* [PATCH] virtio_config: add device ready status check
@ 2024-03-15 10:31 gavin.liu
2024-03-15 11:07 ` Michael S. Tsirkin
2024-03-18 4:21 ` Jason Wang
0 siblings, 2 replies; 3+ messages in thread
From: gavin.liu @ 2024-03-15 10:31 UTC (permalink / raw)
To: jasowang, mst, xuanzhuo, pbonzini
Cc: stefanha, virtualization, angus.chen, yuxue.liu
From: Yuxue Liu <yuxue.liu@jaguarmicro.com>
Under the VDPA framework, when the VDPA driver of a device sets the status
to DRIVER OK,success is not guaranteed. The virtio driver should provide
some protection or notification.
Signed-off-by: Yuxue Liu <yuxue.liu@jaguarmicro.com>
---
drivers/block/virtio_blk.c | 4 +++-
drivers/net/virtio_net.c | 7 ++++++-
include/linux/virtio_config.h | 14 ++++++++++++++
3 files changed, 23 insertions(+), 2 deletions(-)
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 42dea7601d87..14cabf662705 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -1535,7 +1535,9 @@ static int virtblk_probe(struct virtio_device *vdev)
set_disk_ro(vblk->disk, 1);
virtblk_update_capacity(vblk, false);
- virtio_device_ready(vdev);
+ err = virtio_device_ready_safe(vdev);
+ if (err)
+ goto out_cleanup_disk;
/*
* All steps that follow use the VQs therefore they need to be
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index d7ce4a1011ea..0431fc9095ae 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -4808,7 +4808,12 @@ static int virtnet_probe(struct virtio_device *vdev)
goto free_failover;
}
- virtio_device_ready(vdev);
+ err = virtio_device_ready_safe(vdev);
+ if (err) {
+ pr_debug("virtio_net: setting device ready failed\n");
+ rtnl_unlock();
+ goto free_unregister_netdev;
+ }
_virtnet_set_queues(vi, vi->curr_queue_pairs);
diff --git a/include/linux/virtio_config.h b/include/linux/virtio_config.h
index da9b271b54db..de4e864952ef 100644
--- a/include/linux/virtio_config.h
+++ b/include/linux/virtio_config.h
@@ -301,6 +301,20 @@ void virtio_device_ready(struct virtio_device *dev)
dev->config->set_status(dev, status | VIRTIO_CONFIG_S_DRIVER_OK);
}
+static inline
+int virtio_device_ready_safe(struct virtio_device *dev)
+{
+ u8 status;
+
+ virtio_device_ready(dev);
+ status = dev->config->get_status(dev);
+ if (!(status & VIRTIO_CONFIG_S_DRIVER_OK)) {
+ WARN_ON(1);
+ return -EINVAL;
+ }
+ return 0;
+}
+
static inline
const char *virtio_bus_name(struct virtio_device *vdev)
{
--
2.43.0
^ permalink raw reply related [flat|nested] 3+ messages in thread* Re: [PATCH] virtio_config: add device ready status check
2024-03-15 10:31 [PATCH] virtio_config: add device ready status check gavin.liu
@ 2024-03-15 11:07 ` Michael S. Tsirkin
2024-03-18 4:21 ` Jason Wang
1 sibling, 0 replies; 3+ messages in thread
From: Michael S. Tsirkin @ 2024-03-15 11:07 UTC (permalink / raw)
To: gavin.liu
Cc: jasowang, xuanzhuo, pbonzini, stefanha, virtualization,
angus.chen, yuxue.liu
On Fri, Mar 15, 2024 at 06:31:25PM +0800, gavin.liu wrote:
> From: Yuxue Liu <yuxue.liu@jaguarmicro.com>
>
> Under the VDPA framework, when the VDPA driver of a device sets the status
> to DRIVER OK,success is not guaranteed. The virtio driver should provide
> some protection or notification.
>
> Signed-off-by: Yuxue Liu <yuxue.liu@jaguarmicro.com>
> ---
> drivers/block/virtio_blk.c | 4 +++-
> drivers/net/virtio_net.c | 7 ++++++-
> include/linux/virtio_config.h | 14 ++++++++++++++
> 3 files changed, 23 insertions(+), 2 deletions(-)
And then what we are expected to change all drivers for this hack to
work? Suggest you instead have vdpa set broken flag and be done with it.
> diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
> index 42dea7601d87..14cabf662705 100644
> --- a/drivers/block/virtio_blk.c
> +++ b/drivers/block/virtio_blk.c
> @@ -1535,7 +1535,9 @@ static int virtblk_probe(struct virtio_device *vdev)
> set_disk_ro(vblk->disk, 1);
>
> virtblk_update_capacity(vblk, false);
> - virtio_device_ready(vdev);
> + err = virtio_device_ready_safe(vdev);
> + if (err)
> + goto out_cleanup_disk;
>
> /*
> * All steps that follow use the VQs therefore they need to be
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index d7ce4a1011ea..0431fc9095ae 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -4808,7 +4808,12 @@ static int virtnet_probe(struct virtio_device *vdev)
> goto free_failover;
> }
>
> - virtio_device_ready(vdev);
> + err = virtio_device_ready_safe(vdev);
> + if (err) {
> + pr_debug("virtio_net: setting device ready failed\n");
> + rtnl_unlock();
> + goto free_unregister_netdev;
> + }
>
> _virtnet_set_queues(vi, vi->curr_queue_pairs);
>
> diff --git a/include/linux/virtio_config.h b/include/linux/virtio_config.h
> index da9b271b54db..de4e864952ef 100644
> --- a/include/linux/virtio_config.h
> +++ b/include/linux/virtio_config.h
> @@ -301,6 +301,20 @@ void virtio_device_ready(struct virtio_device *dev)
> dev->config->set_status(dev, status | VIRTIO_CONFIG_S_DRIVER_OK);
> }
>
> +static inline
> +int virtio_device_ready_safe(struct virtio_device *dev)
> +{
> + u8 status;
> +
> + virtio_device_ready(dev);
> + status = dev->config->get_status(dev);
> + if (!(status & VIRTIO_CONFIG_S_DRIVER_OK)) {
> + WARN_ON(1);
> + return -EINVAL;
> + }
> + return 0;
> +}
> +
> static inline
> const char *virtio_bus_name(struct virtio_device *vdev)
> {
> --
> 2.43.0
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] virtio_config: add device ready status check
2024-03-15 10:31 [PATCH] virtio_config: add device ready status check gavin.liu
2024-03-15 11:07 ` Michael S. Tsirkin
@ 2024-03-18 4:21 ` Jason Wang
1 sibling, 0 replies; 3+ messages in thread
From: Jason Wang @ 2024-03-18 4:21 UTC (permalink / raw)
To: gavin.liu
Cc: mst, xuanzhuo, pbonzini, stefanha, virtualization, angus.chen,
yuxue.liu
On Fri, Mar 15, 2024 at 6:31 PM gavin.liu <gavin.liu@jaguarmicro.com> wrote:
>
> From: Yuxue Liu <yuxue.liu@jaguarmicro.com>
>
> Under the VDPA framework, when the VDPA driver of a device sets the status
> to DRIVER OK,success is not guaranteed. The virtio driver should provide
> some protection or notification.
>
> Signed-off-by: Yuxue Liu <yuxue.liu@jaguarmicro.com>
> ---
> drivers/block/virtio_blk.c | 4 +++-
> drivers/net/virtio_net.c | 7 ++++++-
> include/linux/virtio_config.h | 14 ++++++++++++++
> 3 files changed, 23 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
> index 42dea7601d87..14cabf662705 100644
> --- a/drivers/block/virtio_blk.c
> +++ b/drivers/block/virtio_blk.c
> @@ -1535,7 +1535,9 @@ static int virtblk_probe(struct virtio_device *vdev)
> set_disk_ro(vblk->disk, 1);
>
> virtblk_update_capacity(vblk, false);
> - virtio_device_ready(vdev);
> + err = virtio_device_ready_safe(vdev);
> + if (err)
> + goto out_cleanup_disk;
>
> /*
> * All steps that follow use the VQs therefore they need to be
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index d7ce4a1011ea..0431fc9095ae 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -4808,7 +4808,12 @@ static int virtnet_probe(struct virtio_device *vdev)
> goto free_failover;
> }
>
> - virtio_device_ready(vdev);
> + err = virtio_device_ready_safe(vdev);
> + if (err) {
> + pr_debug("virtio_net: setting device ready failed\n");
> + rtnl_unlock();
> + goto free_unregister_netdev;
> + }
>
> _virtnet_set_queues(vi, vi->curr_queue_pairs);
>
> diff --git a/include/linux/virtio_config.h b/include/linux/virtio_config.h
> index da9b271b54db..de4e864952ef 100644
> --- a/include/linux/virtio_config.h
> +++ b/include/linux/virtio_config.h
> @@ -301,6 +301,20 @@ void virtio_device_ready(struct virtio_device *dev)
> dev->config->set_status(dev, status | VIRTIO_CONFIG_S_DRIVER_OK);
> }
>
> +static inline
> +int virtio_device_ready_safe(struct virtio_device *dev)
> +{
> + u8 status;
> +
> + virtio_device_ready(dev);
> + status = dev->config->get_status(dev);
> + if (!(status & VIRTIO_CONFIG_S_DRIVER_OK)) {
> + WARN_ON(1);
> + return -EINVAL;
The seems to be a violation of the virtio spec:
Section 3.1.1 Driver Requirements: Device Initialization said
"""
Set the DRIVER_OK status bit. At this point the device is “live”.
"""
Or why did you need to read the status back and check?
Thanks
> + }
> + return 0;
> +}
> +
> static inline
> const char *virtio_bus_name(struct virtio_device *vdev)
> {
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-03-18 4:21 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-03-15 10:31 [PATCH] virtio_config: add device ready status check gavin.liu
2024-03-15 11:07 ` Michael S. Tsirkin
2024-03-18 4:21 ` Jason Wang
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).