diff mbox series

[RFC,V1,07/13] vhost-vdpa: flush workers on suspend

Message ID 1704919215-91319-8-git-send-email-steven.sistare@oracle.com
State New
Headers
Series vdpa live update |

Commit Message

Steven Sistare Jan. 10, 2024, 8:40 p.m. UTC 
  To pass ownership of a live vdpa device to a new process, the user
suspends the device, calls VHOST_NEW_OWNER to change the mm, and calls
VHOST_IOTLB_REMAP to change the user virtual addresses to match the new
mm.  Flush workers in suspend to guarantee that no worker sees the new
mm and old VA in between.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
---
 drivers/vhost/vdpa.c | 4 ++++
 1 file changed, 4 insertions(+)

Comments

Jason Wang Jan. 11, 2024, 3:09 a.m. UTC | #1 
On Thu, Jan 11, 2024 at 4:40 AM Steve Sistare <steven.sistare@oracle.com> wrote:
>
> To pass ownership of a live vdpa device to a new process, the user
> suspends the device, calls VHOST_NEW_OWNER to change the mm, and calls
> VHOST_IOTLB_REMAP to change the user virtual addresses to match the new
> mm.  Flush workers in suspend to guarantee that no worker sees the new
> mm and old VA in between.
>
> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
> ---
>  drivers/vhost/vdpa.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
> index 8fe1562d24af..9673e8e20d11 100644
> --- a/drivers/vhost/vdpa.c
> +++ b/drivers/vhost/vdpa.c
> @@ -591,10 +591,14 @@ static long vhost_vdpa_suspend(struct vhost_vdpa *v)
>  {
>         struct vdpa_device *vdpa = v->vdpa;
>         const struct vdpa_config_ops *ops = vdpa->config;
> +       struct vhost_dev *vdev = &v->vdev;
>
>         if (!ops->suspend)
>                 return -EOPNOTSUPP;
>
> +       if (vdev->use_worker)
> +               vhost_dev_flush(vdev);

It looks to me like it's better to check use_woker in vhost_dev_flush.

Thanks


> +
>         return ops->suspend(vdpa);
>  }
>
> --
> 2.39.3
>
Mike Christie Jan. 11, 2024, 4:17 p.m. UTC | #2 
On 1/10/24 9:09 PM, Jason Wang wrote:
> On Thu, Jan 11, 2024 at 4:40 AM Steve Sistare <steven.sistare@oracle.com> wrote:
>>
>> To pass ownership of a live vdpa device to a new process, the user
>> suspends the device, calls VHOST_NEW_OWNER to change the mm, and calls
>> VHOST_IOTLB_REMAP to change the user virtual addresses to match the new
>> mm.  Flush workers in suspend to guarantee that no worker sees the new
>> mm and old VA in between.
>>
>> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
>> ---
>>  drivers/vhost/vdpa.c | 4 ++++
>>  1 file changed, 4 insertions(+)
>>
>> diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
>> index 8fe1562d24af..9673e8e20d11 100644
>> --- a/drivers/vhost/vdpa.c
>> +++ b/drivers/vhost/vdpa.c
>> @@ -591,10 +591,14 @@ static long vhost_vdpa_suspend(struct vhost_vdpa *v)
>>  {
>>         struct vdpa_device *vdpa = v->vdpa;
>>         const struct vdpa_config_ops *ops = vdpa->config;
>> +       struct vhost_dev *vdev = &v->vdev;
>>
>>         if (!ops->suspend)
>>                 return -EOPNOTSUPP;
>>
>> +       if (vdev->use_worker)
>> +               vhost_dev_flush(vdev);
> 
> It looks to me like it's better to check use_woker in vhost_dev_flush.
> 

