @@ -76,7 +76,7 @@ static int ceph_set_page_dirty(struct page *page)
> >> > if (unlikely(!mapping))
> >> > return !TestSetPageDirty(page);
> >> >
> >> > - if (TestSetPageDirty(page)) {
> >> > + if (!__set_page_dirty_nobuffers(page)) {
> >> it's too early to set the radix tree tag here. We should set page's snapshot
> >> context and increase the i_wrbuffer_ref first. This is because once the tag
> >> is set, writeback thread can find and start flushing the page.
> >
> > Unfortunately I only remember being frustrated by this code. :) Looking
> > at it now, though, it seems like the minimum fix is to set the
> > page->private before marking the page dirty. I don't know the locking
> > rules around that, though. If that is potentially racy, maybe the safest
> > thing would be if __set_page_dirty_nobuffers() took a void* to set
> > page->private to atomically while holding the tree_lock.
> >
>
> Sorry, I don't catch the point of your last sentence... Could you
> please explain it again?
It didn't make much sense. :) I was worried about multiple callers to
set_page_dirty, but as understand it, this all happens under page->lock,
right? (There is a mention of other special cases in mm/page-writeback.c,
but I'm hoping we don't need to worry about that.)
In any case, I suspect what we actually want is something like the below
(untested) patch. The snapc accounting can be ignored here because
invalidatepage will clean it up...
sage
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index afb2fc2..7602e46 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -76,9 +76,10 @@ static int ceph_set_page_dirty(struct page *page)
if (unlikely(!mapping))
return !TestSetPageDirty(page);
- if (TestSetPageDirty(page)) {
+ if (PageDirty(page)) {
dout("%p set_page_dirty %p idx %lu -- already dirty\n",
mapping->host, page, page->index);
+ BUG_ON(!PagePrivate(page));
return 0;
}
@@ -107,35 +108,16 @@ static int ceph_set_page_dirty(struct page *page)
snapc, snapc->seq, snapc->num_snaps);
spin_unlock(&ci->i_ceph_lock);
- /* now adjust page */
- spin_lock_irq(&mapping->tree_lock);
- if (page->mapping) { /* Race with truncate? */
- WARN_ON_ONCE(!PageUptodate(page));
- account_page_dirtied(page, page->mapping);
- radix_tree_tag_set(&mapping->page_tree,
- page_index(page), PAGECACHE_TAG_DIRTY);
-
- /*
- * Reference snap context in page->private. Also set
- * PagePrivate so that we get invalidatepage callback.
- */
- page->private = (unsigned long)snapc;
- SetPagePrivate(page);
- } else {
- dout("ANON set_page_dirty %p (raced truncate?)\n", page);
- undo = 1;
- }
-
- spin_unlock_irq(&mapping->tree_lock);
-
- if (undo)
- /* whoops, we failed to dirty the page */
- ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
-
- __mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
+ /*
+ * Reference snap context in page->private. Also set
+ * PagePrivate so that we get invalidatepage callback.
+ */
+ BUG_ON(PagePrivate(page));
+ page->private = (unsigned long)snapc;
+ SetPagePrivate(page);
- BUG_ON(!PageDirty(page));
- return 1;
+ return __set_page_dirty_nobuffers(page);
}
/*
--
To unsubscribe from this list: send the line "unsubscribe cgroups" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
