> On 2 May 2017, at 19:10, Luis Henriques <lhenriques@xxxxxxxx> wrote: > > On Sun, Apr 30, 2017 at 04:04:39PM +0200, Greg Kroah-Hartman wrote: >> >> The patch below does not apply to the 4.4-stable tree. >> If someone wants it applied there, or to any other stable or longterm >> tree, then please email the backport, including the original git commit >> id to <stable@xxxxxxxxxxxxxxx>. >> > > Here's a backport I've tested with 4.4. > > Cheers, > -- > Luís > > From beab414f9df0c865e73e651393538cde1fa20a3d Mon Sep 17 00:00:00 2001 > From: "Yan, Zheng" <zyan@xxxxxxxxxx> > Date: Wed, 19 Apr 2017 10:01:48 +0800 > Subject: [PATCH] ceph: fix recursion between ceph_set_acl() and > __ceph_setattr() > > commit 8179a101eb5f4ef0ac9a915fcea9a9d3109efa90 upstream. > > ceph_set_acl() calls __ceph_setattr() if the setacl operation needs > to modify inode's i_mode. __ceph_setattr() updates inode's i_mode, > then calls posix_acl_chmod(). > > The problem is that __ceph_setattr() calls posix_acl_chmod() before > sending the setattr request. The get_acl() call in posix_acl_chmod() > can trigger a getxattr request. The reply of the getxattr request > can restore inode's i_mode to its old value. The set_acl() call in > posix_acl_chmod() sees old value of inode's i_mode, so it calls > __ceph_setattr() again. > > Cc: stable@xxxxxxxxxxxxxxx # needs backporting for < 4.9 > Link: http://tracker.ceph.com/issues/19688 > Reported-by: Jerry Lee <leisurelysw24@xxxxxxxxx> > Signed-off-by: "Yan, Zheng" <zyan@xxxxxxxxxx> > Reviewed-by: Jeff Layton <jlayton@xxxxxxxxxx> > Tested-by: Luis Henriques <lhenriques@xxxxxxxx> > Signed-off-by: Ilya Dryomov <idryomov@xxxxxxxxx> > [luis: __ceph_setattr doesn't exist in 4.4, implement all the changes > directly in ceph_setattr. The wrapper function was introduced only in > commit a26feccaba29 ("ceph: Get rid of d_find_alias in ceph_set_acl").] > Signed-off-by: Luis Henriques <lhenriques@xxxxxxxx> > --- > fs/ceph/inode.c | 18 +++++++----------- > 1 file changed, 7 insertions(+), 11 deletions(-) > > diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c > index d98536c8abfc..e3d19c3d246d 100644 > --- a/fs/ceph/inode.c > +++ b/fs/ceph/inode.c > @@ -1975,11 +1975,6 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr) > if (inode_dirty_flags) > __mark_inode_dirty(inode, inode_dirty_flags); > > - if (ia_valid & ATTR_MODE) { > - err = posix_acl_chmod(inode, attr->ia_mode); > - if (err) > - goto out_put; > - } > > if (mask) { > req->r_inode = inode; > @@ -1993,13 +1988,14 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr) > ceph_cap_string(dirtied), mask); > > ceph_mdsc_put_request(req); > - if (mask & CEPH_SETATTR_SIZE) > - __ceph_do_pending_vmtruncate(inode); > - ceph_free_cap_flush(prealloc_cf); > - return err; > -out_put: > - ceph_mdsc_put_request(req); > ceph_free_cap_flush(prealloc_cf); > + > + if (err >= 0 && (mask & CEPH_SETATTR_SIZE)) > + __ceph_do_pending_vmtruncate(inode); > + > + if (err >= 0 && (attr->ia_valid & ATTR_MODE)) > + err = posix_acl_chmod(inode, attr->ia_mode); > + > return err; > } Thank you for the patch. but it does not work because ceph_set_acl() calls ceph_setattr() directly. I think we need to introduce __ceph_setattr() and make ceph_set_acl() call __ceph_setattr() Regards Yan, Zheng