[RFC PATCH 8/8]: PVH: privcmd changes

[RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

---
 drivers/xen/privcmd.c |   68 +++++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 66 insertions(+), 2 deletions(-)

diff --git a/drivers/xen/privcmd.c b/drivers/xen/privcmd.c
index ccee0f1..0a240ab 100644
--- a/drivers/xen/privcmd.c
+++ b/drivers/xen/privcmd.c
@@ -33,6 +33,7 @@
 #include <xen/features.h>
 #include <xen/page.h>
 #include <xen/xen-ops.h>
+#include <xen/balloon.h>
 
 #include "privcmd.h"
 
@@ -199,6 +200,10 @@ static long privcmd_ioctl_mmap(void __user *udata)
 	if (!xen_initial_domain())
 		return -EPERM;
 
+	/* PVH: TBD/FIXME. For now we only support privcmd_ioctl_mmap_batch */
+	if (xen_pvh_domain())
+		return -ENOSYS;
+
 	if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
 		return -EFAULT;
 
@@ -251,6 +256,8 @@ struct mmap_batch_state {
 	xen_pfn_t __user *user;
 };
 
+/* PVH dom0: if domU being created is PV, then mfn is mfn(addr on bus). If
+ * it's PVH then mfn is pfn (input to HAP). */
 static int mmap_batch_fn(void *data, void *state)
 {
 	xen_pfn_t *mfnp = data;
@@ -274,6 +281,40 @@ static int mmap_return_errors(void *data, void *state)
 	return put_user(*mfnp, st->user++);
 }
 
+/* Allocate pfns that are then mapped with gmfns from foreign domid. Update
+ * the vma with the page info to use later.
+ * Returns: 0 if success, otherwise -errno
+ */
+static int pvh_privcmd_resv_pfns(struct vm_area_struct *vma, int numpgs)
+{
+	int rc;
+	struct xen_pvh_sav_pfn_info *savp;
+
+	savp = kzalloc(sizeof(struct xen_pvh_sav_pfn_info), GFP_KERNEL);
+	if (savp == NULL)
+		return -ENOMEM;
+
+	savp->sp_paga = kcalloc(numpgs, sizeof(savp->sp_paga[0]), GFP_KERNEL);
+	if (savp->sp_paga == NULL) {
+		kfree(savp);
+		return -ENOMEM;
+	}
+
+	rc = alloc_xenballooned_pages(numpgs, savp->sp_paga, 0);
+	if (rc != 0) {
+		pr_warn("%s Could not alloc %d pfns rc:%d\n", __FUNCTION__,
+			numpgs, rc);
+		kfree(savp->sp_paga);
+		kfree(savp);
+		return -ENOMEM;
+	}
+	savp->sp_num_pgs = numpgs;
+	BUG_ON(vma->vm_private_data);
+	vma->vm_private_data = savp;
+
+	return 0;
+}
+
 static struct vm_operations_struct privcmd_vm_ops;
 
 static long privcmd_ioctl_mmap_batch(void __user *udata)
@@ -315,6 +356,12 @@ static long privcmd_ioctl_mmap_batch(void __user *udata)
 		goto out;
 	}
 
+	if (xen_pvh_domain()) {
+		if ((ret=pvh_privcmd_resv_pfns(vma, m.num))) {
+			up_write(&mm->mmap_sem);
+			goto out;
+		}
+	}
 	state.domain = m.dom;
 	state.vma = vma;
 	state.va = m.addr;
@@ -365,6 +412,19 @@ static long privcmd_ioctl(struct file *file,
 	return ret;
 }
 
+static void privcmd_close(struct vm_area_struct *vma)
+{
+	struct xen_pvh_sav_pfn_info *savp;
+
+	if (!xen_pvh_domain() || ((savp=vma->vm_private_data) == NULL))
+		return;
+
+	while (savp->sp_next_todo--) {
+		xen_pfn_t pfn = page_to_pfn(savp->sp_paga[savp->sp_next_todo]);
+		pvh_rem_xen_p2m(pfn, 1);
+	}
+}
+
 static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 {
 	printk(KERN_DEBUG "privcmd_fault: vma=%p %lx-%lx, pgoff=%lx, uv=%p\n",
@@ -375,13 +435,14 @@ static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 }
 
 static struct vm_operations_struct privcmd_vm_ops = {
+	.close = privcmd_close,
 	.fault = privcmd_fault
 };
 
 static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
 {
-	/* Unsupported for auto-translate guests. */
-	if (xen_feature(XENFEAT_auto_translated_physmap))
+	/* Unsupported for auto-translate guests unless PVH */
+	if (xen_feature(XENFEAT_auto_translated_physmap) && !xen_pvh_domain())
 		return -ENOSYS;
 
 	/* DONTCOPY is essential for Xen because copy_page_range doesn't know
@@ -395,6 +456,9 @@ static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
 
 static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
 {
+	if (xen_pvh_domain())
+		return (vma->vm_private_data == NULL);
+
 	return (xchg(&vma->vm_private_data, (void *)1) == NULL);
 }
 
-- 
1.7.2.3

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Ian Campbell]

