On Fri, Jul 31, 2015 at 03:52:38PM -0700, Linus Torvalds wrote:
> Is that correct? Maybe, I haven't checked. And maybe it's a big bad
> bug. Regardless, it sure as hell isn't just changing the order of the
> access to those fields. That "DCACHE_ENTRY_TYPE | DCACHE_FALLTHRU"
> clearing came from __d_instantiate(), but now it hits __d_obtain_alias
> too.
Actually, the shit had hit the fan earlier. Look: in
commit b18825a7c8e37a7cf6abb97a12a6ad71af160de7
Author: David Howells <dhowells@xxxxxxxxxx>
Date: Thu Sep 12 19:22:53 2013 +0100
VFS: Put a small type field into struct dentry::d_flags
we have this:
@@ -1823,7 +1794,7 @@ static int link_path_walk(const char *name, struct nameidata *nd)
if (err)
return err;
}
- if (!can_lookup(nd->inode)) {
+ if (!d_is_directory(nd->path.dentry)) {
err = -ENOTDIR;
break;
}
And that has turned the check done to an inode that *was* ours at some
point (i.e. fetching it had been followed by checking that ->d_seq had
been still valid) into something completely unprotected. Suppose we
are in lazy mode and somebody had evicted nd->path.dentry after we'd looked
it up and before that check. Sure, its ->d_seq had been bumped by that,
and we would've failed anyway. With ECHILD. Which, unlike ENOTDIR, is
"repeat in non-lazy mode".
AFAICS, that's where the problem is. It affects only RCU mode and only
the places where dentry isn't pinned. That place in link_path_walk() is
trivial - we just need to do
if (unlikely(!d_can_lookup(nd->path.dentry))) {
if (nd->flags & LOOKUP_RCU) {
if (unlazy_walk(nd, NULL, 0))
return -ECHILD;
}
return -ENOTDIR;
}
there. AFAICS, other places of that sort are not a problem anymore.
Folks, could you check if this fixes the problems you are seeing?
diff --git a/fs/namei.c b/fs/namei.c
index ae4e4c1..b16c3a7 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1954,7 +1954,11 @@ OK:
continue;
}
}
- if (unlikely(!d_can_lookup(nd->path.dentry)))
+ if (unlikely(!d_can_lookup(nd->path.dentry))) {
+ if (nd->flags & LOOKUP_RCU) {
+ if (unlazy_walk(nd, NULL, 0))
+ return -ECHILD;
+ }
return -ENOTDIR;
}
}
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
