Re: per-bdi-throttling: synchronous writepage doesn't work correctly

[Peter Zijlstra <peterz@infradead.org>]

On Thu, 2007-11-01 at 20:19 +0100, Miklos Szeredi wrote:
> > > 
> > >       See the file "Locking" for more details.
> > > 
> > > 
> > > The "should set PG_Writeback" bit threw me off I guess.
> > 
> > Hmm, set_page_writeback() is also the one clearing the radix tree dirty
> > tag. So if that is not called, we get in a bit of a mess, no?
> > 
> > Which makes me think hostfs is buggy.
> 
> Yes, looks like that sort of usage is not valid.  But not clearing the
> dirty tag won't cause any malfunction, it'll just waste some CPU when
> looking for dirty pages to write back.  This is probably why this
> wasn't noticed earlier.

Documentation/filesystems/Locking is also quite clear on the need to
call set_page_writeback() and end_page_writeback().

minimal fix for hostfs

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---

diff --git a/fs/hostfs/hostfs_kern.c b/fs/hostfs/hostfs_kern.c
index 8966b05..b6c1e12 100644
--- a/fs/hostfs/hostfs_kern.c
+++ b/fs/hostfs/hostfs_kern.c
@@ -415,6 +415,7 @@ int hostfs_writepage(struct page *page, struct writeback_control *wbc)
 	int end_index = inode->i_size >> PAGE_CACHE_SHIFT;
 	int err;
 
+	set_page_writeback(page);
 	if (page->index >= end_index)
 		count = inode->i_size & (PAGE_CACHE_SIZE-1);
 
@@ -438,6 +439,7 @@ int hostfs_writepage(struct page *page, struct writeback_control *wbc)
 	kunmap(page);
 
 	unlock_page(page);
+	end_page_writeback(page);
 	return err;
 }