On Fri, May 15, 2020 at 01:16:59PM -0500, Eric W. Biederman wrote:

Tetsuo Handa [off-list ref] writes:

quoted

This is

        if (sb->s_magic == PROC_SUPER_MAGIC && *pos == '/') {
                char *ep;
                const pid_t pid = (pid_t) simple_strtoul(pos + 1, &ep, 10);
                struct pid_namespace *proc_pidns = proc_pid_ns(d_inode(dentry)); // <= here

                if (*ep == '/' && pid && pid ==
                    task_tgid_nr_ns(current, proc_pidns)) {

which was added by commit c59f415a7cb6e1e1 ("Use proc_pid_ns() to get pid_namespace from the proc superblock").

@@ -161,9 +162,10 @@ static char *tomoyo_get_local_path(struct dentry *dentry, char * const buffer,
        if (sb->s_magic == PROC_SUPER_MAGIC && *pos == '/') {
                char *ep;
                const pid_t pid = (pid_t) simple_strtoul(pos + 1, &ep, 10);
+               struct pid_namespace *proc_pidns = proc_pid_ns(d_inode(dentry));

                if (*ep == '/' && pid && pid ==
-                   task_tgid_nr_ns(current, sb->s_fs_info)) {
+                   task_tgid_nr_ns(current, proc_pidns)) {
                        pos = ep - 5;
                        if (pos < buffer)
                                goto out;

Alexey and Eric, any clue?

Looking at the stack backtrace this is happening as part of creating a
file or a device node.  The dentry that is passed in most likely
comes from d_alloc_parallel.  So we have d_inode == NULL.

I want to suggest doing the very simple fix:

-	if (sb->s_magic == PROC_SUPER_MAGIC && *pos == '/') {
+	if (sb->s_magic == PROC_SUPER_MAGIC && *pos == '/' && denty->d_inode) {

But I don't know if there are any other security hooks early in lookup,
that could be called for an already existing dentry.

So it looks like we need a version proc_pid_ns that works for a dentry,
or a superblock.

Alex do you think you can code up an patch against my proc-next branch
to fix this?

Sure.

-- 
Rgrds, legion