Re: [Xen-devel] [RFC][v2][PATCH 07/14] xen/passthrough: extend hypercall to support rdm reservation policy

2015-05-25 Thread Julien Grall 



On 25/05/2015 04:09, Chen, Tiejun wrote:

On 2015/5/22 18:33, Julien Grall wrote:

Hi,

On 22/05/2015 10:35, Tiejun Chen wrote:

diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index 0c0ea4a..203c80e 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -499,6 +499,11 @@ struct xen_domctl_assign_device {
  XEN_GUEST_HANDLE_64(char) path; /* path to the device
tree node */
  } dt;
  } u;
+/* IN */
+#define XEN_DOMCTL_DEV_NO_RDM   0
+#define XEN_DOMCTL_DEV_RDM_RELAXED  1
+#define XEN_DOMCTL_DEV_RDM_STRICT   2
+uint32_t  flag;   /* flag of assigned device */


You don't plumb this value for DT neither in the toolstack (see
xc_assign_dt_device) and Xen. Please add a comment saying it's only used


I think we should do this,

@@ -1801,6 +1801,8 @@ int xc_assign_dt_device(

  domctl.u.assign_device.dev = XEN_DOMCTL_DEV_DT;
  domctl.u.assign_device.u.dt.size = size;
+/* DT doesn't own any RDM. */
+domctl.u.assign_device.flag = XEN_DOMCTL_DEV_NO_RDM;
  set_xen_guest_handle(domctl.u.assign_device.u.dt.path, path);

  rc = do_domctl(xch, domctl);


I would be fine with plumbing in drivers/passthrough/device_tree.c and a 
check that the value is not different.


Regards,

--
Julien Grall

___
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC][v2][PATCH 07/14] xen/passthrough: extend hypercall to support rdm reservation policy

2015-05-25 Thread Chen, Tiejun 


On 2015/5/25 18:02, Julien Grall wrote:



On 25/05/2015 04:09, Chen, Tiejun wrote:

On 2015/5/22 18:33, Julien Grall wrote:

Hi,

On 22/05/2015 10:35, Tiejun Chen wrote:

diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index 0c0ea4a..203c80e 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -499,6 +499,11 @@ struct xen_domctl_assign_device {
  XEN_GUEST_HANDLE_64(char) path; /* path to the device
tree node */
  } dt;
  } u;
+/* IN */
+#define XEN_DOMCTL_DEV_NO_RDM   0
+#define XEN_DOMCTL_DEV_RDM_RELAXED  1
+#define XEN_DOMCTL_DEV_RDM_STRICT   2
+uint32_t  flag;   /* flag of assigned device */


You don't plumb this value for DT neither in the toolstack (see
xc_assign_dt_device) and Xen. Please add a comment saying it's only used


I think we should do this,

@@ -1801,6 +1801,8 @@ int xc_assign_dt_device(

  domctl.u.assign_device.dev = XEN_DOMCTL_DEV_DT;
  domctl.u.assign_device.u.dt.size = size;
+/* DT doesn't own any RDM. */
+domctl.u.assign_device.flag = XEN_DOMCTL_DEV_NO_RDM;
  set_xen_guest_handle(domctl.u.assign_device.u.dt.path, path);

  rc = do_domctl(xch, domctl);


I would be fine with plumbing in drivers/passthrough/device_tree.c and a
check that the value is not different.



Are you saying something like this?

@@ -149,6 +149,14 @@ int iommu_do_dt_domctl(struct xen_domctl *domctl, 
struct domain *d,

 if ( domctl-u.assign_device.dev != XEN_DOMCTL_DEV_DT )
 break;

+if ( domctl-u.assign_device.dev == XEN_DOMCTL_DEV_NO_RDM )
+{
+printk(XENLOG_G_ERR XEN_DOMCTL_assign_device: assign \%s\
+to dom%u failed (%d) since we don't support RDM.\n,
+   dt_node_full_name(dev), d-domain_id, ret);
+break;
+}
+
 if ( unlikely(d-is_dying) )
 {
 ret = -EINVAL;

