| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20200624012901.GC18460@fieldses.org>
Date: Tue, 23 Jun 2020 21:29:01 -0400
From: "J. Bruce Fields" <bfields@...ldses.org>
To: Luo Xiaogang <lxgrxd@....com>
Cc: chuck.lever@...cle.com, linux-nfs@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] nfsd: fix kernel crash when load nfsd in docker
On Mon, Jun 15, 2020 at 03:12:11PM +0800, Luo Xiaogang wrote:
> We load nfsd module in the docker container, kernel crash as following.
>
> The 'current->nsproxy->net_ns->gen->ptr[nfsd_net_id]' is overflow in the
> nfsd_init_net.
>
> We should use the net_ns which is being init in the nfsd_init_net,
> not the 'current->nsproxy->net_ns'.
Thanks! Actually, I think my problem was that net init and exit are
just the wrong place to be doing this--I moved them to nfsd start/stop
instead.
And then that exposed the fact that I had an inode leak.
Do the following two patches help?
--b.
>From 16f954bd5c481596a63271a91963bf260e2f3f46 Mon Sep 17 00:00:00 2001
From: "J. Bruce Fields" <bfields@...hat.com>
Date: Tue, 23 Jun 2020 16:00:33 -0400
Subject: [PATCH 1/2] nfsd4: fix nfsdfs reference count loop
We don't drop the reference on the nfsdfs filesystem with
mntput(nn->nfsd_mnt) until nfsd_exit_net(), but that won't be called
until the nfsd module's unloaded, and we can't unload the module as long
as there's a reference on nfsdfs. So this prevents module unloading.
Signed-off-by: J. Bruce Fields <bfields@...hat.com>
---
fs/nfsd/nfs4state.c | 8 +++++++-
fs/nfsd/nfsctl.c | 22 ++++++++++++----------
fs/nfsd/nfsd.h | 3 +++
3 files changed, 22 insertions(+), 11 deletions(-)
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index bb3d2c32664a..cce2510b2cca 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -7912,9 +7912,14 @@ nfs4_state_start_net(struct net *net)
struct nfsd_net *nn = net_generic(net, nfsd_net_id);
int ret;
- ret = nfs4_state_create_net(net);
+ ret = get_nfsdfs(net);
if (ret)
return ret;
+ ret = nfs4_state_create_net(net);
+ if (ret) {
+ mntput(nn->nfsd_mnt);
+ return ret;
+ }
locks_start_grace(net, &nn->nfsd4_manager);
nfsd4_client_tracking_init(net);
if (nn->track_reclaim_completes && nn->reclaim_str_hashtbl_size == 0)
@@ -7984,6 +7989,7 @@ nfs4_state_shutdown_net(struct net *net)
nfsd4_client_tracking_exit(net);
nfs4_state_destroy_net(net);
+ mntput(nn->nfsd_mnt);
}
void
diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index b68e96681522..cf98a81ca1ea 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -1424,6 +1424,18 @@ static struct file_system_type nfsd_fs_type = {
};
MODULE_ALIAS_FS("nfsd");
+int get_nfsdfs(struct net *net)
+{
+ struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+ struct vfsmount *mnt;
+
+ mnt = vfs_kern_mount(&nfsd_fs_type, SB_KERNMOUNT, "nfsd", NULL);
+ if (IS_ERR(mnt))
+ return PTR_ERR(mnt);
+ nn->nfsd_mnt = mnt;
+ return 0;
+}
+
#ifdef CONFIG_PROC_FS
static int create_proc_exports_entry(void)
{
@@ -1451,7 +1463,6 @@ unsigned int nfsd_net_id;
static __net_init int nfsd_init_net(struct net *net)
{
int retval;
- struct vfsmount *mnt;
struct nfsd_net *nn = net_generic(net, nfsd_net_id);
retval = nfsd_export_init(net);
@@ -1478,16 +1489,8 @@ static __net_init int nfsd_init_net(struct net *net)
init_waitqueue_head(&nn->ntf_wq);
seqlock_init(&nn->boot_lock);
- mnt = vfs_kern_mount(&nfsd_fs_type, SB_KERNMOUNT, "nfsd", NULL);
- if (IS_ERR(mnt)) {
- retval = PTR_ERR(mnt);
- goto out_mount_err;
- }
- nn->nfsd_mnt = mnt;
return 0;
-out_mount_err:
- nfsd_reply_cache_shutdown(nn);
out_drc_error:
nfsd_idmap_shutdown(net);
out_idmap_error:
@@ -1500,7 +1503,6 @@ static __net_exit void nfsd_exit_net(struct net *net)
{
struct nfsd_net *nn = net_generic(net, nfsd_net_id);
- mntput(nn->nfsd_mnt);
nfsd_reply_cache_shutdown(nn);
nfsd_idmap_shutdown(net);
nfsd_export_shutdown(net);
diff --git a/fs/nfsd/nfsd.h b/fs/nfsd/nfsd.h
index 36cdd81b6688..57c832d1b30f 100644
--- a/fs/nfsd/nfsd.h
+++ b/fs/nfsd/nfsd.h
@@ -90,6 +90,8 @@ void nfsd_destroy(struct net *net);
bool i_am_nfsd(void);
+int get_nfsdfs(struct net *);
+
struct nfsdfs_client {
struct kref cl_ref;
void (*cl_release)(struct kref *kref);
@@ -100,6 +102,7 @@ struct dentry *nfsd_client_mkdir(struct nfsd_net *nn,
struct nfsdfs_client *ncl, u32 id, const struct tree_descr *);
void nfsd_client_rmdir(struct dentry *dentry);
+
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
#ifdef CONFIG_NFSD_V2_ACL
extern const struct svc_version nfsd_acl_version2;
--
2.26.2
>From 51de3b460b39e862f7dcfd4d600e8de0afe73e29 Mon Sep 17 00:00:00 2001
From: "J. Bruce Fields" <bfields@...hat.com>
Date: Tue, 23 Jun 2020 21:01:19 -0400
Subject: [PATCH 2/2] nfsd: fix nfsdfs inode reference count leak
I don't understand this code well, but I'm seeing a warning about a
still-referenced inode on unmount, and every other similar filesystem
does a dput() here.
Signed-off-by: J. Bruce Fields <bfields@...hat.com>
---
fs/nfsd/nfsctl.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index cf98a81ca1ea..cd05732f8eaa 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -1335,6 +1335,7 @@ void nfsd_client_rmdir(struct dentry *dentry)
WARN_ON_ONCE(ret);
fsnotify_rmdir(dir, dentry);
d_delete(dentry);
+ dput(dentry);
inode_unlock(dir);
}
--
2.26.2
Powered by blists - more mailing lists