Re: BUG_ON hit in "reimplement path_mountpoint() with less magic" [Was: Re: linux-next: Tree for Jan 13]

[Jason A. Donenfeld]

Hey Al,

On Mon, Jan 13, 2020 at 2:23 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> On Mon, Jan 13, 2020 at 01:27:34PM -0500, Jason A. Donenfeld wrote:
> > Hi Al,
> >
> > Your "reimplement path_mountpoint() with less magic" commit is causing
> > oopses on the linux-next tests running on https://build.wireguard.com/ .
> > Here's one such crash with debug symbols:
>
> Check if that persists with the version currently in #fixes (commit
> 7a955b7363b8).  It ought to be fixed there; quick check is to look
> at fs/namei.c:path_mountpoint() in your tree - it should read
>         if (!err && (nd->flags & LOOKUP_RCU))
>                 err = unlazy_walk(nd);
> not
>         if (!err)
>                 err = unlazy_walk(nd);
> as in the broken variant.

That does the trick. Glad that's sorted. Thanks.

Jason