On Mon, 28 Jun 2010 18:56:09 +0530
Suresh Jayaraman <sjayara...@suse.de> wrote:
> On 06/28/2010 04:40 PM, Jeff Layton wrote:
> > Reduce false inode collisions by using the CreationTime like an
> > i_generation field. This way, even if the server ends up reusing
> > a uniqueid after a delete/create cycle, we can avoid matching
> > the inode incorrectly.
> >
> > Signed-off-by: Jeff Layton <jlay...@redhat.com>
> > ---
> > fs/cifs/cifsfs.c | 2 ++
> > fs/cifs/cifsglob.h | 2 ++
> > fs/cifs/inode.c | 6 ++++++
> > fs/cifs/readdir.c | 1 +
> > 4 files changed, 11 insertions(+), 0 deletions(-)
> >
> > diff --git a/fs/cifs/cifsfs.c b/fs/cifs/cifsfs.c
> > index 484e52b..e409216 100644
> > --- a/fs/cifs/cifsfs.c
> > +++ b/fs/cifs/cifsfs.c
> > @@ -314,6 +314,8 @@ cifs_alloc_inode(struct super_block *sb)
> > cifs_inode->invalid_mapping = false;
> > cifs_inode->vfs_inode.i_blkbits = 14; /* 2**14 = CIFS_MAX_MSGSIZE */
> > cifs_inode->server_eof = 0;
> > + cifs_inode->uniqueid = 0;
> > + cifs_inode->createtime = 0;
> >
> > /* Can not set i_flags here - they get immediately overwritten
> > to zero by the VFS */
> > diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h
> > index e15d7a5..a0fd2c2 100644
> > --- a/fs/cifs/cifsglob.h
> > +++ b/fs/cifs/cifsglob.h
> > @@ -382,6 +382,7 @@ struct cifsInodeInfo {
> > bool invalid_mapping:1; /* pagecache is invalid */
> > u64 server_eof; /* current file size on server */
> > u64 uniqueid; /* server inode number */
> > + u64 createtime; /* creation time on server */
> > struct inode vfs_inode;
> > };
> >
> > @@ -499,6 +500,7 @@ struct cifs_fattr {
> > u64 cf_uniqueid;
> > u64 cf_eof;
> > u64 cf_bytes;
> > + u64 cf_createtime;
> > uid_t cf_uid;
> > gid_t cf_gid;
> > umode_t cf_mode;
> > diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
> > index f64b95f..35e2466 100644
> > --- a/fs/cifs/inode.c
> > +++ b/fs/cifs/inode.c
> > @@ -487,6 +487,7 @@ cifs_all_info_to_fattr(struct cifs_fattr *fattr,
> > FILE_ALL_INFO *info,
> >
> > fattr->cf_eof = le64_to_cpu(info->EndOfFile);
> > fattr->cf_bytes = le64_to_cpu(info->AllocationSize);
> > + fattr->cf_createtime = le64_to_cpu(info->CreationTime);
> >
> > if (fattr->cf_cifsattrs & ATTR_DIRECTORY) {
> > fattr->cf_mode = S_IFDIR | cifs_sb->mnt_dir_mode;
> > @@ -727,6 +728,10 @@ cifs_find_inode(struct inode *inode, void *opaque)
> > if (CIFS_I(inode)->uniqueid != fattr->cf_uniqueid)
> > return 0;
> >
> > + /* use createtime like an i_generation field */
> > + if (CIFS_I(inode)->createtime != fattr->cf_createtime)
> > + return 0;
> > +
> > /* don't match inode of different type */
> > if ((inode->i_mode & S_IFMT) != (fattr->cf_mode & S_IFMT))
> > return 0;
> > @@ -750,6 +755,7 @@ cifs_init_inode(struct inode *inode, void *opaque)
> > struct cifs_fattr *fattr = (struct cifs_fattr *) opaque;
> >
> > CIFS_I(inode)->uniqueid = fattr->cf_uniqueid;
> > + CIFS_I(inode)->createtime = fattr->cf_createtime;
> > return 0;
> > }
> >
> > diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
> > index daf1753..8bd18be 100644
> > --- a/fs/cifs/readdir.c
> > +++ b/fs/cifs/readdir.c
> > @@ -160,6 +160,7 @@ cifs_dir_info_to_fattr(struct cifs_fattr *fattr,
> > FILE_DIRECTORY_INFO *info,
> > fattr->cf_cifsattrs = le32_to_cpu(info->ExtFileAttributes);
> > fattr->cf_eof = le64_to_cpu(info->EndOfFile);
> > fattr->cf_bytes = le64_to_cpu(info->AllocationSize);
> > + fattr->cf_createtime = le64_to_cpu(info->CreationTime);
>
> cifs_std_info_to_fattr() needs this too?
>
I don't think so. If we're using that function then we're probably
going to be using the legacy SMBQueryInformation calls too. Those do
not return a CreateTime and it ends up being set to 0. This *should*
do the right thing I think...
It would be nice to do this in a more definite way somehow though. The
legacy QPathInfo codepath is a bit of a mess...
> > fattr->cf_atime = cifs_NTtimeToUnix(info->LastAccessTime);
> > fattr->cf_ctime = cifs_NTtimeToUnix(info->ChangeTime);
> > fattr->cf_mtime = cifs_NTtimeToUnix(info->LastWriteTime);
>
>
--
Jeff Layton <jlay...@redhat.com>
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
