From: Chengguang Xu <cgxu519@mykernel.net>
To: "Miklos Szeredi" <miklos@szeredi.hu>
Cc: "Miklos Szeredi" <mszeredi@redhat.com>,
"overlayfs" <linux-unionfs@vger.kernel.org>,
"Huang Jianan" <huangjianan@oppo.com>
Subject: Re: [PATCH] ovl: set overlayfs inode's a_ops->direct_IO properly
Date: Sun, 03 Oct 2021 22:41:34 +0800 [thread overview]
Message-ID: <17c469a5f3f.e5bfa83020210.6858947926351314597@mykernel.net> (raw)
In-Reply-To: <CAJfpegsHH1wpLXDJXemVM1mpcRACRwew8pc2X62KkyuwS91jKQ@mail.gmail.com>
---- 在 星期四, 2021-09-30 20:55:54 Miklos Szeredi <miklos@szeredi.hu> 撰写 ----
> On Tue, 28 Sept 2021 at 14:48, Chengguang Xu <cgxu519@mykernel.net> wrote:
> >
> > Loop device checks the ability of DIRECT-IO by checking
> > a_ops->direct_IO of inode, in order to avoid this kind of
> > false detection we set a_ops->direct_IO for overlayfs inode
> > only when underlying inode really has DIRECT-IO ability.
> >
> > Reported-by: Huang Jianan <huangjianan@oppo.com>
> > Signed-off-by: Chengguang Xu <cgxu519@mykernel.net>
>
> Can you please add Fixes: and Cc: stable@vger.kernel.org tags?
>
> > ---
> > fs/overlayfs/dir.c | 2 ++
> > fs/overlayfs/inode.c | 4 ++--
> > fs/overlayfs/overlayfs.h | 1 +
> > fs/overlayfs/util.c | 14 ++++++++++++++
> > 4 files changed, 19 insertions(+), 2 deletions(-)
> >
> > diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c
> > index 1fefb2b8960e..32a60f9e3f9e 100644
> > --- a/fs/overlayfs/dir.c
> > +++ b/fs/overlayfs/dir.c
> > @@ -648,6 +648,8 @@ static int ovl_create_object(struct dentry *dentry, int mode, dev_t rdev,
> > /* Did we end up using the preallocated inode? */
> > if (inode != d_inode(dentry))
> > iput(inode);
> > + else
> > + ovl_inode_set_aops(inode);
>
> This is too late, since the dentry was instantiated and can be found
> through a cached lookup already.
>
> Anyway, I think this can be dropped, since ovl_inode_init() should be
> called for inodes preallocated by ovl_create_object() as well:
> inode_insert5() will set I_NEW on the preallocated inode.
>
> It is interesting that ovl_fill_inode() will be called a second time
> on the preallocated inode. This is something that should probably be
> cleaned up, but that's a separate patch.
>
> >
> > out_drop_write:
> > ovl_drop_write(dentry);
> > diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c
> > index 832b17589733..a7a327e4f790 100644
> > --- a/fs/overlayfs/inode.c
> > +++ b/fs/overlayfs/inode.c
> > @@ -659,7 +659,7 @@ static const struct inode_operations ovl_special_inode_operations = {
> > .update_time = ovl_update_time,
> > };
> >
> > -static const struct address_space_operations ovl_aops = {
> > +const struct address_space_operations ovl_aops = {
> > /* For O_DIRECT dentry_open() checks f_mapping->a_ops->direct_IO */
> > .direct_IO = noop_direct_IO,
> > };
> > @@ -786,6 +786,7 @@ void ovl_inode_init(struct inode *inode, struct ovl_inode_params *oip,
> > ovl_copyattr(realinode, inode);
> > ovl_copyflags(realinode, inode);
> > ovl_map_ino(inode, ino, fsid);
> > + ovl_inode_set_aops(inode);
>
> OVL_UPPERDATA is only set after ovl_get_inode() in all callers. This
> needs to be moved into ovl_inode_init() before calling
> ovl_inode_set_aops() otherwise this won't work correctly for a copied
> up file.
>
Hi Miklos,
I found it's not convenient to move setting OVL_UPPERDATA into ovl_inode_init() because
we should detect different conditions for different callers. How about calling ovl_inode_set_aops()
after setting OVL_UPPERDATA?
Thanks,
Chengguang
next prev parent reply other threads:[~2021-10-03 14:42 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-28 12:47 [PATCH] ovl: set overlayfs inode's a_ops->direct_IO properly Chengguang Xu
2021-09-30 6:52 ` Huang Jianan
2021-09-30 6:54 ` Chengguang Xu
2021-09-30 8:02 ` Miklos Szeredi
2021-09-30 8:11 ` Huang Jianan
2021-09-30 12:55 ` Miklos Szeredi
2021-10-03 14:41 ` Chengguang Xu [this message]
2021-10-07 12:09 ` Miklos Szeredi
2021-10-12 7:41 ` Chengguang Xu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=17c469a5f3f.e5bfa83020210.6858947926351314597@mykernel.net \
--to=cgxu519@mykernel.net \
--cc=huangjianan@oppo.com \
--cc=linux-unionfs@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=mszeredi@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).