On Thu, Dec 09, 2021 at 02:38:13PM -0500, James Bottomley wrote: > On Thu, 2021-12-09 at 10:30 -0500, James Bottomley wrote: > > On Thu, 2021-12-09 at 15:37 +0100, Christian Brauner wrote: > > > On Thu, Dec 09, 2021 at 03:34:28PM +0100, Christian Brauner wrote: > > > > On Wed, Dec 08, 2021 at 05:18:17PM -0500, Stefan Berger wrote: > > > > > Move the dentries into the ima_namespace for reuse by > > > > > virtualized > > > > > SecurityFS. Implement function freeing the dentries in order of > > > > > files and symlinks before directories. > > > > > > > > > > Signed-off-by: Stefan Berger <stefanb@xxxxxxxxxxxxx> > > > > > --- > > > > > > > > This doesn't work as implemented, I think. > > > > > > > > What I would have preferred and what I tried to explain in the > > > > earlier review was: > > > > Keep the dentry stashing global since it is only needed for > > > > init_ima_ns. > > > > Then struct ima_namespace becomes way smaller and simpler. > > > > If you do that then it makes sense to remove the additional > > > > dget() in securityfs_create_dentry() for non-init_ima_ns. > > > > Then you can rely on auto-cleanup in .kill_sb() or on > > > > ima_securityfs_init() failure and you only need to call > > > > ima_fs_ns_free_dentries() if ns != init_ima_ns. > > > > > > > > IIuc, it seems you're currently doing one dput() too many since > > > > you're calling securityfs_remove() in the error path for non- > > > > init_ima_ns which relies on the previous increased dget() which > > > > we removed. > > > > > > If you really want to move the dentry stashing into struct > > > ima_namespace even though it's really unnecessary then you may as > > > well not care about the auto-cleanup and keep that additional > > > ima_fs_ns_free_dentries(ns) call in .kill_sb(). But I really think > > > not dragging dentry stashing into struct ima_namespace is the > > > correct way to go about this. > > > > We, unfortunately, do have one case we can't avoid stashing for the > > policy file. It's this code in ima_release_policy: > > > > > #if !defined(CONFIG_IMA_WRITE_POLICY) && > > > !defined(CONFIG_IMA_READ_POLICY) > > > securityfs_remove(ns->dentry[IMAFS_DENTRY_IMA_POLICY]); > > > ns->dentry[IMAFS_DENTRY_IMA_POLICY] = NULL; > > > > > > > What it does is that in certain config options, the policy file entry > > gets removed from the securityfs ima directory after you write to it. > > This is what I have incremental to v5 that corrects all of this. It > actually keeps every dentry reference (including init_user_ns ones) at > 1 so they can be reaped on unmount. For the remove case it does > d_delete and then puts the only reference. This means > securityfs_remove() works for the namespaced policy file as well. > > I also got rid of the spurious initialized check in ima_securityfs_init > because it prevents you doing a mount;umount;mount on securityfs within > a namespace. > > There's still the problem that if you write the policy, making the file > disappear then unmount and remount securityfs it will come back. My > guess for fixing this is that we only stash the policy file reference, > create it if NULL but then set the pointer to PTR_ERR(-EINVAL) or > something and refuse to create it for that value. Some sort of indicator that gets stashed in struct ima_ns that the file does not get recreated on consecutive mounts. That shouldn't be hard to fix. > > James > > --- > > From 7de285a81ff06b6e0eb2c6db24810aeef9f6dd17 Mon Sep 17 00:00:00 2001 > From: James Bottomley <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> > Date: Thu, 9 Dec 2021 19:33:49 +0000 > Subject: [PATCH] fix dentry ref counting > > --- > security/inode.c | 12 ++---------- > security/integrity/ima/ima_fs.c | 4 ---- > 2 files changed, 2 insertions(+), 14 deletions(-) > > diff --git a/security/inode.c b/security/inode.c > index eaccba7017d9..b53152f7a625 100644 > --- a/security/inode.c > +++ b/security/inode.c > @@ -178,8 +178,6 @@ static struct dentry *securityfs_create_dentry(const char *name, umode_t mode, > inode->i_fop = fops; > } > d_instantiate(dentry, inode); > - if (ns == &init_user_ns) > - dget(dentry); > inode_unlock(dir); > return dentry; > > @@ -317,21 +315,15 @@ EXPORT_SYMBOL_GPL(securityfs_create_symlink); > void securityfs_remove(struct dentry *dentry) > { > struct user_namespace *ns = dentry->d_sb->s_user_ns; > - struct inode *dir; > > if (!dentry || IS_ERR(dentry)) > return; > > - dir = d_inode(dentry->d_parent); > - inode_lock(dir); > if (simple_positive(dentry)) { > - if (d_is_dir(dentry)) > - simple_rmdir(dir, dentry); > - else > - simple_unlink(dir, dentry); > + d_delete(dentry); Not, that doesn't work. You can't just call d_delete() and dput() and even if I wouldn't advise it. And you also can't do this without taking the inode lock on the directory. simple_rmdir()/simple_unlink() take care to update various inode fields in the parent dir and handle link counts. This really wants to be sm like struct inode *parent_inode; parent_inode = d_inode(dentry->d_parent); inode_lock(parent_inode); if (simple_positive(dentry)) { dget(dentry); if (d_is_dir(dentry) simple_unlink(parent_inode, dentry); else simple_unlink(parent_inode, dentry); d_delete(dentry); dput(dentry); } inode_unlock(parent_inode); > dput(dentry); > } > - inode_unlock(dir); > + > if (ns == &init_user_ns) > simple_release_fs(&init_securityfs_mount, > &init_securityfs_mount_count); > diff --git a/security/integrity/ima/ima_fs.c b/security/integrity/ima/ima_fs.c > index 778983fd9a73..077a6ff46858 100644 > --- a/security/integrity/ima/ima_fs.c > +++ b/security/integrity/ima/ima_fs.c > @@ -466,10 +466,6 @@ int ima_securityfs_init(struct user_namespace *user_ns, struct dentry *root) > struct ima_namespace *ns = user_ns->ima_ns; > struct dentry *ima_dir; > > - /* already initialized? */ > - if (ns->dentry[IMAFS_DENTRY_INTEGRITY_DIR]) > - return 0; > - > /* FIXME: update when evm and integrity are namespaced */ > if (user_ns != &init_user_ns) { > ns->dentry[IMAFS_DENTRY_INTEGRITY_DIR] = > -- > 2.33.0 > > >