lkml.org 
[lkml]   [2015]   [Jul]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 1/2] f2fs: reorganize __f2fs_add_link
Hi Chao,

BTW, isn't there any problem on update_inode/mark_inode_dirty stuffs?
And, is there a hole to write uncompleted node pages unnecessarily?

On Fri, Jul 24, 2015 at 08:52:00AM -0700, Jaegeuk Kim wrote:
> Hi Chao,
>
> On Fri, Jul 24, 2015 at 06:31:16PM +0800, Chao Yu wrote:
> > Hi Jaegeuk,
> >
> > > -----Original Message-----
> > > From: Jaegeuk Kim [mailto:jaegeuk@kernel.org]
> > > Sent: Friday, July 24, 2015 2:15 AM
> > > To: Chao Yu
> > > Cc: linux-f2fs-devel@lists.sourceforge.net; linux-kernel@vger.kernel.org
> > > Subject: Re: [PATCH 1/2] f2fs: reorganize __f2fs_add_link
> > >
> > > Hi Chao,
> > >
> > > On Wed, Jul 22, 2015 at 06:21:52PM +0800, Chao Yu wrote:
> > > > Firstly, spliting init_inode_metadata into two parts as below since they
> > > > are relatively independent.
> > > > 1) init_inode_metadata: init datas belong to newly created inode;
> > > > 2) update_inode_metadata: update inode info for linking inode to new
> > > > dentry.
> > > >
> > > > Secondly, move init_inode_metadata to the front of __f2fs_add_link,
> > > > So the flow of __f2fs_add_link will be reorganized as:
> > > > 1) init inode meta data for new creatation;
> > > > 2) lookup dentry page and insert new directory entry in the page;
> > > > 3) update meta data of inode and parent.
> > > >
> > > > Signed-off-by: Chao Yu <chao2.yu@samsung.com>
> > > > ---
> > > > fs/f2fs/dir.c | 129 ++++++++++++++++++++++++++++++++-----------------------
> > > > fs/f2fs/f2fs.h | 4 +-
> > > > fs/f2fs/inline.c | 2 +-
> > > > 3 files changed, 78 insertions(+), 57 deletions(-)
> > > >
> > > > diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
> > > > index a34ebd8..0ad9a1c 100644
> > > > --- a/fs/f2fs/dir.c
> > > > +++ b/fs/f2fs/dir.c
> > > > @@ -381,43 +381,68 @@ static int make_empty_dir(struct inode *inode,
> > > > return 0;
> > > > }
> > > >
> > > > -struct page *init_inode_metadata(struct inode *inode, struct inode *dir,
> > > > +int init_inode_metadata(struct inode *inode, struct inode *dir,
> > > > const struct qstr *name, struct page *dpage)
> > > > {
> > > > struct page *page;
> > > > int err;
> > > >
> > > > - if (is_inode_flag_set(F2FS_I(inode), FI_NEW_INODE)) {
> > > > - page = new_inode_page(inode);
> > > > - if (IS_ERR(page))
> > > > - return page;
> > > > + if (!is_inode_flag_set(F2FS_I(inode), FI_NEW_INODE))
> > > > + return 0;
> > > >
> > > > - if (S_ISDIR(inode->i_mode)) {
> > > > - err = make_empty_dir(inode, dir, page);
> > > > - if (err)
> > > > - goto error;
> > > > - }
> > > > + page = new_inode_page(inode);
> > > > + if (IS_ERR(page))
> > > > + return PTR_ERR(page);
> > > >
> > > > - err = f2fs_init_acl(inode, dir, page, dpage);
> > > > + if (S_ISDIR(inode->i_mode)) {
> > > > + err = make_empty_dir(inode, dir, page);
> > > > if (err)
> > > > - goto put_error;
> > > > + goto error;
> > > > + }
> > > >
> > > > - err = f2fs_init_security(inode, dir, name, page);
> > > > + err = f2fs_init_acl(inode, dir, page, dpage);
> > > > + if (err)
> > > > + goto put_error;
> > > > +
> > > > + err = f2fs_init_security(inode, dir, name, page);
> > > > + if (err)
> > > > + goto put_error;
> > > > +
> > > > + if (f2fs_encrypted_inode(dir) && f2fs_may_encrypt(inode)) {
> > > > + err = f2fs_inherit_context(dir, inode, page);
> > > > if (err)
> > > > goto put_error;
> > > > + }
> > > >
> > > > - if (f2fs_encrypted_inode(dir) && f2fs_may_encrypt(inode)) {
> > > > - err = f2fs_inherit_context(dir, inode, page);
> > > > - if (err)
> > > > - goto put_error;
> > > > - }
> > > > - } else {
> > > > - page = get_node_page(F2FS_I_SB(dir), inode->i_ino);
> > > > - if (IS_ERR(page))
> > > > - return page;
> > > > + if (f2fs_encrypted_inode(dir))
> > > > + file_set_enc_name(inode);
> > > > +
> > > > + update_inode(inode, page);
> > > > + f2fs_put_page(page, 1);
> > >
> > > We should not put the inode page here.
> > > Since checkpoint can be called and flush all the node pages with valid
> > > NAT entres. Once power-off happens after then, we can see unreachable NAT
> > > entries.
> >
> > IMO, all callers of init_inode_metadata such as __f2fs_add_link/f2fs_do_tmpfile,
> > they are all protected well by cp_rwsem, so checkpoint & SPO will not easily
> > bother us.
> >
> > 1) Lately in __f2fs_add_link, ino of ipage will be inserted into directory entry
> > of parent's inode, this makes our nat entry of ipage reachable.
> > 2) And also in __f2fs_tmpfile, ino will be added in orphan list, so all data
> > resource of the inode will be evicted after CP->SPO->RFR procedure.
> >
> > Is there anything I missed?
>
> Let me think about this for a while.
> I think two patches are clean-ups with a little bit big changes.
> Currently, we've touched many parts including extent_cache, so I need to focus
> on stabilizing them first.
> After then, I'd like to dig two clean-up patches. Ok?
>
> Thanks,
>
> >
> > Thanks,
> >
> > >
> > > Thanks,
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/


\
 
 \ /
  Last update: 2015-07-24 18:41    [W:0.067 / U:0.908 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site