linux-kernel - Re: [PATCH] vp_vdpa: harden the logic of set status

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 11 Nov 2022 16:14:59 +0100
From:   Stefano Garzarella <sgarzare@...hat.com>
To:     "Longpeng(Mike)" <longpeng2@...wei.com>
Cc:     stefanha@...hat.com, mst@...hat.com, jasowang@...hat.com,
        virtualization@...ts.linux-foundation.org, arei.gonglei@...wei.com,
        yechuan@...wei.com, huangzhichao@...wei.com,
        linux-kernel@...r.kernel.org, xiehong@...wei.com
Subject: Re: [PATCH] vp_vdpa: harden the logic of set status

On Fri, Nov 11, 2022 at 10:55:05PM +0800, Longpeng(Mike) wrote:
>From: Longpeng <longpeng2@...wei.com>
>
>1. We should not set status to 0 when invoking vp_vdpa_set_status().
>
>2. The driver MUST wait for a read of device_status to return 0 before
>   reinitializing the device.
>
>Signed-off-by: Longpeng <longpeng2@...wei.com>
>---
> drivers/vdpa/virtio_pci/vp_vdpa.c | 11 ++++++++++-
> 1 file changed, 10 insertions(+), 1 deletion(-)
>
>diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
>index d448db0c4de3..d35fac5cde11 100644
>--- a/drivers/vdpa/virtio_pci/vp_vdpa.c
>+++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
>@@ -212,8 +212,12 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
> {
> 	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
> 	struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
>-	u8 s = vp_vdpa_get_status(vdpa);

Is this change really needed?

>+	u8 s;
>+
>+	/* We should never be setting status to 0. */
>+	BUG_ON(status == 0);

IMHO panicking the kernel seems excessive in this case, please use 
WARN_ON and maybe return earlier.

>
>+	s = vp_vdpa_get_status(vdpa);
> 	if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
> 	    !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
> 		vp_vdpa_request_irq(vp_vdpa);
>@@ -229,6 +233,11 @@ static int vp_vdpa_reset(struct vdpa_device *vdpa)
> 	u8 s = vp_vdpa_get_status(vdpa);
>
> 	vp_modern_set_status(mdev, 0);
>+	/* After writing 0 to device_status, the driver MUST wait for a read of
>+	 * device_status to return 0 before reinitializing the device.
>+	 */
>+	while (vp_modern_get_status(mdev))
>+		msleep(1);

Should we set a limit after which we give up? A malfunctioning device 
could keep us here forever.

Thanks,
Stefano

>
> 	if (s & VIRTIO_CONFIG_S_DRIVER_OK)
> 		vp_vdpa_free_irq(vp_vdpa);
>-- 
>2.23.0
>