Re: [PATCH RFC don't apply] vdpa_sim: endian-ness for config space

2020-07-16 Thread Jason Wang 



On 2020/7/16 下午1:42, Michael S. Tsirkin wrote:

On Wed, Jul 15, 2020 at 10:02:32PM +0800, Jason Wang wrote:

On 2020/7/15 下午9:58, Michael S. Tsirkin wrote:

VDPA sim stores config space as native endian, but that
is wrong: modern guests expect LE.
I coded up the following to fix it up, but it is wrong too:
vdpasim_create is called before guest features are known.

So what should we do? New ioctl to specify the interface used?
More ideas?

Signed-off-by: Michael S. Tsirkin 


Can we do the endian conversion in set_config/get_config()?

Thanks

That is problematic at least from static checking point of view.



I don't get here. I thought it's just leXX_to_cpu() conversion in 
set_config()/get_config() as what is done in MMIO's vm_set/vm_get.




It would be reasonable to do it in vdpasim_set_features, except
legacy guests might not set features at all.
So my proposal is:
- set config in vdpasim_set_features



Just to make sure I understand here. What's the meaning of "set config" 
do you mean setup the config space? If yes, and if there's a device 
config space, we still need to do the endian conversion?




- document that this is where devices should initialize config
- vdpa core will maintain a "features set" flag, if get/set config
   is called without set features, core will call set features
   automatically with 0 value.



Ok, I think you meant checking FEATURE_OK via get_status in vDPA core?

Thanks





Thoughts?




---
   drivers/vdpa/vdpa_sim/vdpa_sim.c | 22 --
   1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c b/drivers/vdpa/vdpa_sim/vdpa_sim.c
index a9bc5e0fb353..cc754ae0ec15 100644
--- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
+++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
@@ -24,6 +24,7 @@
   #include 
   #include 
   #include 
+#include 
   #include 
   #include 
   #include 
@@ -72,6 +73,23 @@ struct vdpasim {
u64 features;
   };
+/* TODO: cross-endian support */
+static inline bool vdpasim_is_little_endian(struct vdpasim *vdpasim)
+{
+   return virtio_legacy_is_little_endian() ||
+   (vdpasim->features & (1ULL << VIRTIO_F_VERSION_1));
+}
+
+static inline u16 vdpasim16_to_cpu(struct vdpasim *vdpasim, __virtio16 val)
+{
+   return __virtio16_to_cpu(vdpasim_is_little_endian(vdpasim), val);
+}
+
+static inline __virtio16 cpu_to_vdpasim16(struct vdpasim *vdpasim, u16 val)
+{
+   return __cpu_to_virtio16(vdpasim_is_little_endian(vdpasim), val);
+}
+
   static struct vdpasim *vdpasim_dev;
   static struct vdpasim *vdpa_to_sim(struct vdpa_device *vdpa)
@@ -332,8 +350,8 @@ static struct vdpasim *vdpasim_create(void)
goto err_iommu;
config = >config;
-   config->mtu = 1500;
-   config->status = VIRTIO_NET_S_LINK_UP;
+   config->mtu = cpu_to_vdpasim16(vdpasim, 1500);
+   config->status = cpu_to_vdpasim16(vdpasim, VIRTIO_NET_S_LINK_UP);
eth_random_addr(config->mac);
vringh_set_iotlb(>vqs[0].vring, vdpasim->iommu);

Re: [PATCH RFC don't apply] vdpa_sim: endian-ness for config space

2020-07-15 Thread Michael S. Tsirkin 
On Wed, Jul 15, 2020 at 10:02:32PM +0800, Jason Wang wrote:
> 
> On 2020/7/15 下午9:58, Michael S. Tsirkin wrote:
> > VDPA sim stores config space as native endian, but that
> > is wrong: modern guests expect LE.
> > I coded up the following to fix it up, but it is wrong too:
> > vdpasim_create is called before guest features are known.
> > 
> > So what should we do? New ioctl to specify the interface used?
> > More ideas?
> > 
> > Signed-off-by: Michael S. Tsirkin 
> 
> 
> Can we do the endian conversion in set_config/get_config()?
> 
> Thanks

That is problematic at least from static checking point of view.
It would be reasonable to do it in vdpasim_set_features, except
legacy guests might not set features at all.
So my proposal is:
- set config in vdpasim_set_features
- document that this is where devices should initialize config
- vdpa core will maintain a "features set" flag, if get/set config
  is called without set features, core will call set features
  automatically with 0 value.

Thoughts?


> 
> > 
> > 
> > ---
> >   drivers/vdpa/vdpa_sim/vdpa_sim.c | 22 --
> >   1 file changed, 20 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c 
> > b/drivers/vdpa/vdpa_sim/vdpa_sim.c
> > index a9bc5e0fb353..cc754ae0ec15 100644
> > --- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
> > +++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
> > @@ -24,6 +24,7 @@
> >   #include 
> >   #include 
> >   #include 
> > +#include 
> >   #include 
> >   #include 
> >   #include 
> > @@ -72,6 +73,23 @@ struct vdpasim {
> > u64 features;
> >   };
> > +/* TODO: cross-endian support */
> > +static inline bool vdpasim_is_little_endian(struct vdpasim *vdpasim)
> > +{
> > +   return virtio_legacy_is_little_endian() ||
> > +   (vdpasim->features & (1ULL << VIRTIO_F_VERSION_1));
> > +}
> > +
> > +static inline u16 vdpasim16_to_cpu(struct vdpasim *vdpasim, __virtio16 val)
> > +{
> > +   return __virtio16_to_cpu(vdpasim_is_little_endian(vdpasim), val);
> > +}
> > +
> > +static inline __virtio16 cpu_to_vdpasim16(struct vdpasim *vdpasim, u16 val)
> > +{
> > +   return __cpu_to_virtio16(vdpasim_is_little_endian(vdpasim), val);
> > +}
> > +
> >   static struct vdpasim *vdpasim_dev;
> >   static struct vdpasim *vdpa_to_sim(struct vdpa_device *vdpa)
> > @@ -332,8 +350,8 @@ static struct vdpasim *vdpasim_create(void)
> > goto err_iommu;
> > config = >config;
> > -   config->mtu = 1500;
> > -   config->status = VIRTIO_NET_S_LINK_UP;
> > +   config->mtu = cpu_to_vdpasim16(vdpasim, 1500);
> > +   config->status = cpu_to_vdpasim16(vdpasim, VIRTIO_NET_S_LINK_UP);
> > eth_random_addr(config->mac);
> > vringh_set_iotlb(>vqs[0].vring, vdpasim->iommu);

Re: [PATCH RFC don't apply] vdpa_sim: endian-ness for config space

2020-07-15 Thread Jason Wang 



On 2020/7/15 下午9:58, Michael S. Tsirkin wrote:

VDPA sim stores config space as native endian, but that
is wrong: modern guests expect LE.
I coded up the following to fix it up, but it is wrong too:
vdpasim_create is called before guest features are known.

So what should we do? New ioctl to specify the interface used?
More ideas?

Signed-off-by: Michael S. Tsirkin 



Can we do the endian conversion in set_config/get_config()?

Thanks





---
  drivers/vdpa/vdpa_sim/vdpa_sim.c | 22 --
  1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c b/drivers/vdpa/vdpa_sim/vdpa_sim.c
index a9bc5e0fb353..cc754ae0ec15 100644
--- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
+++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
@@ -24,6 +24,7 @@
  #include 
  #include 
  #include 
+#include 
  #include 
  #include 
  #include 
@@ -72,6 +73,23 @@ struct vdpasim {
u64 features;
  };
  
+/* TODO: cross-endian support */

+static inline bool vdpasim_is_little_endian(struct vdpasim *vdpasim)
+{
+   return virtio_legacy_is_little_endian() ||
+   (vdpasim->features & (1ULL << VIRTIO_F_VERSION_1));
+}
+
+static inline u16 vdpasim16_to_cpu(struct vdpasim *vdpasim, __virtio16 val)
+{
+   return __virtio16_to_cpu(vdpasim_is_little_endian(vdpasim), val);
+}
+
+static inline __virtio16 cpu_to_vdpasim16(struct vdpasim *vdpasim, u16 val)
+{
+   return __cpu_to_virtio16(vdpasim_is_little_endian(vdpasim), val);
+}
+
  static struct vdpasim *vdpasim_dev;
  
  static struct vdpasim *vdpa_to_sim(struct vdpa_device *vdpa)

@@ -332,8 +350,8 @@ static struct vdpasim *vdpasim_create(void)
goto err_iommu;
  
  	config = >config;

-   config->mtu = 1500;
-   config->status = VIRTIO_NET_S_LINK_UP;
+   config->mtu = cpu_to_vdpasim16(vdpasim, 1500);
+   config->status = cpu_to_vdpasim16(vdpasim, VIRTIO_NET_S_LINK_UP);
eth_random_addr(config->mac);
  
  	vringh_set_iotlb(>vqs[0].vring, vdpasim->iommu);

[PATCH RFC don't apply] vdpa_sim: endian-ness for config space

2020-07-15 Thread Michael S. Tsirkin 
VDPA sim stores config space as native endian, but that
is wrong: modern guests expect LE.
I coded up the following to fix it up, but it is wrong too:
vdpasim_create is called before guest features are known.

So what should we do? New ioctl to specify the interface used?
More ideas?

Signed-off-by: Michael S. Tsirkin 


---
 drivers/vdpa/vdpa_sim/vdpa_sim.c | 22 --
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c b/drivers/vdpa/vdpa_sim/vdpa_sim.c
index a9bc5e0fb353..cc754ae0ec15 100644
--- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
+++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
@@ -24,6 +24,7 @@
 #include 
 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
@@ -72,6 +73,23 @@ struct vdpasim {
u64 features;
 };
 
+/* TODO: cross-endian support */
+static inline bool vdpasim_is_little_endian(struct vdpasim *vdpasim)
+{
+   return virtio_legacy_is_little_endian() ||
+   (vdpasim->features & (1ULL << VIRTIO_F_VERSION_1));
+}
+
+static inline u16 vdpasim16_to_cpu(struct vdpasim *vdpasim, __virtio16 val)
+{
+   return __virtio16_to_cpu(vdpasim_is_little_endian(vdpasim), val);
+}
+
+static inline __virtio16 cpu_to_vdpasim16(struct vdpasim *vdpasim, u16 val)
+{
+   return __cpu_to_virtio16(vdpasim_is_little_endian(vdpasim), val);
+}
+
 static struct vdpasim *vdpasim_dev;
 
 static struct vdpasim *vdpa_to_sim(struct vdpa_device *vdpa)
@@ -332,8 +350,8 @@ static struct vdpasim *vdpasim_create(void)
goto err_iommu;
 
config = >config;
-   config->mtu = 1500;
-   config->status = VIRTIO_NET_S_LINK_UP;
+   config->mtu = cpu_to_vdpasim16(vdpasim, 1500);
+   config->status = cpu_to_vdpasim16(vdpasim, VIRTIO_NET_S_LINK_UP);
eth_random_addr(config->mac);
 
vringh_set_iotlb(>vqs[0].vring, vdpasim->iommu);
-- 
MST