Thanks
Tiejun

___
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC][v2][PATCH 07/14] xen/passthrough: extend hypercall to support rdm reservation policy

2015-05-25 Thread Julien Grall 

Hi,

On 25/05/2015 12:50, Chen, Tiejun wrote:


On 2015/5/25 18:02, Julien Grall wrote:



On 25/05/2015 04:09, Chen, Tiejun wrote:

On 2015/5/22 18:33, Julien Grall wrote:

Hi,

On 22/05/2015 10:35, Tiejun Chen wrote:

diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index 0c0ea4a..203c80e 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -499,6 +499,11 @@ struct xen_domctl_assign_device {
  XEN_GUEST_HANDLE_64(char) path; /* path to the device
tree node */
  } dt;
  } u;
+/* IN */
+#define XEN_DOMCTL_DEV_NO_RDM   0
+#define XEN_DOMCTL_DEV_RDM_RELAXED  1
+#define XEN_DOMCTL_DEV_RDM_STRICT   2
+uint32_t  flag;   /* flag of assigned device */


You don't plumb this value for DT neither in the toolstack (see
xc_assign_dt_device) and Xen. Please add a comment saying it's only
used


I think we should do this,

@@ -1801,6 +1801,8 @@ int xc_assign_dt_device(

  domctl.u.assign_device.dev = XEN_DOMCTL_DEV_DT;
  domctl.u.assign_device.u.dt.size = size;
+/* DT doesn't own any RDM. */
+domctl.u.assign_device.flag = XEN_DOMCTL_DEV_NO_RDM;
  set_xen_guest_handle(domctl.u.assign_device.u.dt.path, path);

  rc = do_domctl(xch, domctl);


I would be fine with plumbing in drivers/passthrough/device_tree.c and a
check that the value is not different.



Are you saying something like this?

@@ -149,6 +149,14 @@ int iommu_do_dt_domctl(struct xen_domctl *domctl,
struct domain *d,
  if ( domctl-u.assign_device.dev != XEN_DOMCTL_DEV_DT )
  break;

+if ( domctl-u.assign_device.dev == XEN_DOMCTL_DEV_NO_RDM )


wrong field here. Other than that it looks good to me.


+{
+printk(XENLOG_G_ERR XEN_DOMCTL_assign_device: assign \%s\
+to dom%u failed (%d) since we don't support RDM.\n,
+   dt_node_full_name(dev), d-domain_id, ret);
+break;
+}
+
  if ( unlikely(d-is_dying) )
  {
  ret = -EINVAL;


Regards,

--
Julien Grall

___
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC][v2][PATCH 07/14] xen/passthrough: extend hypercall to support rdm reservation policy

2015-05-25 Thread Chen, Tiejun 

On 2015/5/25 19:42, Julien Grall wrote:

Hi,

On 25/05/2015 12:50, Chen, Tiejun wrote:


On 2015/5/25 18:02, Julien Grall wrote:



On 25/05/2015 04:09, Chen, Tiejun wrote:

On 2015/5/22 18:33, Julien Grall wrote:

Hi,

On 22/05/2015 10:35, Tiejun Chen wrote:

diff --git a/xen/include/public/domctl.h
b/xen/include/public/domctl.h
index 0c0ea4a..203c80e 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -499,6 +499,11 @@ struct xen_domctl_assign_device {
  XEN_GUEST_HANDLE_64(char) path; /* path to the device
tree node */
  } dt;
  } u;
+/* IN */
+#define XEN_DOMCTL_DEV_NO_RDM   0
+#define XEN_DOMCTL_DEV_RDM_RELAXED  1
+#define XEN_DOMCTL_DEV_RDM_STRICT   2
+uint32_t  flag;   /* flag of assigned device */


You don't plumb this value for DT neither in the toolstack (see
xc_assign_dt_device) and Xen. Please add a comment saying it's only
used


I think we should do this,

@@ -1801,6 +1801,8 @@ int xc_assign_dt_device(

  domctl.u.assign_device.dev = XEN_DOMCTL_DEV_DT;
  domctl.u.assign_device.u.dt.size = size;
+/* DT doesn't own any RDM. */
+domctl.u.assign_device.flag = XEN_DOMCTL_DEV_NO_RDM;
  set_xen_guest_handle(domctl.u.assign_device.u.dt.path, path);

  rc = do_domctl(xch, domctl);


I would be fine with plumbing in drivers/passthrough/device_tree.c and a
check that the value is not different.



