From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Layton Subject: Re: [PATCH v2] ceph: fix recursively call between ceph_set_acl and __ceph_setattr Date: Wed, 19 Apr 2017 06:40:41 -0400 Message-ID: <1492598441.2783.4.camel@redhat.com> References: <20170419103622.52060-1-zyan@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: Received: from mail-qt0-f173.google.com ([209.85.216.173]:35631 "EHLO mail-qt0-f173.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1762291AbdDSKkn (ORCPT ); Wed, 19 Apr 2017 06:40:43 -0400 Received: by mail-qt0-f173.google.com with SMTP id y33so15331083qta.2 for ; Wed, 19 Apr 2017 03:40:43 -0700 (PDT) In-Reply-To: <20170419103622.52060-1-zyan@redhat.com> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: "Yan, Zheng" , ceph-devel@vger.kernel.org On Wed, 2017-04-19 at 18:36 +0800, Yan, Zheng wrote: > 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. > > Link: http://tracker.ceph.com/issues/19688 > Reported-by: Jerry Lee > Tested-by: Luis Henriques > Signed-off-by: "Yan, Zheng" > --- > fs/ceph/inode.c | 24 ++++++++++++------------ > 1 file changed, 12 insertions(+), 12 deletions(-) > > diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c > index efee88c..976fd3a 100644 > --- a/fs/ceph/inode.c > +++ b/fs/ceph/inode.c > @@ -2078,11 +2078,6 @@ int __ceph_setattr(struct inode *inode, 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; > @@ -2096,13 +2091,13 @@ int __ceph_setattr(struct inode *inode, 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) { > + if (mask & CEPH_SETATTR_SIZE) { > + __ceph_do_pending_vmtruncate(inode); > + } > + } > return err; > } > nit: Could drop some of the curly braces and nesting above: if (err >= 0 && (mask & CEPH_SETATTR_SIZE)) __ceph_do_pending_vmtruncate(inode); > @@ -2121,7 +2116,12 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr) > if (err != 0) > return err; > > - return __ceph_setattr(inode, attr); > + err = __ceph_setattr(inode, attr); > + > + if (err >= 0 && (attr->ia_valid & ATTR_MODE)) { > + err = posix_acl_chmod(inode, attr->ia_mode); > + } > + return err; > } > > /* Much better without the bool argument though :) Reviewed-by: Jeff Layton