From:   ebiederm@xmission.com (Eric W. Biederman)
To:     Mike Christie <michael.christie@oracle.com>
Cc:     geert@linux-m68k.org, vverma@digitalocean.com, hdanton@sina.com,
        hch@infradead.org, stefanha@redhat.com, jasowang@redhat.com,
        mst@redhat.com, sgarzare@redhat.com,
        virtualization@lists.linux-foundation.org,
        christian.brauner@ubuntu.com, axboe@kernel.dk,
        linux-kernel@vger.kernel.org, Christoph Hellwig <hch@lst.de>
References: <20211129194707.5863-1-michael.christie@oracle.com>
        <20211129194707.5863-11-michael.christie@oracle.com>
Date:   Fri, 17 Dec 2021 13:01:51 -0600
In-Reply-To: <20211129194707.5863-11-michael.christie@oracle.com> (Mike
        Christie's message of "Mon, 29 Nov 2021 13:47:07 -0600")
Message-ID: <87bl1fcaxs.fsf@email.froward.int.ebiederm.org>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Subject: Re: [PATCH V6 10/10] vhost: use user_worker to check RLIMITs
Precedence: bulk

Mike Christie <michael.christie@oracle.com> writes:

> For vhost workers we use the kthread API which inherit's its values from
> and checks against the kthreadd thread. This results in the wrong RLIMITs
> being checked. This patch has us use the user_worker helpers which will
> inherit its values/checks from the thread that owns the device similar to
> if we did a clone in userspace.
>
> Signed-off-by: Mike Christie <michael.christie@oracle.com>
> Acked-by: Michael S. Tsirkin <mst@redhat.com>
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> ---
>  drivers/vhost/vhost.c | 65 +++++++++++++++----------------------------
>  drivers/vhost/vhost.h |  7 ++++-
>  2 files changed, 28 insertions(+), 44 deletions(-)
>
> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> index c9a1f706989c..8cf259d798c0 100644
> --- a/drivers/vhost/vhost.c
> +++ b/drivers/vhost/vhost.c
> @@ -22,7 +22,6 @@
>  #include <linux/slab.h>
>  #include <linux/vmalloc.h>
>  #include <linux/kthread.h>
> -#include <linux/cgroup.h>
>  #include <linux/module.h>
>  #include <linux/sort.h>
>  #include <linux/sched/mm.h>
> @@ -344,17 +343,14 @@ static void vhost_vq_reset(struct vhost_dev *dev,
>  static int vhost_worker(void *data)
>  {
>  	struct vhost_worker *worker = data;
> -	struct vhost_dev *dev = worker->dev;
>  	struct vhost_work *work, *work_next;
>  	struct llist_node *node;
>  
> -	kthread_use_mm(dev->mm);
> -
>  	for (;;) {
>  		/* mb paired w/ kthread_stop */
>  		set_current_state(TASK_INTERRUPTIBLE);
>  
> -		if (kthread_should_stop()) {
> +		if (test_bit(VHOST_WORKER_FLAG_STOP, &worker->flags)) {
>  			__set_current_state(TASK_RUNNING);
>  			break;
>  		}
> @@ -376,8 +372,9 @@ static int vhost_worker(void *data)
>  				schedule();
>  		}
>  	}
> -	kthread_unuse_mm(dev->mm);
> -	return 0;
> +
> +	complete(worker->exit_done);
> +	do_exit(0);

This code worries me.

It has the potential for a caller to do:

	vhost_worker_stop()
        module_put();

Then the exiting work thread tries to do:
	do_exit()

Except the code that calls do_exit has already been removed from the
kernel.  Maybe the vhost code can never be removed from the kernel
but otherwise I expect that is possible.

Eric