On Tue, Oct 10, 2023 at 01:58:44PM -0300, Jason Gunthorpe wrote:
> On Thu, Sep 21, 2023 at 12:51:22AM -0700, Yi Liu wrote:
> > diff --git a/include/linux/iommu.h b/include/linux/iommu.h
> > index 660dc1931dc9..12e12e5563e6 100644
> > --- a/include/linux/iommu.h
> > +++ b/include/linux/iommu.h
> > @@ -14,6 +14,7 @@
> > #include <linux/err.h>
> > #include <linux/of.h>
> > #include <uapi/linux/iommu.h>
> > +#include <uapi/linux/iommufd.h>
>
> Oh we should definately avoid doing that!
>
> Maybe this is a good moment to start a new header file exclusively for
> iommu drivers and core subsystem to include?
>
> include/linux/iommu-driver.h
>
> ?
>
> Put iommu_copy_user_data() and struct iommu_user_data in there
>
> Avoid this include in this file.
By looking closer, it seems that we included the uapi header for:
+ struct iommu_domain *(*domain_alloc_user)(struct device *dev, u32 flags,
+ enum iommu_hwpt_data_type data_type,
+ struct iommu_domain *parent,
+ const struct iommu_user_data *user_data);
So we could drop the include, and instead add this next to structs:
+enum iommu_hwpt_data_type;
Then it's not that necessary to have a new header? We could mark a
section of "driver exclusively functions" in iommu.h, I think.
> > #define IOMMU_READ (1 << 0)
> > #define IOMMU_WRITE (1 << 1)
> > @@ -227,6 +228,41 @@ struct iommu_iotlb_gather {
> > bool queued;
> > };
> >
> > +/**
> > + * struct iommu_user_data - iommu driver specific user space data info
> > + * @uptr: Pointer to the user buffer for copy_from_user()
> > + * @len: The length of the user buffer in bytes
> > + *
> > + * A user space data is an uAPI that is defined in include/uapi/linux/iommufd.h
> > + * Both @uptr and @len should be just copied from an iommufd core uAPI structure
> > + */
> > +struct iommu_user_data {
> > + void __user *uptr;
> > + size_t len;
> > +};
>
> Put the "hwpt_type" in here and just call it type
Ah, then domain_alloc_user can omit the enum iommu_hwpt_data_type.
Thanks!
Nic
