From: Jason Wang <jasowang@redhat.com>
To: Greg Kurz <groug@kaod.org>, "Michael S. Tsirkin" <mst@redhat.com>
Cc: kvm@vger.kernel.org, virtualization@lists.linux-foundation.org,
netdev@vger.kernel.org, qemu-devel@nongnu.org,
Laurent Vivier <laurent@vivier.eu>,
David Gibson <david@gibson.dropbear.id.au>
Subject: Re: [PATCH v2 1/2] vhost: Don't call access_ok() when using IOTLB
Date: Sat, 3 Oct 2020 09:51:33 +0800 [thread overview]
Message-ID: <21349052-fefc-4437-4233-f803caceeb38@redhat.com> (raw)
In-Reply-To: <160139703153.162128.16860679176471296230.stgit@bahia.lan>
On 2020/9/30 上午12:30, Greg Kurz wrote:
> When the IOTLB device is enabled, the vring addresses we get
> from userspace are GIOVAs. It is thus wrong to pass them down
> to access_ok() which only takes HVAs.
>
> Access validation is done at prefetch time with IOTLB. Teach
> vq_access_ok() about that by moving the (vq->iotlb) check
> from vhost_vq_access_ok() to vq_access_ok(). This prevents
> vhost_vring_set_addr() to fail when verifying the accesses.
> No behavior change for vhost_vq_access_ok().
>
> BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1883084
> Fixes: 6b1e6cc7855b ("vhost: new device IOTLB API")
> Cc: jasowang@redhat.com
> CC: stable@vger.kernel.org # 4.14+
> Signed-off-by: Greg Kurz <groug@kaod.org>
> ---
> drivers/vhost/vhost.c | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> index b45519ca66a7..c3b49975dc28 100644
> --- a/drivers/vhost/vhost.c
> +++ b/drivers/vhost/vhost.c
> @@ -1290,6 +1290,11 @@ static bool vq_access_ok(struct vhost_virtqueue *vq, unsigned int num,
> vring_used_t __user *used)
>
> {
> + /* If an IOTLB device is present, the vring addresses are
> + * GIOVAs. Access validation occurs at prefetch time. */
> + if (vq->iotlb)
> + return true;
> +
> return access_ok(desc, vhost_get_desc_size(vq, num)) &&
> access_ok(avail, vhost_get_avail_size(vq, num)) &&
> access_ok(used, vhost_get_used_size(vq, num));
> @@ -1383,10 +1388,6 @@ bool vhost_vq_access_ok(struct vhost_virtqueue *vq)
> if (!vq_log_access_ok(vq, vq->log_base))
> return false;
>
> - /* Access validation occurs at prefetch time with IOTLB */
> - if (vq->iotlb)
> - return true;
> -
> return vq_access_ok(vq, vq->num, vq->desc, vq->avail, vq->used);
> }
> EXPORT_SYMBOL_GPL(vhost_vq_access_ok);
>
Acked-by: Jason Wang <jasowang@redhat.com>
Thanks
next prev parent reply other threads:[~2020-10-03 1:51 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-09-29 16:30 [PATCH v2 0/2] vhost: Skip access checks on GIOVAs Greg Kurz
2020-09-29 16:30 ` [PATCH v2 1/2] vhost: Don't call access_ok() when using IOTLB Greg Kurz
2020-10-03 1:51 ` Jason Wang [this message]
2020-09-29 16:30 ` [PATCH v2 2/2] vhost: Don't call log_access_ok() " Greg Kurz
2020-10-03 1:58 ` Jason Wang
2020-10-03 8:38 ` Greg Kurz
2020-10-01 12:46 ` [PATCH v2 0/2] vhost: Skip access checks on GIOVAs Michael S. Tsirkin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=21349052-fefc-4437-4233-f803caceeb38@redhat.com \
--to=jasowang@redhat.com \
--cc=david@gibson.dropbear.id.au \
--cc=groug@kaod.org \
--cc=kvm@vger.kernel.org \
--cc=laurent@vivier.eu \
--cc=mst@redhat.com \
--cc=netdev@vger.kernel.org \
--cc=qemu-devel@nongnu.org \
--cc=virtualization@lists.linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).