Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220221072852.31820-1-mail@anirudhrb.com>
In-Reply-To: <20220221072852.31820-1-mail@anirudhrb.com>
From:   Jason Wang <jasowang@redhat.com>
Date:   Mon, 21 Feb 2022 15:56:29 +0800
Message-ID: <CACGkMEs6HLM3ok29rm4u=Tq2preno_60Z6cvKw2T7=nak2yzkQ@mail.gmail.com>
Subject: Re: [PATCH] vhost: handle zero regions in vhost_set_memory
To:     Anirudh Rayabharam <mail@anirudhrb.com>
Cc:     "Michael S. Tsirkin" <mst@redhat.com>,
        syzbot+0abd373e2e50d704db87@syzkaller.appspotmail.com,
        kvm <kvm@vger.kernel.org>,
        virtualization <virtualization@lists.linux-foundation.org>,
        netdev <netdev@vger.kernel.org>,
        linux-kernel <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Feb 21, 2022 at 3:45 PM Anirudh Rayabharam <mail@anirudhrb.com> wrote:
>
> Return early when userspace sends zero regions in the VHOST_SET_MEM_TABLE
> ioctl.
>
> Otherwise, this causes an erroneous entry to be added to the iotlb. This
> entry has a range size of 0 (due to u64 overflow). This then causes
> iotlb_access_ok() to loop indefinitely resulting in a hung thread.
> Syzbot has reported this here:

Interesting, I think iotlb_access_ok() won't be called for memory
table entries, or anything I missed?

(If this is not true, we need a kernel patch as well).

Thanks

>
> https://syzkaller.appspot.com/bug?extid=0abd373e2e50d704db87
>
> Reported-and-tested-by: syzbot+0abd373e2e50d704db87@syzkaller.appspotmail.com
> Signed-off-by: Anirudh Rayabharam <mail@anirudhrb.com>
> ---
>  drivers/vhost/vhost.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> index 59edb5a1ffe2..821aba60eac2 100644
> --- a/drivers/vhost/vhost.c
> +++ b/drivers/vhost/vhost.c
> @@ -1428,6 +1428,8 @@ static long vhost_set_memory(struct vhost_dev *d, struct vhost_memory __user *m)
>                 return -EFAULT;
>         if (mem.padding)
>                 return -EOPNOTSUPP;
> +       if (mem.nregions == 0)
> +               return 0;
>         if (mem.nregions > max_mem_regions)
>                 return -E2BIG;
>         newmem = kvzalloc(struct_size(newmem, regions, mem.nregions),
> --
> 2.35.1
>