Received-SPF: pass (google.com: domain of linux-kernel+bounces-51922-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240202170537.409508-1-meetakshisetiyaoss@gmail.com>
In-Reply-To: <20240202170537.409508-1-meetakshisetiyaoss@gmail.com>
From: Steve French <smfrench@gmail.com>
Date: Sun, 4 Feb 2024 15:57:43 -0600
Message-ID: <CAH2r5msEtpJGTvK2F2TY0jR9hU0e7LYJekiyV1zD7MM+i9Oxgg@mail.gmail.com>
Subject: Re: [PATCH] smb: client: do not defer close open handles to deleted files
To: meetakshisetiyaoss@gmail.com
Cc: pc@manguebit.com, ronniesahlberg@gmail.com, sprasad@microsoft.com, 
	nspmangalore@gmail.com, tom@talpey.com, linux-cifs@vger.kernel.org, 
	linux-kernel@vger.kernel.org, samba-technical@lists.samba.org, 
	bharathsm.hsk@gmail.com, msetiya@microsoft.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

This was puzzling because I tried some experiments with deleting files
that were deferred close and noticed that we already making sure that
they weren't deferred close in cifs_unlink by calling
cifs_close_deferred_file_under_dentry(tcon, full_path) even in
Rohith's original code from a few years ago - so I didn't see cases
where we had a file marked for delete that was deferred close.  How
did you reproduce that problem?

On Fri, Feb 2, 2024 at 11:39=E2=80=AFAM <meetakshisetiyaoss@gmail.com> wrot=
e:
>
> From: Meetakshi Setiya <msetiya@microsoft.com>
>
> When a file/dentry has been deleted before closing all its open
> handles, currently, closing them can add them to the deferred close
> list. This can lead to problems in creating file with the same name
> when the file is re-created before the deferred close completes. This
> issue was seen while reusing a client's already existing lease on a
> file for compound operations and xfstest 591 failed because of the
> deferred close handle that remained valid even after the file was
> deleted and was being reused to create a file with the same name. The
> server in this case returns an error on open with
> STATUS_DELETE_PENDING. Recreating the file would fail till the
> deferred handles are closed (duration specified in closetimeo).
>
> This patch fixes the issue by flagging all open handles for the
> deleted file (file path to be precise) with FILE_DELETED at the end
> of the unlink operation. When doing close in cifs_close for each of
> these handles, check the flag and do not defer close these handles if
> the corresponding filepath has been deleted.
>
> Signed-off-by: Meetakshi Setiya <msetiya@microsoft.com>
> ---
>  fs/smb/client/cifsglob.h  |  1 +
>  fs/smb/client/cifsproto.h |  4 ++++
>  fs/smb/client/file.c      |  2 +-
>  fs/smb/client/inode.c     |  2 ++
>  fs/smb/client/misc.c      | 22 ++++++++++++++++++++++
>  5 files changed, 30 insertions(+), 1 deletion(-)
>
> diff --git a/fs/smb/client/cifsglob.h b/fs/smb/client/cifsglob.h
> index decf80131bbe..826da216a9e1 100644
> --- a/fs/smb/client/cifsglob.h
> +++ b/fs/smb/client/cifsglob.h
> @@ -1399,6 +1399,7 @@ struct cifs_fid_locks {
>         struct list_head locks;         /* locks held by fid above */
>  };
>
> +#define FILE_DELETED   0x550
>  struct cifsFileInfo {
>         /* following two lists are protected by tcon->open_file_lock */
>         struct list_head tlist; /* pointer to next fid owned by tcon */
> diff --git a/fs/smb/client/cifsproto.h b/fs/smb/client/cifsproto.h
> index 770db9026850..67f75497db02 100644
> --- a/fs/smb/client/cifsproto.h
> +++ b/fs/smb/client/cifsproto.h
> @@ -292,6 +292,10 @@ extern void cifs_close_all_deferred_files(struct cif=
s_tcon *cifs_tcon);
>
>  extern void cifs_close_deferred_file_under_dentry(struct cifs_tcon *cifs=
_tcon,
>                                 const char *path);
> +
> +extern void cifs_mark_open_handles_for_deleted_file(struct cifs_tcon
> +                               *cifs_tcon, const char *path);
> +
>  extern struct TCP_Server_Info *
>  cifs_get_tcp_session(struct smb3_fs_context *ctx,
>                      struct TCP_Server_Info *primary_server);
> diff --git a/fs/smb/client/file.c b/fs/smb/client/file.c
> index b75282c204da..4c87ddbe8a5a 100644
> --- a/fs/smb/client/file.c
> +++ b/fs/smb/client/file.c
> @@ -1085,7 +1085,7 @@ int cifs_close(struct inode *inode, struct file *fi=
le)
>                 if ((cifs_sb->ctx->closetimeo && cinode->oplock =3D=3D CI=
FS_CACHE_RHW_FLG)
>                     && cinode->lease_granted &&
>                     !test_bit(CIFS_INO_CLOSE_ON_LOCK, &cinode->flags) &&
> -                   dclose) {
> +                   dclose && !(cfile->f_flags & FILE_DELETED)) {
>                         if (test_and_clear_bit(CIFS_INO_MODIFIED_ATTR, &c=
inode->flags)) {
>                                 inode_set_mtime_to_ts(inode,
>                                                       inode_set_ctime_cur=
rent(inode));
> diff --git a/fs/smb/client/inode.c b/fs/smb/client/inode.c
> index 24489e1e238a..be5bc01319d8 100644
> --- a/fs/smb/client/inode.c
> +++ b/fs/smb/client/inode.c
> @@ -1822,6 +1822,8 @@ int cifs_unlink(struct inode *dir, struct dentry *d=
entry)
>         cifs_inode =3D CIFS_I(dir);
>         CIFS_I(dir)->time =3D 0;  /* force revalidate of dir as well */
>  unlink_out:
> +       if (rc =3D=3D 0)
> +               cifs_mark_open_handles_for_deleted_file(tcon, full_path);
>         free_dentry_path(page);
>         kfree(attrs);
>         free_xid(xid);
> diff --git a/fs/smb/client/misc.c b/fs/smb/client/misc.c
> index 0748d7b757b9..ebc867bfc034 100644
> --- a/fs/smb/client/misc.c
> +++ b/fs/smb/client/misc.c
> @@ -853,6 +853,28 @@ cifs_close_deferred_file_under_dentry(struct cifs_tc=
on *tcon, const char *path)
>         free_dentry_path(page);
>  }
>
> +/*
> + * If a dentry has been deleted, all corresponding open handles should k=
now the
> + * same, so that we do not defer close them.
> + */
> +void
> +cifs_mark_open_handles_for_deleted_file(struct cifs_tcon *tcon, const ch=
ar
> +*path)
> +{
> +       struct cifsFileInfo *cfile;
> +       void *page;
> +       const char *full_path;
> +
> +       page =3D alloc_dentry_path();
> +       spin_lock(&tcon->open_file_lock);
> +       list_for_each_entry(cfile, &tcon->openFileList, tlist) {
> +               full_path =3D build_path_from_dentry(cfile->dentry, page)=
;
> +               if (strcmp(full_path, path) =3D=3D 0)
> +                       cfile->f_flags |=3D FILE_DELETED;
> +       }
> +       spin_unlock(&tcon->open_file_lock);
> +}
> +
>  /* parses DFS referral V3 structure
>   * caller is responsible for freeing target_nodes
>   * returns:
> --
> 2.39.2
>
>


--=20
Thanks,

Steve