Are you saying something like this?

@@ -149,6 +149,14 @@ int iommu_do_dt_domctl(struct xen_domctl *domctl,
struct domain *d,
  if ( domctl-u.assign_device.dev != XEN_DOMCTL_DEV_DT )
  break;

+if ( domctl-u.assign_device.dev == XEN_DOMCTL_DEV_NO_RDM )


wrong field here. Other than that it looks good to me.


Sorry for this typo, s/.dev/.flag.

Thanks
Tiejun




+{
+printk(XENLOG_G_ERR XEN_DOMCTL_assign_device: assign
\%s\
+to dom%u failed (%d) since we don't support
RDM.\n,
+   dt_node_full_name(dev), d-domain_id, ret);
+break;
+}
+
  if ( unlikely(d-is_dying) )
  {
  ret = -EINVAL;


Regards,



___
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC][v2][PATCH 07/14] xen/passthrough: extend hypercall to support rdm reservation policy

2015-05-24 Thread Chen, Tiejun 

On 2015/5/22 18:33, Julien Grall wrote:

Hi,

On 22/05/2015 10:35, Tiejun Chen wrote:

diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index 0c0ea4a..203c80e 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -499,6 +499,11 @@ struct xen_domctl_assign_device {
  XEN_GUEST_HANDLE_64(char) path; /* path to the device
tree node */
  } dt;
  } u;
+/* IN */
+#define XEN_DOMCTL_DEV_NO_RDM   0
+#define XEN_DOMCTL_DEV_RDM_RELAXED  1
+#define XEN_DOMCTL_DEV_RDM_STRICT   2
+uint32_t  flag;   /* flag of assigned device */


You don't plumb this value for DT neither in the toolstack (see
xc_assign_dt_device) and Xen. Please add a comment saying it's only used


I think we should do this,

@@ -1801,6 +1801,8 @@ int xc_assign_dt_device(

 domctl.u.assign_device.dev = XEN_DOMCTL_DEV_DT;
 domctl.u.assign_device.u.dt.size = size;
+/* DT doesn't own any RDM. */
+domctl.u.assign_device.flag = XEN_DOMCTL_DEV_NO_RDM;
 set_xen_guest_handle(domctl.u.assign_device.u.dt.path, path);

 rc = do_domctl(xch, domctl);

Thanks
Tiejun


by PCI and/or the value should always be XEN_DOMCTL_DEV_NO_RDM for DT.

Regards,



___
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Xen-devel] [RFC][v2][PATCH 07/14] xen/passthrough: extend hypercall to support rdm reservation policy

2015-05-22 Thread Julien Grall 

Hi,

On 22/05/2015 10:35, Tiejun Chen wrote:

diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index 0c0ea4a..203c80e 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -499,6 +499,11 @@ struct xen_domctl_assign_device {
  XEN_GUEST_HANDLE_64(char) path; /* path to the device tree node */
  } dt;
  } u;
+/* IN */
+#define XEN_DOMCTL_DEV_NO_RDM   0
+#define XEN_DOMCTL_DEV_RDM_RELAXED  1
+#define XEN_DOMCTL_DEV_RDM_STRICT   2
+uint32_t  flag;   /* flag of assigned device */


You don't plumb this value for DT neither in the toolstack (see 
xc_assign_dt_device) and Xen. Please add a comment saying it's only used 
by PCI and/or the value should always be XEN_DOMCTL_DEV_NO_RDM for DT.


Regards,

--
--
Julien Grall

___
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel