On Wed, 2021-07-07 at 15:56 +0100, Luis Henriques wrote:
> On Wed, Jul 07, 2021 at 03:32:13PM +0100, Luis Henriques wrote:
> > On Wed, Jul 07, 2021 at 08:19:25AM -0400, Jeff Layton wrote:
> > > On Wed, 2021-07-07 at 19:19 +0800, Xiubo Li wrote:
> > > > On 7/7/21 6:47 PM, Luis Henriques wrote:
> > > > > On Fri, Jun 25, 2021 at 09:58:16AM -0400, Jeff Layton wrote:
> > > > > > ...and store them in the ceph_inode_info.
> > > > > >
> > > > > > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> > > > > > ---
> > > > > > fs/ceph/file.c | 2 ++
> > > > > > fs/ceph/inode.c | 18 ++++++++++++++++++
> > > > > > fs/ceph/mds_client.c | 44 ++++++++++++++++++++++++++++++++++++++++++++
> > > > > > fs/ceph/mds_client.h | 4 ++++
> > > > > > fs/ceph/super.h | 6 ++++++
> > > > > > 5 files changed, 74 insertions(+)
> > > > > >
> > > > > > diff --git a/fs/ceph/file.c b/fs/ceph/file.c
> > > > > > index 2cda398ba64d..ea0e85075b7b 100644
> > > > > > --- a/fs/ceph/file.c
> > > > > > +++ b/fs/ceph/file.c
> > > > > > @@ -592,6 +592,8 @@ static int ceph_finish_async_create(struct inode *dir, struct inode *inode,
> > > > > > iinfo.xattr_data = xattr_buf;
> > > > > > memset(iinfo.xattr_data, 0, iinfo.xattr_len);
> > > > > >
> > > > > > + /* FIXME: set fscrypt_auth and fscrypt_file */
> > > > > > +
> > > > > > in.ino = cpu_to_le64(vino.ino);
> > > > > > in.snapid = cpu_to_le64(CEPH_NOSNAP);
> > > > > > in.version = cpu_to_le64(1); // ???
> > > > > > diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
> > > > > > index f62785e4dbcb..b620281ea65b 100644
> > > > > > --- a/fs/ceph/inode.c
> > > > > > +++ b/fs/ceph/inode.c
> > > > > > @@ -611,6 +611,13 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
> > > > > >
> > > > > > ci->i_meta_err = 0;
> > > > > >
> > > > > > +#ifdef CONFIG_FS_ENCRYPTION
> > > > > > + ci->fscrypt_auth = NULL;
> > > > > > + ci->fscrypt_auth_len = 0;
> > > > > > + ci->fscrypt_file = NULL;
> > > > > > + ci->fscrypt_file_len = 0;
> > > > > > +#endif
> > > > > > +
> > > > > > return &ci->vfs_inode;
> > > > > > }
> > > > > >
> > > > > > @@ -619,6 +626,9 @@ void ceph_free_inode(struct inode *inode)
> > > > > > struct ceph_inode_info *ci = ceph_inode(inode);
> > > > > >
> > > > > > kfree(ci->i_symlink);
> > > > > > +#ifdef CONFIG_FS_ENCRYPTION
> > > > > > + kfree(ci->fscrypt_auth);
> > > > > > +#endif
> > > > > > kmem_cache_free(ceph_inode_cachep, ci);
> > > > > > }
> > > > > >
> > > > > > @@ -1021,6 +1031,14 @@ int ceph_fill_inode(struct inode *inode, struct page *locked_page,
> > > > > > xattr_blob = NULL;
> > > > > > }
> > > > > >
> > > > > > + if (iinfo->fscrypt_auth_len && !ci->fscrypt_auth) {
> > > > > > + ci->fscrypt_auth_len = iinfo->fscrypt_auth_len;
> > > > > > + ci->fscrypt_auth = iinfo->fscrypt_auth;
> > > > > > + iinfo->fscrypt_auth = NULL;
> > > > > > + iinfo->fscrypt_auth_len = 0;
> > > > > > + inode_set_flags(inode, S_ENCRYPTED, S_ENCRYPTED);
> > > > > > + }
> > > > > I think we also need to free iinfo->fscrypt_auth here if ci->fscrypt_auth
> > > > > is already set. Something like:
> > > > >
> > > > > if (iinfo->fscrypt_auth_len) {
> > > > > if (!ci->fscrypt_auth) {
> > > > > ...
> > > > > } else {
> > > > > kfree(iinfo->fscrypt_auth);
> > > > > iinfo->fscrypt_auth = NULL;
> > > > > }
> > > > > }
> > > > >
> > > > IMO, this should be okay because it will be freed in
> > > > destroy_reply_info() when putting the request.
> > > >
> > > >
> > >
> > > Yes. All of that should get cleaned up with the request.
> >
> > Hmm... ok, so maybe I missed something because I *did* saw kmemleak
> > complaining. Maybe it was on the READDIR path. /me goes look again.
>
> Ah, that was indeed the problem. So, here's a quick hack to fix
> destroy_reply_info() so that it also frees the extra memory from READDIR:
>
> @@ -686,12 +686,23 @@ static int parse_reply_info(struct ceph_mds_session *s, struct ceph_msg *msg,
>
> static void destroy_reply_info(struct ceph_mds_reply_info_parsed *info)
> {
> + int i = 0;
> +
> kfree(info->diri.fscrypt_auth);
> kfree(info->diri.fscrypt_file);
> kfree(info->targeti.fscrypt_auth);
> kfree(info->targeti.fscrypt_file);
> if (!info->dir_entries)
> return;
> +
> + for (i = 0; i < info->dir_nr; i++) {
> + struct ceph_mds_reply_dir_entry *rde = info->dir_entries + i;
> + if (rde->inode.fscrypt_auth_len)
> + kfree(rde->inode.fscrypt_auth);
> + }
> +
> free_pages((unsigned long)info->dir_entries, get_order(info->dir_buf_size));
> }
>
>
Well spotted! I'll plan to incorporate a fix like that soon.
--
Jeff Layton <jlayton@xxxxxxxxxx>
|