Re: [RFC PATCH V11 12/21] Btrfs: subpagesize-blocksize: Search for all ordered extents that could span across a page.

From: Liu Bo <bo.li.liu@oracle.com>
To: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Cc: clm@fb.com, jbacik@fb.com, dsterba@suse.cz,
	linux-btrfs@vger.kernel.org, chandan@mykolab.com
Subject: Re: [RFC PATCH V11 12/21] Btrfs: subpagesize-blocksize: Search for all ordered extents that could span across a page.
Date: Mon, 6 Jul 2015 11:17:38 +0800	[thread overview]
Message-ID: <20150706025900.GA6105@localhost.localdomain> (raw)
In-Reply-To: <1662899.ajn0RKgkkt@localhost.localdomain>

On Fri, Jul 03, 2015 at 03:38:00PM +0530, Chandan Rajendra wrote:
> On Wednesday 01 Jul 2015 22:47:10 Liu Bo wrote:
> > On Mon, Jun 01, 2015 at 08:52:47PM +0530, Chandan Rajendra wrote:
> > > In subpagesize-blocksize scenario it is not sufficient to search using the
> > > first byte of the page to make sure that there are no ordered extents
> > > present across the page. Fix this.
> > > 
> > > Signed-off-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
> > > ---
> > > 
> > >  fs/btrfs/extent_io.c | 3 ++-
> > >  fs/btrfs/inode.c     | 4 ++--
> > >  2 files changed, 4 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
> > > index 14b4e05..0b017e1 100644
> > > --- a/fs/btrfs/extent_io.c
> > > +++ b/fs/btrfs/extent_io.c
> > > @@ -3244,7 +3244,8 @@ static int __extent_read_full_page(struct
> > > extent_io_tree *tree,> 
> > >  	while (1) {
> > >  	
> > >  		lock_extent(tree, start, end);
> > > 
> > > -		ordered = btrfs_lookup_ordered_extent(inode, start);
> > > +		ordered = btrfs_lookup_ordered_range(inode, start,
> > > +						PAGE_CACHE_SIZE);
> > 
> > A minor suggestion, it'd be better to include the new prototype in the
> > same patch, which will be benefit to later cherry-picking or reverting.
> > 
> 
> Liu, The definition of btrfs_lookup_ordered_range() is already part of
> the mainline kernel.

Ah, I didn't recognize the difference of btrfs_lookup_ordered_extent and
btrfs_lookup_ordered_range, sorry.

> 
> > 
> > >  		if (!ordered)
> > >  		
> > >  			break;
> > >  		
> > >  		unlock_extent(tree, start, end);
> > > 
> > > diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
> > > index e9bab73..8b4aaed 100644
> > > --- a/fs/btrfs/inode.c
> > > +++ b/fs/btrfs/inode.c
> > > 
> > > @@ -1976,7 +1976,7 @@ again:
> > >  	if (PagePrivate2(page))
> > >  	
> > >  		goto out;
> > > 
> > > -	ordered = btrfs_lookup_ordered_extent(inode, page_start);
> > > +	ordered = btrfs_lookup_ordered_range(inode, page_start,
> > > PAGE_CACHE_SIZE);
> > > 
> > >  	if (ordered) {
> > >  	
> > >  		unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start,
> > >  		
> > >  				     page_end, &cached_state, GFP_NOFS);
> > > 
> > > @@ -8513,7 +8513,7 @@ static void btrfs_invalidatepage(struct page *page,
> > > unsigned int offset,> 
> > >  	if (!inode_evicting)
> > >  	
> > >  		lock_extent_bits(tree, page_start, page_end, 0, 
> &cached_state);
> > > 
> > > -	ordered = btrfs_lookup_ordered_extent(inode, page_start);
> > > +	ordered = btrfs_lookup_ordered_range(inode, page_start,
> > > PAGE_CACHE_SIZE);

It's possible for a page to hold two (or more) ordered extents here, a
while loop is necessary to ensure that every ordered extent is processed
properly.

Thanks,

-liubo

> > > 
> > >  	if (ordered) {
> > >  	
> > >  		/*
> > >  		
> > >  		 * IO on this page will never be started, so we need
> 
> -- 
> chandan
> 