Re: [PATCH ghak100 V2 2/2] audit: ignore fcaps on umount
On Fri, Feb 1, 2019 at 3:42 PM Nathan Chancellor
<natechancellor(a)gmail.com> wrote:
On Wed, Jan 23, 2019 at 01:35:00PM -0500, Richard Guy Briggs wrote:
> Don't fetch fcaps when umount2 is called to avoid a process hang while
> it waits for the missing resource to (possibly never) re-appear.
>
> Note the comment above user_path_mountpoint_at():
> * A umount is a special case for path walking. We're not actually interested
> * in the inode in this situation, and ESTALE errors can be a problem. We
> * simply want track down the dentry and vfsmount attached at the mountpoint
> * and avoid revalidating the last component.
>
> This can happen on ceph, cifs, 9p, lustre, fuse (gluster) or NFS.
>
> Please see the github issue tracker
> https://github.com/linux-audit/audit-kernel/issues/100
>
> Signed-off-by: Richard Guy Briggs <rgb(a)redhat.com>
> ---
> fs/namei.c | 2 +-
> fs/namespace.c | 2 ++
> include/linux/audit.h | 15 ++++++++++-----
> include/linux/namei.h | 3 +++
> kernel/audit.c | 10 +++++++++-
> kernel/audit.h | 2 +-
> kernel/auditsc.c | 6 +++---
> 7 files changed, 29 insertions(+), 11 deletions(-)
...
> /* Copy inode data into an audit_names. */
> void audit_copy_inode(struct audit_names *name, const struct dentry *dentry,
> - struct inode *inode)
> + struct inode *inode, unsigned int flags)
> {
> name->ino = inode->i_ino;
> name->dev = inode->i_sb->s_dev;
> @@ -2120,6 +2124,10 @@ void audit_copy_inode(struct audit_names *name, const struct
dentry *dentry,
> name->gid = inode->i_gid;
> name->rdev = inode->i_rdev;
> security_inode_getsecid(inode, &name->osid);
> + if (flags & AUDIT_INODE_NOEVAL) {
I don't know if this has been reported or if I am missing something but
on next-20190201, this line causes an error with arm allyesconfig (as
CONFIG_AUDITSYCALL doesn't get selected):
...
CC kernel/audit.o
kernel/audit.c: In function 'audit_copy_inode':
kernel/audit.c:2130:14: error: 'AUDIT_INODE_NOEVAL' undeclared (first use in this
function); did you mean 'AUDIT_TYPE_NORMAL'?
if (flags & AUDIT_INODE_NOEVAL) {
^~~~~~~~~~~~~~~~~~
AUDIT_TYPE_NORMAL
kernel/audit.c:2130:14: note: each undeclared identifier is reported only once for each
function it appears in
make[2]: *** [scripts/Makefile.build:277: kernel/audit.o] Error 1
make[1]: *** [Makefile:1699: kernel/audit.o] Error 2
make: *** [Makefile:296: __build_one_by_one] Error 2
I hadn't seen this reported to the audit list yet, thanks for letting us now.
Richard, please submit a patch to fix this ASAP. Looking at this, the
obvious fix is to move audit_copy_inode() to auditsc.c, but I'm not
sure if that itself is going to cause problems (it doesn't look like
it). Actually, thinking out loud, I wonder if we shouldn't move
audit_log_cap(), audit_log_fcaps(), audit_copy_fcaps(), and
audit_log_name() too?
--
paul moore
www.paul-moore.com