You can now just call vhost_dev_flush and it will do the right thing.
The xa_for_each loop will only flush workers if they have been setup,
so for vdpa it will not find/flush anything.
Jason Wang Jan. 12, 2024, 2:28 a.m. UTC | #3 
On Fri, Jan 12, 2024 at 12:18 AM Mike Christie
<michael.christie@oracle.com> wrote:
>
> On 1/10/24 9:09 PM, Jason Wang wrote:
> > On Thu, Jan 11, 2024 at 4:40 AM Steve Sistare <steven.sistare@oracle.com> wrote:
> >>
> >> To pass ownership of a live vdpa device to a new process, the user
> >> suspends the device, calls VHOST_NEW_OWNER to change the mm, and calls
> >> VHOST_IOTLB_REMAP to change the user virtual addresses to match the new
> >> mm.  Flush workers in suspend to guarantee that no worker sees the new
> >> mm and old VA in between.
> >>
> >> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
> >> ---
> >>  drivers/vhost/vdpa.c | 4 ++++
> >>  1 file changed, 4 insertions(+)
> >>
> >> diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
> >> index 8fe1562d24af..9673e8e20d11 100644
> >> --- a/drivers/vhost/vdpa.c
> >> +++ b/drivers/vhost/vdpa.c
> >> @@ -591,10 +591,14 @@ static long vhost_vdpa_suspend(struct vhost_vdpa *v)
> >>  {
> >>         struct vdpa_device *vdpa = v->vdpa;
> >>         const struct vdpa_config_ops *ops = vdpa->config;
> >> +       struct vhost_dev *vdev = &v->vdev;
> >>
> >>         if (!ops->suspend)
> >>                 return -EOPNOTSUPP;
> >>
> >> +       if (vdev->use_worker)
> >> +               vhost_dev_flush(vdev);
> >
> > It looks to me like it's better to check use_woker in vhost_dev_flush.
> >
>
> You can now just call vhost_dev_flush and it will do the right thing.
> The xa_for_each loop will only flush workers if they have been setup,
> so for vdpa it will not find/flush anything.

Right.

Thanks

>
>
>
Steven Sistare Jan. 17, 2024, 8:30 p.m. UTC | #4 
On 1/11/2024 9:28 PM, Jason Wang wrote:
> On Fri, Jan 12, 2024 at 12:18 AM Mike Christie
> <michael.christie@oracle.com> wrote:
>>
>> On 1/10/24 9:09 PM, Jason Wang wrote:
>>> On Thu, Jan 11, 2024 at 4:40 AM Steve Sistare <steven.sistare@oracle.com> wrote:
>>>>
>>>> To pass ownership of a live vdpa device to a new process, the user
>>>> suspends the device, calls VHOST_NEW_OWNER to change the mm, and calls
>>>> VHOST_IOTLB_REMAP to change the user virtual addresses to match the new
>>>> mm.  Flush workers in suspend to guarantee that no worker sees the new
>>>> mm and old VA in between.
>>>>
>>>> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
>>>> ---
>>>>  drivers/vhost/vdpa.c | 4 ++++
>>>>  1 file changed, 4 insertions(+)
>>>>
>>>> diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
>>>> index 8fe1562d24af..9673e8e20d11 100644
>>>> --- a/drivers/vhost/vdpa.c
>>>> +++ b/drivers/vhost/vdpa.c
>>>> @@ -591,10 +591,14 @@ static long vhost_vdpa_suspend(struct vhost_vdpa *v)
>>>>  {
>>>>         struct vdpa_device *vdpa = v->vdpa;
>>>>         const struct vdpa_config_ops *ops = vdpa->config;
>>>> +       struct vhost_dev *vdev = &v->vdev;
>>>>
>>>>         if (!ops->suspend)
>>>>                 return -EOPNOTSUPP;
>>>>
>>>> +       if (vdev->use_worker)
>>>> +               vhost_dev_flush(vdev);
>>>
>>> It looks to me like it's better to check use_woker in vhost_dev_flush.
>>
>> You can now just call vhost_dev_flush and it will do the right thing.
>> The xa_for_each loop will only flush workers if they have been setup,
>> so for vdpa it will not find/flush anything.

Very good, I will drop this patch.

- Steve
diff mbox series

Patch

diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
index 8fe1562d24af..9673e8e20d11 100644
--- a/drivers/vhost/vdpa.c
+++ b/drivers/vhost/vdpa.c
@@ -591,10 +591,14 @@  static long vhost_vdpa_suspend(struct vhost_vdpa *v)
 {
 	struct vdpa_device *vdpa = v->vdpa;
 	const struct vdpa_config_ops *ops = vdpa->config;
+	struct vhost_dev *vdev = &v->vdev;
 
 	if (!ops->suspend)
 		return -EOPNOTSUPP;
 
+	if (vdev->use_worker)
+		vhost_dev_flush(vdev);
+
 	return ops->suspend(vdpa);
 }