On Thu, 2012-08-16 at 02:07 +0100, Mukesh Rathor wrote:
> ---
>  drivers/xen/privcmd.c |   68 +++++++++++++++++++++++++++++++++++++++++++++++-
>  1 files changed, 66 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/xen/privcmd.c b/drivers/xen/privcmd.c
> index ccee0f1..0a240ab 100644
> --- a/drivers/xen/privcmd.c
> +++ b/drivers/xen/privcmd.c
> @@ -33,6 +33,7 @@
>  #include <xen/features.h>
>  #include <xen/page.h>
>  #include <xen/xen-ops.h>
> +#include <xen/balloon.h>
>  
>  #include "privcmd.h"
>  
> @@ -199,6 +200,10 @@ static long privcmd_ioctl_mmap(void __user *udata)
>  	if (!xen_initial_domain())
>  		return -EPERM;
>  
> +	/* PVH: TBD/FIXME. For now we only support privcmd_ioctl_mmap_batch */
> +	if (xen_pvh_domain())
> +		return -ENOSYS;
> +
>  	if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
>  		return -EFAULT;
>  
> @@ -251,6 +256,8 @@ struct mmap_batch_state {
>  	xen_pfn_t __user *user;
>  };
>  
> +/* PVH dom0: if domU being created is PV, then mfn is mfn(addr on bus). If
> + * it's PVH then mfn is pfn (input to HAP). */
>  static int mmap_batch_fn(void *data, void *state)
>  {
>  	xen_pfn_t *mfnp = data;
> @@ -274,6 +281,40 @@ static int mmap_return_errors(void *data, void *state)
>  	return put_user(*mfnp, st->user++);
>  }
>  
> +/* Allocate pfns that are then mapped with gmfns from foreign domid. Update
> + * the vma with the page info to use later.
> + * Returns: 0 if success, otherwise -errno
> + */
> +static int pvh_privcmd_resv_pfns(struct vm_area_struct *vma, int numpgs)
> +{
> +	int rc;
> +	struct xen_pvh_sav_pfn_info *savp;
> +
> +	savp = kzalloc(sizeof(struct xen_pvh_sav_pfn_info), GFP_KERNEL);
> +	if (savp == NULL)
> +		return -ENOMEM;
> +
> +	savp->sp_paga = kcalloc(numpgs, sizeof(savp->sp_paga[0]), GFP_KERNEL);
> +	if (savp->sp_paga == NULL) {
> +		kfree(savp);
> +		return -ENOMEM;
> +	}
> +
> +	rc = alloc_xenballooned_pages(numpgs, savp->sp_paga, 0);
> +	if (rc != 0) {
> +		pr_warn("%s Could not alloc %d pfns rc:%d\n", __FUNCTION__,
> +			numpgs, rc);
> +		kfree(savp->sp_paga);
> +		kfree(savp);
> +		return -ENOMEM;
> +	}
> +	savp->sp_num_pgs = numpgs;
> +	BUG_ON(vma->vm_private_data);
> +	vma->vm_private_data = savp;
> +
> +	return 0;
> +}
> +
>  static struct vm_operations_struct privcmd_vm_ops;
>  
>  static long privcmd_ioctl_mmap_batch(void __user *udata)
> @@ -315,6 +356,12 @@ static long privcmd_ioctl_mmap_batch(void __user *udata)
>  		goto out;
>  	}
>  
> +	if (xen_pvh_domain()) {
> +		if ((ret=pvh_privcmd_resv_pfns(vma, m.num))) {
> +			up_write(&mm->mmap_sem);
> +			goto out;
> +		}
> +	}
>  	state.domain = m.dom;
>  	state.vma = vma;
>  	state.va = m.addr;
> @@ -365,6 +412,19 @@ static long privcmd_ioctl(struct file *file,
>  	return ret;
>  }
>  
> +static void privcmd_close(struct vm_area_struct *vma)
> +{
> +	struct xen_pvh_sav_pfn_info *savp;
> +
> +	if (!xen_pvh_domain() || ((savp=vma->vm_private_data) == NULL))
> +		return;
> +
> +	while (savp->sp_next_todo--) {
> +		xen_pfn_t pfn = page_to_pfn(savp->sp_paga[savp->sp_next_todo]);
> +		pvh_rem_xen_p2m(pfn, 1);
> +	}
> +}
> +
>  static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
>  {
>  	printk(KERN_DEBUG "privcmd_fault: vma=%p %lx-%lx, pgoff=%lx, uv=%p\n",
> @@ -375,13 +435,14 @@ static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
>  }
>  
>  static struct vm_operations_struct privcmd_vm_ops = {
> +	.close = privcmd_close,
>  	.fault = privcmd_fault
>  };
>  
>  static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
>  {
> -	/* Unsupported for auto-translate guests. */
> -	if (xen_feature(XENFEAT_auto_translated_physmap))
> +	/* Unsupported for auto-translate guests unless PVH */
> +	if (xen_feature(XENFEAT_auto_translated_physmap) && !xen_pvh_domain())
>  		return -ENOSYS;
>  
>  	/* DONTCOPY is essential for Xen because copy_page_range doesn't know
> @@ -395,6 +456,9 @@ static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
>  
>  static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
>  {
> +	if (xen_pvh_domain())
> +		return (vma->vm_private_data == NULL);
> +
>  	return (xchg(&vma->vm_private_data, (void *)1) == NULL);

How come this is different for pvh?

Your version doesn't appear to enforce anything, since it doesn't set
it. Oh I see, you set it to some actual useful data in
pvh_privcmd_resv_pfns. I have a feeling you might want to use some sort
of atomic construct for that.

Which I suspect then means you need to be prepared for it to fail.

Can we set vm_private_data => 1 in privcmd_enforce_singleshot_mapping to
"open the transaction" and then up in pvh_privcmd_resv_pfns set it to
the final pointer up in pvh_privcmd_resv_pfns?

That was it would be non-NULL both while the mapping is being
constructed and in use. Might need some checks elsewhere that it is != 1
when you expect a real pointer.

Ian.

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Ian Campbell]

On Thu, 2012-08-16 at 02:07 +0100, Mukesh Rathor wrote:
> ---
>  drivers/xen/privcmd.c |   68 +++++++++++++++++++++++++++++++++++++++++++++++-
>  1 files changed, 66 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/xen/privcmd.c b/drivers/xen/privcmd.c
> index ccee0f1..0a240ab 100644
> --- a/drivers/xen/privcmd.c
> +++ b/drivers/xen/privcmd.c
> @@ -33,6 +33,7 @@
>  #include <xen/features.h>
>  #include <xen/page.h>
>  #include <xen/xen-ops.h>
> +#include <xen/balloon.h>
>  
>  #include "privcmd.h"
>  
> @@ -199,6 +200,10 @@ static long privcmd_ioctl_mmap(void __user *udata)
>  	if (!xen_initial_domain())
>  		return -EPERM;
>  
> +	/* PVH: TBD/FIXME. For now we only support privcmd_ioctl_mmap_batch */
> +	if (xen_pvh_domain())
> +		return -ENOSYS;
> +
>  	if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
>  		return -EFAULT;
>  
> @@ -251,6 +256,8 @@ struct mmap_batch_state {
>  	xen_pfn_t __user *user;
>  };
>  
> +/* PVH dom0: if domU being created is PV, then mfn is mfn(addr on bus). If
> + * it's PVH then mfn is pfn (input to HAP). */
>  static int mmap_batch_fn(void *data, void *state)
>  {
>  	xen_pfn_t *mfnp = data;
> @@ -274,6 +281,40 @@ static int mmap_return_errors(void *data, void *state)
>  	return put_user(*mfnp, st->user++);
>  }
>  
> +/* Allocate pfns that are then mapped with gmfns from foreign domid. Update
> + * the vma with the page info to use later.
> + * Returns: 0 if success, otherwise -errno
> + */
> +static int pvh_privcmd_resv_pfns(struct vm_area_struct *vma, int numpgs)
> +{
> +	int rc;
> +	struct xen_pvh_sav_pfn_info *savp;
> +
> +	savp = kzalloc(sizeof(struct xen_pvh_sav_pfn_info), GFP_KERNEL);
> +	if (savp == NULL)
> +		return -ENOMEM;
> +
> +	savp->sp_paga = kcalloc(numpgs, sizeof(savp->sp_paga[0]), GFP_KERNEL);
> +	if (savp->sp_paga == NULL) {
> +		kfree(savp);
> +		return -ENOMEM;
> +	}
> +
> +	rc = alloc_xenballooned_pages(numpgs, savp->sp_paga, 0);
> +	if (rc != 0) {
> +		pr_warn("%s Could not alloc %d pfns rc:%d\n", __FUNCTION__,
> +			numpgs, rc);
> +		kfree(savp->sp_paga);
> +		kfree(savp);
> +		return -ENOMEM;
> +	}

I've just been building on this patch to make proper mmap foreign
support on ARM and I was looking for the place which freed this, both
the pages back to the balloon and then the array itself. There is code
in privcmd_close which unmaps the P2M, but I can't find the code which
frees things back to the balloon. Have I missed something?

I think we also need to think about the layering / abstraction a bit
here too. Currently on setup the caller allocates the array iff pvh and
stores it in the opaque vma->vm_private, then calls
xen_remap_domain_mfn_range which iff pvh calls pvh_add_to_xen_p2m which
assumes that the caller has seeded the vm_private with the correct
struct. xen_remap_domain_mfn_range also sets up the stage 1 PT mappings.

On teardown the iff pvh is in the caller which calls pvh_rem_xen_p2m
directly (this API is unbalanced with the setup side). The stage 1 PT
mappings are torn down implicitly somewhere else (in generic code, I
think you said).

(BTW, the terminology I'm using is stage 1 == guest OS page tables,
stage 2 == HAP)

I think you can't rely on the implicit teardown here since you need to
unmap before you hand the page back to the balloon. The reason this
doesn't look necessary now is that you don't give the page back.

Also not ordering the stage 1 and stage 2 teardown correctly is
dangerous, depending on the eventual ordering you potentially turn an
erroneous access to a virtual address, which should result in a guest OS
level page fault (and e.g. a seg fault to the offending process) into a
hypervisor shoots the guest due to an unexpected stage 2 fault type
failure, which is somewhat undesirable ;-)

With that in mind I think you do in the end need to add
xen_unmap_domain_mfn_range which does the unmap from both stage 1 and
stage 2 -- that balances out the interface (making pvh_rem_xen_p2m
internal) too, which is nice. This function may turn out to be a nop
on !pvh, but that's ok (although maybe there would be no harm in doing
explicit unmaps, for consistency?).

WRT passing data between interfaces in vma->vm_private, which is pretty
subtle, can we push that whole thing down into
xen_{remap,unmap}_domain_mfn_range too? This would make the requirement
on the caller be simple "never use vm_private", as opposed to now where
the requirement is "sometimes you might have to allocate some stuff and
stick it in here". The downside is that it pushes code which could be
generic down into per-arch stuff, although with suitable generic helper
functions this isn't so bad (whatever happened to
{alloc,free}_empty_pages_and_pagevec from the classic kernels? Those did
exactly what we want here, I think)

Ian.

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

On Tue, 11 Sep 2012 15:10:23 +0100
Ian Campbell <Ian.Campbell@citrix.com> wrote:

> On Thu, 2012-08-16 at 02:07 +0100, Mukesh Rathor wrote:
> > ---
> >  drivers/xen/privcmd.c |   68
> > +	if (rc != 0) {
> > +		pr_warn("%s Could not alloc %d pfns rc:%d\n",
> > __FUNCTION__,
> > +			numpgs, rc);
> > +		kfree(savp->sp_paga);
> > +		kfree(savp);
> > +		return -ENOMEM;
> > +	}
> 
> I've just been building on this patch to make proper mmap foreign
> support on ARM and I was looking for the place which freed this, both
> the pages back to the balloon and then the array itself. There is code
> in privcmd_close which unmaps the P2M, but I can't find the code which
> frees things back to the balloon. Have I missed something?

You are right, I missed the free. Let me revisit it and make some changes.

thanks,
mukesh

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Ian Campbell]

On Wed, 2012-09-12 at 02:32 +0100, Mukesh Rathor wrote:
> On Tue, 11 Sep 2012 15:10:23 +0100
> Ian Campbell <Ian.Campbell@citrix.com> wrote:
> 
> > On Thu, 2012-08-16 at 02:07 +0100, Mukesh Rathor wrote:
> > > ---
> > >  drivers/xen/privcmd.c |   68
> > > +	if (rc != 0) {
> > > +		pr_warn("%s Could not alloc %d pfns rc:%d\n",
> > > __FUNCTION__,
> > > +			numpgs, rc);
> > > +		kfree(savp->sp_paga);
> > > +		kfree(savp);
> > > +		return -ENOMEM;
> > > +	}
> > 
> > I've just been building on this patch to make proper mmap foreign
> > support on ARM and I was looking for the place which freed this, both
> > the pages back to the balloon and then the array itself. There is code
> > in privcmd_close which unmaps the P2M, but I can't find the code which
> > frees things back to the balloon. Have I missed something?
> 
> You are right, I missed the free. Let me revisit it and make some changes.

Thanks.

Any comments on the rest of my mail? We need to agree what the interface
between the generic and the per-arch code is going to look like here.

Ian.

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

On Wed, 12 Sep 2012 08:36:59 +0100
Ian Campbell <Ian.Campbell@citrix.com> wrote:

> On Wed, 2012-09-12 at 02:32 +0100, Mukesh Rathor wrote:
> > On Tue, 11 Sep 2012 15:10:23 +0100
> > Ian Campbell <Ian.Campbell@citrix.com> wrote:
> > 
> > > On Thu, 2012-08-16 at 02:07 +0100, Mukesh Rathor wrote:
> > > > ---
> > > >  drivers/xen/privcmd.c |   68
> > > > +	if (rc != 0) {
> > > > +		pr_warn("%s Could not alloc %d pfns rc:%d\n",
> > > > __FUNCTION__,
> > > > +			numpgs, rc);
> > > > +		kfree(savp->sp_paga);
> > > > +		kfree(savp);
> > > > +		return -ENOMEM;
> > > > +	}
> > > 
> > > I've just been building on this patch to make proper mmap foreign
> > > support on ARM and I was looking for the place which freed this,
> > > both the pages back to the balloon and then the array itself.
> > > There is code in privcmd_close which unmaps the P2M, but I can't
> > > find the code which frees things back to the balloon. Have I
> > > missed something?
> > 
> > You are right, I missed the free. Let me revisit it and make some
> > changes.
> 
> Thanks.
> 
> Any comments on the rest of my mail? We need to agree what the
> interface between the generic and the per-arch code is going to look
> like here.
> 
> Ian.

Right, I am paging it all in the brain right now, as I made the changes
a while ago :). I will attempt to change the code according to
your email, to come up with generic interface. Also I am going to come
up with xen_unmap_domain_mfn_range. Mostly agree with your email. More
soon.

thanks,
Mukesh

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

[-- Attachment #1: Type: text/plain, Size: 3136 bytes --]

On Tue, 11 Sep 2012 15:10:23 +0100
Ian Campbell <Ian.Campbell@citrix.com> wrote:

> I think you can't rely on the implicit teardown here since you need to
> unmap before you hand the page back to the balloon. The reason this
> doesn't look necessary now is that you don't give the page back.
> Also not ordering the stage 1 and stage 2 teardown correctly is
> dangerous, depending on the eventual ordering you potentially turn an
> erroneous access to a virtual address, which should result in a guest
> OS level page fault (and e.g. a seg fault to the offending process)
> into a hypervisor shoots the guest due to an unexpected stage 2 fault
> type failure, which is somewhat undesirable ;-)
> 
> With that in mind I think you do in the end need to add
> xen_unmap_domain_mfn_range which does the unmap from both stage 1 and
> stage 2 -- that balances out the interface (making pvh_rem_xen_p2m
> internal) too, which is nice. This function may turn out to be a nop
> on !pvh, but that's ok (although maybe there would be no harm in doing
> explicit unmaps, for consistency?).
 
Ok, I added xen_unmap_domain_mfn_range(). Take a look. It appears that
by the time privcmd_close() is called, the kernel has already freed 
process resources and lookup_address() returns NULL. Now I am wondering
if the kernel/mmu does anything to the page while shooting the pte
entry. Well the page was orig from balloon, so the cleanup hopefully
leaves it alone.

I had looked for other hooks initially when I did this, but 
vm_operations_struct->close was the only one to pan out.

I can't really move pvh_privcmd_resv_pfns to mmu.c because the 
xen_remap_domain_mfn_range is called one page at a time, and I need
to allocate the array first. I'd have to change it to linked list, worth
it? Or I'd have to move and export it.


> WRT passing data between interfaces in vma->vm_private, which is
> pretty subtle, can we push that whole thing down into
> xen_{remap,unmap}_domain_mfn_range too? This would make the
> requirement on the caller be simple "never use vm_private", as
> opposed to now where the requirement is "sometimes you might have to
> allocate some stuff and stick it in here". The downside is that it
> pushes code which could be generic down into per-arch stuff, although
> with suitable generic helper functions this isn't so bad (whatever
> happened to {alloc,free}_empty_pages_and_pagevec from the classic
> kernels? Those did exactly what we want here, I think)

Well, it has to hang off of vma->vm_private. The alternative would be to
have a hash table by process id or something, again not sure if worth it. 

Take a look at my latest files attached. Now the alloc_balloon and free
are split between privcmd and mmu.c. The alternative would be to call 
xen_unmap_domain_mfn_range one pfn at a time and have it call 
pvh_rem_xen_p2m(), and move free_xenballooned_pages to privcmd. But
that would be same as just changing the name of pvh_rem_xen_p2m to
xen_unmap_domain_mfn_range(). Also, remap and unmap won't be much
symmetric then.

Not sure if there's a lot we could do here to be honest. LMK what you 
think.

thanks,
Mukesh


[-- Attachment #2: privcmd.c --]
[-- Type: text/x-c++src, Size: 10748 bytes --]

/******************************************************************************
 * privcmd.c
 *
 * Interface to privileged domain-0 commands.
 *
 * Copyright (c) 2002-2004, K A Fraser, B Dragovic
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include <linux/uaccess.h>
#include <linux/swap.h>
#include <linux/highmem.h>
#include <linux/pagemap.h>
#include <linux/seq_file.h>
#include <linux/miscdevice.h>

#include <asm/pgalloc.h>
#include <asm/pgtable.h>
#include <asm/tlb.h>
#include <asm/xen/hypervisor.h>
#include <asm/xen/hypercall.h>

#include <xen/xen.h>
#include <xen/privcmd.h>
#include <xen/interface/xen.h>
#include <xen/features.h>
#include <xen/page.h>
#include <xen/xen-ops.h>
#include <xen/balloon.h>

#include "privcmd.h"

MODULE_LICENSE("GPL");

#ifndef HAVE_ARCH_PRIVCMD_MMAP
static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma);
#endif

static long privcmd_ioctl_hypercall(void __user *udata)
{
	struct privcmd_hypercall hypercall;
	long ret;

	if (copy_from_user(&hypercall, udata, sizeof(hypercall)))
		return -EFAULT;

	ret = privcmd_call(hypercall.op,
			   hypercall.arg[0], hypercall.arg[1],
			   hypercall.arg[2], hypercall.arg[3],
			   hypercall.arg[4]);

	return ret;
}

static void free_page_list(struct list_head *pages)
{
	struct page *p, *n;

	list_for_each_entry_safe(p, n, pages, lru)
		__free_page(p);

	INIT_LIST_HEAD(pages);
}

/*
 * Given an array of items in userspace, return a list of pages
 * containing the data.  If copying fails, either because of memory
 * allocation failure or a problem reading user memory, return an
 * error code; its up to the caller to dispose of any partial list.
 */
static int gather_array(struct list_head *pagelist,
			unsigned nelem, size_t size,
			void __user *data)
{
	unsigned pageidx;
	void *pagedata;
	int ret;

	if (size > PAGE_SIZE)
		return 0;

	pageidx = PAGE_SIZE;
	pagedata = NULL;	/* quiet, gcc */
	while (nelem--) {
		if (pageidx > PAGE_SIZE-size) {
			struct page *page = alloc_page(GFP_KERNEL);

			ret = -ENOMEM;
			if (page == NULL)
				goto fail;

			pagedata = page_address(page);

			list_add_tail(&page->lru, pagelist);
			pageidx = 0;
		}

		ret = -EFAULT;
		if (copy_from_user(pagedata + pageidx, data, size))
			goto fail;

		data += size;
		pageidx += size;
	}

	ret = 0;

fail:
	return ret;
}

/*
 * Call function "fn" on each element of the array fragmented
 * over a list of pages.
 */
static int traverse_pages(unsigned nelem, size_t size,
			  struct list_head *pos,
			  int (*fn)(void *data, void *state),
			  void *state)
{
	void *pagedata;
	unsigned pageidx;
	int ret = 0;

	BUG_ON(size > PAGE_SIZE);

	pageidx = PAGE_SIZE;
	pagedata = NULL;	/* hush, gcc */

	while (nelem--) {
		if (pageidx > PAGE_SIZE-size) {
			struct page *page;
			pos = pos->next;
			page = list_entry(pos, struct page, lru);
			pagedata = page_address(page);
			pageidx = 0;
		}

		ret = (*fn)(pagedata + pageidx, state);
		if (ret)
			break;
		pageidx += size;
	}

	return ret;
}

struct mmap_mfn_state {
	unsigned long va;
	struct vm_area_struct *vma;
	domid_t domain;
};

static int mmap_mfn_range(void *data, void *state)
{
	struct privcmd_mmap_entry *msg = data;
	struct mmap_mfn_state *st = state;
	struct vm_area_struct *vma = st->vma;
	int rc;

	/* Do not allow range to wrap the address space. */
	if ((msg->npages > (LONG_MAX >> PAGE_SHIFT)) ||
	    ((unsigned long)(msg->npages << PAGE_SHIFT) >= -st->va))
		return -EINVAL;

	/* Range chunks must be contiguous in va space. */
	if ((msg->va != st->va) ||
	    ((msg->va+(msg->npages<<PAGE_SHIFT)) > vma->vm_end))
		return -EINVAL;

	rc = xen_remap_domain_mfn_range(vma,
					msg->va & PAGE_MASK,
					msg->mfn, msg->npages,
					vma->vm_page_prot,
					st->domain);
	if (rc < 0)
		return rc;

	st->va += msg->npages << PAGE_SHIFT;

	return 0;
}

static long privcmd_ioctl_mmap(void __user *udata)
{
	struct privcmd_mmap mmapcmd;
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	int rc;
	LIST_HEAD(pagelist);
	struct mmap_mfn_state state;

	if (!xen_initial_domain())
		return -EPERM;

	/* PVH: TBD/FIXME. For now we only support privcmd_ioctl_mmap_batch */
	if (xen_pv_domain() && xen_feature(XENFEAT_auto_translated_physmap))
		return -ENOSYS;

	if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
		return -EFAULT;

	rc = gather_array(&pagelist,
			  mmapcmd.num, sizeof(struct privcmd_mmap_entry),
			  mmapcmd.entry);

	if (rc || list_empty(&pagelist))
		goto out;

	down_write(&mm->mmap_sem);

	{
		struct page *page = list_first_entry(&pagelist,
						     struct page, lru);
		struct privcmd_mmap_entry *msg = page_address(page);

		vma = find_vma(mm, msg->va);
		rc = -EINVAL;

		if (!vma || (msg->va != vma->vm_start) ||
		    !privcmd_enforce_singleshot_mapping(vma))
			goto out_up;
	}

	state.va = vma->vm_start;
	state.vma = vma;
	state.domain = mmapcmd.dom;

	rc = traverse_pages(mmapcmd.num, sizeof(struct privcmd_mmap_entry),
			    &pagelist,
			    mmap_mfn_range, &state);


out_up:
	up_write(&mm->mmap_sem);

out:
	free_page_list(&pagelist);

	return rc;
}

struct mmap_batch_state {
	domid_t domain;
	unsigned long va;
	struct vm_area_struct *vma;
	int err;

	xen_pfn_t __user *user;
};

/* PVH dom0 fyi: if domU being created is PV, then mfn is mfn(addr on bus). If 
 * it's PVH then mfn is pfn (input to HAP). */
static int mmap_batch_fn(void *data, void *state)
{
	xen_pfn_t *mfnp = data;
	struct mmap_batch_state *st = state;

	if (xen_remap_domain_mfn_range(st->vma, st->va & PAGE_MASK, *mfnp, 1,
				       st->vma->vm_page_prot, st->domain) < 0) {
		*mfnp |= 0xf0000000U;
		st->err++;
	}
	st->va += PAGE_SIZE;

	return 0;
}

static int mmap_return_errors(void *data, void *state)
{
	xen_pfn_t *mfnp = data;
	struct mmap_batch_state *st = state;

	return put_user(*mfnp, st->user++);
}

/* Allocate pfns that are then mapped with gmfns from foreign domid. Update
 * the vma with the page info to use later.
 * Returns: 0 if success, otherwise -errno
 */ 
static int pvh_privcmd_resv_pfns(struct vm_area_struct *vma, int numpgs)
{
	int rc;
	struct xen_pvh_sav_pfn_info *savp;

	savp = kzalloc(sizeof(struct xen_pvh_sav_pfn_info), GFP_KERNEL);
	if (savp == NULL)
		return -ENOMEM;

	savp->sp_paga = kcalloc(numpgs, sizeof(savp->sp_paga[0]), GFP_KERNEL);
	if (savp->sp_paga == NULL) {
		kfree(savp);
		return -ENOMEM;
	}

	rc = alloc_xenballooned_pages(numpgs, savp->sp_paga, 0);
	if (rc != 0) {
		pr_warn("%s Could not alloc %d pfns rc:%d\n", __FUNCTION__, 
			numpgs, rc);
		kfree(savp->sp_paga);
		kfree(savp);
		return -ENOMEM;
	}
	savp->sp_num_pgs = numpgs;
	BUG_ON(vma->vm_private_data);
	vma->vm_private_data = savp;

	return 0;
}

static struct vm_operations_struct privcmd_vm_ops;

static long privcmd_ioctl_mmap_batch(void __user *udata)
{
	int ret;
	struct privcmd_mmapbatch m;
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	unsigned long nr_pages;
	LIST_HEAD(pagelist);
	struct mmap_batch_state state;

	if (!xen_initial_domain())
		return -EPERM;

	if (copy_from_user(&m, udata, sizeof(m)))
		return -EFAULT;

	nr_pages = m.num;
	if ((m.num <= 0) || (nr_pages > (LONG_MAX >> PAGE_SHIFT)))
		return -EINVAL;

	ret = gather_array(&pagelist, m.num, sizeof(xen_pfn_t),
			   m.arr);

	if (ret || list_empty(&pagelist))
		goto out;

	down_write(&mm->mmap_sem);

	vma = find_vma(mm, m.addr);
	ret = -EINVAL;
	if (!vma ||
	    vma->vm_ops != &privcmd_vm_ops ||
	    (m.addr != vma->vm_start) ||
	    ((m.addr + (nr_pages << PAGE_SHIFT)) != vma->vm_end) ||
	    !privcmd_enforce_singleshot_mapping(vma)) {
		up_write(&mm->mmap_sem);
		goto out;
	}

	if (xen_pv_domain() && xen_feature(XENFEAT_auto_translated_physmap)) {
		if ((ret=pvh_privcmd_resv_pfns(vma, m.num))) {
			up_write(&mm->mmap_sem);
			goto out;
		}
	}
	state.domain = m.dom;
	state.vma = vma;
	state.va = m.addr;
	state.err = 0;

	ret = traverse_pages(m.num, sizeof(xen_pfn_t),
			     &pagelist, mmap_batch_fn, &state);

	up_write(&mm->mmap_sem);

	if (state.err > 0) {
		state.user = m.arr;
		ret = traverse_pages(m.num, sizeof(xen_pfn_t),
			       &pagelist,
			       mmap_return_errors, &state);
	}

out:
	free_page_list(&pagelist);

	return ret;
}

static long privcmd_ioctl(struct file *file,
			  unsigned int cmd, unsigned long data)
{
	int ret = -ENOSYS;
	void __user *udata = (void __user *) data;

	switch (cmd) {
	case IOCTL_PRIVCMD_HYPERCALL:
		ret = privcmd_ioctl_hypercall(udata);
		break;

	case IOCTL_PRIVCMD_MMAP:
		ret = privcmd_ioctl_mmap(udata);
		break;

	case IOCTL_PRIVCMD_MMAPBATCH:
		ret = privcmd_ioctl_mmap_batch(udata);
		break;

	default:
		ret = -EINVAL;
		break;
	}

	return ret;
}

static void privcmd_close(struct vm_area_struct *vma)
{
	struct xen_pvh_sav_pfn_info *savp = vma->vm_private_data;

	if (!xen_pv_domain()  || 
	    !xen_feature(XENFEAT_auto_translated_physmap)  ||
	    !vma->vm_private_data)
		return;

	xen_unmap_domain_mfn_range(vma);
	kfree(savp->sp_paga);
	kfree(savp);
}

static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
	printk(KERN_DEBUG "privcmd_fault: vma=%p %lx-%lx, pgoff=%lx, uv=%p\n",
	       vma, vma->vm_start, vma->vm_end,
	       vmf->pgoff, vmf->virtual_address);

	return VM_FAULT_SIGBUS;
}

static struct vm_operations_struct privcmd_vm_ops = {
	.close = privcmd_close,
	.fault = privcmd_fault
};

static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
{
	/* DONTCOPY is essential for Xen because copy_page_range doesn't know
	 * how to recreate these mappings */
	vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY | VM_PFNMAP;
	vma->vm_ops = &privcmd_vm_ops;
	vma->vm_private_data = NULL;

	return 0;
}

static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
{
	if (xen_pv_domain() && xen_feature(XENFEAT_auto_translated_physmap))
		/* NOTE : make this atomic ********************** */
		return (vma->vm_private_data == NULL);

	return (xchg(&vma->vm_private_data, (void *)1) == NULL);
}

const struct file_operations xen_privcmd_fops = {
	.owner = THIS_MODULE,
	.unlocked_ioctl = privcmd_ioctl,
	.mmap = privcmd_mmap,
};
EXPORT_SYMBOL_GPL(xen_privcmd_fops);

static struct miscdevice privcmd_dev = {
	.minor = MISC_DYNAMIC_MINOR,
	.name = "xen/privcmd",
	.fops = &xen_privcmd_fops,
};

static int __init privcmd_init(void)
{
	int err;

	if (!xen_domain())
		return -ENODEV;

	err = misc_register(&privcmd_dev);
	if (err != 0) {
		printk(KERN_ERR "Could not register Xen privcmd device\n");
		return err;
	}
	return 0;
}

static void __exit privcmd_exit(void)
{
	misc_deregister(&privcmd_dev);
}

module_init(privcmd_init);
module_exit(privcmd_exit);

[-- Attachment #3: mmu.c --]
[-- Type: text/x-c++src, Size: 63760 bytes --]

/*
 * Xen mmu operations
 *
 * This file contains the various mmu fetch and update operations.
 * The most important job they must perform is the mapping between the
 * domain's pfn and the overall machine mfns.
 *
 * Xen allows guests to directly update the pagetable, in a controlled
 * fashion.  In other words, the guest modifies the same pagetable
 * that the CPU actually uses, which eliminates the overhead of having
 * a separate shadow pagetable.
 *
 * In order to allow this, it falls on the guest domain to map its
 * notion of a "physical" pfn - which is just a domain-local linear
 * address - into a real "machine address" which the CPU's MMU can
 * use.
 *
 * A pgd_t/pmd_t/pte_t will typically contain an mfn, and so can be
 * inserted directly into the pagetable.  When creating a new
 * pte/pmd/pgd, it converts the passed pfn into an mfn.  Conversely,
 * when reading the content back with __(pgd|pmd|pte)_val, it converts
 * the mfn back into a pfn.
 *
 * The other constraint is that all pages which make up a pagetable
 * must be mapped read-only in the guest.  This prevents uncontrolled
 * guest updates to the pagetable.  Xen strictly enforces this, and
 * will disallow any pagetable update which will end up mapping a
 * pagetable page RW, and will disallow using any writable page as a
 * pagetable.
 *
 * Naively, when loading %cr3 with the base of a new pagetable, Xen
 * would need to validate the whole pagetable before going on.
 * Naturally, this is quite slow.  The solution is to "pin" a
 * pagetable, which enforces all the constraints on the pagetable even
 * when it is not actively in use.  This menas that Xen can be assured
 * that it is still valid when you do load it into %cr3, and doesn't
 * need to revalidate it.
 *
 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
 */
#include <linux/sched.h>
#include <linux/highmem.h>
#include <linux/debugfs.h>
#include <linux/bug.h>
#include <linux/vmalloc.h>
#include <linux/module.h>
#include <linux/gfp.h>
#include <linux/memblock.h>
#include <linux/seq_file.h>

#include <trace/events/xen.h>

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/fixmap.h>
#include <asm/mmu_context.h>
#include <asm/setup.h>
#include <asm/paravirt.h>
#include <asm/e820.h>
#include <asm/linkage.h>
#include <asm/page.h>
#include <asm/init.h>
#include <asm/pat.h>
#include <asm/smp.h>

#include <asm/xen/hypercall.h>
#include <asm/xen/hypervisor.h>

#include <xen/xen.h>
#include <xen/page.h>
#include <xen/interface/xen.h>
#include <xen/interface/hvm/hvm_op.h>
#include <xen/interface/version.h>
#include <xen/interface/memory.h>
#include <xen/hvc-console.h>
#include <xen/balloon.h>

#include "multicalls.h"
#include "mmu.h"
#include "debugfs.h"

/*
 * Protects atomic reservation decrease/increase against concurrent increases.
 * Also protects non-atomic updates of current_pages and balloon lists.
 */
DEFINE_SPINLOCK(xen_reservation_lock);

/*
 * Identity map, in addition to plain kernel map.  This needs to be
 * large enough to allocate page table pages to allocate the rest.
 * Each page can map 2MB.
 */
#define LEVEL1_IDENT_ENTRIES	(PTRS_PER_PTE * 4)
static RESERVE_BRK_ARRAY(pte_t, level1_ident_pgt, LEVEL1_IDENT_ENTRIES);

#ifdef CONFIG_X86_64
/* l3 pud for userspace vsyscall mapping */
static pud_t level3_user_vsyscall[PTRS_PER_PUD] __page_aligned_bss;
#endif /* CONFIG_X86_64 */

/*
 * Note about cr3 (pagetable base) values:
 *
 * xen_cr3 contains the current logical cr3 value; it contains the
 * last set cr3.  This may not be the current effective cr3, because
 * its update may be being lazily deferred.  However, a vcpu looking
 * at its own cr3 can use this value knowing that it everything will
 * be self-consistent.
 *
 * xen_current_cr3 contains the actual vcpu cr3; it is set once the
 * hypercall to set the vcpu cr3 is complete (so it may be a little
 * out of date, but it will never be set early).  If one vcpu is
 * looking at another vcpu's cr3 value, it should use this variable.
 */
DEFINE_PER_CPU(unsigned long, xen_cr3);	 /* cr3 stored as physaddr */
DEFINE_PER_CPU(unsigned long, xen_current_cr3);	 /* actual vcpu cr3 */


/*
 * Just beyond the highest usermode address.  STACK_TOP_MAX has a
 * redzone above it, so round it up to a PGD boundary.
 */
#define USER_LIMIT	((STACK_TOP_MAX + PGDIR_SIZE - 1) & PGDIR_MASK)

unsigned long arbitrary_virt_to_mfn(void *vaddr)
{
	xmaddr_t maddr = arbitrary_virt_to_machine(vaddr);

	return PFN_DOWN(maddr.maddr);
}

xmaddr_t arbitrary_virt_to_machine(void *vaddr)
{
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;
	pte_t *pte;
	unsigned offset;

	/*
	 * if the PFN is in the linear mapped vaddr range, we can just use
	 * the (quick) virt_to_machine() p2m lookup
	 */
	if (virt_addr_valid(vaddr))
		return virt_to_machine(vaddr);

	/* otherwise we have to do a (slower) full page-table walk */

	pte = lookup_address(address, &level);
	BUG_ON(pte == NULL);
	offset = address & ~PAGE_MASK;
	return XMADDR(((phys_addr_t)pte_mfn(*pte) << PAGE_SHIFT) + offset);
}
EXPORT_SYMBOL_GPL(arbitrary_virt_to_machine);

void make_lowmem_page_readonly(void *vaddr)
{
	pte_t *pte, ptev;
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;

	pte = lookup_address(address, &level);
	if (pte == NULL)
		return;		/* vaddr missing */

	ptev = pte_wrprotect(*pte);

	if (HYPERVISOR_update_va_mapping(address, ptev, 0))
		BUG();
}

void make_lowmem_page_readwrite(void *vaddr)
{
	pte_t *pte, ptev;
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;

	pte = lookup_address(address, &level);
	if (pte == NULL)
		return;		/* vaddr missing */

	ptev = pte_mkwrite(*pte);

	if (HYPERVISOR_update_va_mapping(address, ptev, 0))
		BUG();
}


static bool xen_page_pinned(void *ptr)
{
	struct page *page = virt_to_page(ptr);

	return PagePinned(page);
}

void xen_set_domain_pte(pte_t *ptep, pte_t pteval, unsigned domid)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	trace_xen_mmu_set_domain_pte(ptep, pteval, domid);

	mcs = xen_mc_entry(sizeof(*u));
	u = mcs.args;

	/* ptep might be kmapped when using 32-bit HIGHPTE */
	u->ptr = virt_to_machine(ptep).maddr;
	u->val = pte_val_ma(pteval);

	MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, domid);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}
EXPORT_SYMBOL_GPL(xen_set_domain_pte);

static void xen_extend_mmu_update(const struct mmu_update *update)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	mcs = xen_mc_extend_args(__HYPERVISOR_mmu_update, sizeof(*u));

	if (mcs.mc != NULL) {
		mcs.mc->args[1]++;
	} else {
		mcs = __xen_mc_entry(sizeof(*u));
		MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
	}

	u = mcs.args;
	*u = *update;
}

static void xen_extend_mmuext_op(const struct mmuext_op *op)
{
	struct multicall_space mcs;
	struct mmuext_op *u;

	mcs = xen_mc_extend_args(__HYPERVISOR_mmuext_op, sizeof(*u));

	if (mcs.mc != NULL) {
		mcs.mc->args[1]++;
	} else {
		mcs = __xen_mc_entry(sizeof(*u));
		MULTI_mmuext_op(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
	}

	u = mcs.args;
	*u = *op;
}

static void xen_set_pmd_hyper(pmd_t *ptr, pmd_t val)
{
	struct mmu_update u;

	preempt_disable();

	xen_mc_batch();

	/* ptr may be ioremapped for 64-bit pagetable setup */
	u.ptr = arbitrary_virt_to_machine(ptr).maddr;
	u.val = pmd_val_ma(val);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_set_pmd(pmd_t *ptr, pmd_t val)
{
	trace_xen_mmu_set_pmd(ptr, val);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		return;
	}

	xen_set_pmd_hyper(ptr, val);
}

/*
 * Associate a virtual page frame with a given physical page frame
 * and protection flags for that frame.
 */
void set_pte_mfn(unsigned long vaddr, unsigned long mfn, pgprot_t flags)
{
	set_pte_vaddr(vaddr, mfn_pte(mfn, flags));
}

static bool xen_batched_set_pte(pte_t *ptep, pte_t pteval)
{
	struct mmu_update u;

	if (paravirt_get_lazy_mode() != PARAVIRT_LAZY_MMU)
		return false;

	xen_mc_batch();

	u.ptr = virt_to_machine(ptep).maddr | MMU_NORMAL_PT_UPDATE;
	u.val = pte_val_ma(pteval);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	return true;
}

static inline void __xen_set_pte(pte_t *ptep, pte_t pteval)
{
	if (!xen_batched_set_pte(ptep, pteval)) {
		/*
		 * Could call native_set_pte() here and trap and
		 * emulate the PTE write but with 32-bit guests this
		 * needs two traps (one for each of the two 32-bit
		 * words in the PTE) so do one hypercall directly
		 * instead.
		 */
		struct mmu_update u;

		u.ptr = virt_to_machine(ptep).maddr | MMU_NORMAL_PT_UPDATE;
		u.val = pte_val_ma(pteval);
		HYPERVISOR_mmu_update(&u, 1, NULL, DOMID_SELF);
	}
}

static void xen_set_pte(pte_t *ptep, pte_t pteval)
{
	trace_xen_mmu_set_pte(ptep, pteval);
	__xen_set_pte(ptep, pteval);
}

void xen_set_clr_mmio_pvh_pte(unsigned long pfn, unsigned long mfn, 
			      int nr_mfns, int add_mapping)
{
	struct physdev_map_iomem iomem;

	iomem.first_gfn = pfn;
	iomem.first_mfn = mfn;
	iomem.nr_mfns = nr_mfns;
	iomem.add_mapping = add_mapping;

	if (HYPERVISOR_physdev_op(PHYSDEVOP_pvh_map_iomem, &iomem))
		BUG();
}

static void xen_dom0pvh_set_pte_at(struct mm_struct *mm, unsigned long addr,
		    		   pte_t *ptep, pte_t pteval)
{
	native_set_pte(ptep, pteval);
}

static void xen_set_pte_at(struct mm_struct *mm, unsigned long addr,
		    pte_t *ptep, pte_t pteval)
{
	trace_xen_mmu_set_pte_at(mm, addr, ptep, pteval);
	__xen_set_pte(ptep, pteval);
}

pte_t xen_ptep_modify_prot_start(struct mm_struct *mm,
				 unsigned long addr, pte_t *ptep)
{
	/* Just return the pte as-is.  We preserve the bits on commit */
	trace_xen_mmu_ptep_modify_prot_start(mm, addr, ptep, *ptep);
	return *ptep;
}

void xen_ptep_modify_prot_commit(struct mm_struct *mm, unsigned long addr,
				 pte_t *ptep, pte_t pte)
{
	struct mmu_update u;

	trace_xen_mmu_ptep_modify_prot_commit(mm, addr, ptep, pte);
	xen_mc_batch();

	u.ptr = virt_to_machine(ptep).maddr | MMU_PT_UPDATE_PRESERVE_AD;
	u.val = pte_val_ma(pte);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}

/* Assume pteval_t is equivalent to all the other *val_t types. */
static pteval_t pte_mfn_to_pfn(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long mfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		unsigned long pfn = mfn_to_pfn(mfn);

		pteval_t flags = val & PTE_FLAGS_MASK;
		if (unlikely(pfn == ~0))
			val = flags & ~_PAGE_PRESENT;
		else
			val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t pte_pfn_to_mfn(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		pteval_t flags = val & PTE_FLAGS_MASK;
		unsigned long mfn;

		if (!xen_feature(XENFEAT_auto_translated_physmap))
			mfn = get_phys_to_machine(pfn);
		else
			mfn = pfn;
		/*
		 * If there's no mfn for the pfn, then just create an
		 * empty non-present pte.  Unfortunately this loses
		 * information about the original pfn, so
		 * pte_mfn_to_pfn is asymmetric.
		 */
		if (unlikely(mfn == INVALID_P2M_ENTRY)) {
			mfn = 0;
			flags = 0;
		} else {
			/*
			 * Paramount to do this test _after_ the
			 * INVALID_P2M_ENTRY as INVALID_P2M_ENTRY &
			 * IDENTITY_FRAME_BIT resolves to true.
			 */
			mfn &= ~FOREIGN_FRAME_BIT;
			if (mfn & IDENTITY_FRAME_BIT) {
				mfn &= ~IDENTITY_FRAME_BIT;
				flags |= _PAGE_IOMAP;
			}
		}
		val = ((pteval_t)mfn << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t iomap_pte(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		pteval_t flags = val & PTE_FLAGS_MASK;

		/* We assume the pte frame number is a MFN, so
		   just use it as-is. */
		val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t xen_pte_val(pte_t pte)
{
	pteval_t pteval = pte.pte;
#if 0
	/* If this is a WC pte, convert back from Xen WC to Linux WC */
	if ((pteval & (_PAGE_PAT | _PAGE_PCD | _PAGE_PWT)) == _PAGE_PAT) {
		WARN_ON(!pat_enabled);
		pteval = (pteval & ~_PAGE_PAT) | _PAGE_PWT;
	}
#endif
	if (xen_initial_domain() && (pteval & _PAGE_IOMAP))
		return pteval;

	return pte_mfn_to_pfn(pteval);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pte_val);

static pgdval_t xen_pgd_val(pgd_t pgd)
{
	return pte_mfn_to_pfn(pgd.pgd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pgd_val);

/*
 * Xen's PAT setup is part of its ABI, though I assume entries 6 & 7
 * are reserved for now, to correspond to the Intel-reserved PAT
 * types.
 *
 * We expect Linux's PAT set as follows:
 *
 * Idx  PTE flags        Linux    Xen    Default
 * 0                     WB       WB     WB
 * 1            PWT      WC       WT     WT
 * 2        PCD          UC-      UC-    UC-
 * 3        PCD PWT      UC       UC     UC
 * 4    PAT              WB       WC     WB
 * 5    PAT     PWT      WC       WP     WT
 * 6    PAT PCD          UC-      UC     UC-
 * 7    PAT PCD PWT      UC       UC     UC
 */

void xen_set_pat(u64 pat)
{
	/* We expect Linux to use a PAT setting of
	 * UC UC- WC WB (ignoring the PAT flag) */
	WARN_ON(pat != 0x0007010600070106ull);
}

static pte_t xen_make_pte(pteval_t pte)
{
	phys_addr_t addr = (pte & PTE_PFN_MASK);
#if 0
	/* If Linux is trying to set a WC pte, then map to the Xen WC.
	 * If _PAGE_PAT is set, then it probably means it is really
	 * _PAGE_PSE, so avoid fiddling with the PAT mapping and hope
	 * things work out OK...
	 *
	 * (We should never see kernel mappings with _PAGE_PSE set,
	 * but we could see hugetlbfs mappings, I think.).
	 */
	if (pat_enabled && !WARN_ON(pte & _PAGE_PAT)) {
		if ((pte & (_PAGE_PCD | _PAGE_PWT)) == _PAGE_PWT)
			pte = (pte & ~(_PAGE_PCD | _PAGE_PWT)) | _PAGE_PAT;
	}
#endif
	/*
	 * Unprivileged domains are allowed to do IOMAPpings for
	 * PCI passthrough, but not map ISA space.  The ISA
	 * mappings are just dummy local mappings to keep other
	 * parts of the kernel happy.
	 */
	if (unlikely(pte & _PAGE_IOMAP) &&
	    (xen_initial_domain() || addr >= ISA_END_ADDRESS)) {
		pte = iomap_pte(pte);
	} else {
		pte &= ~_PAGE_IOMAP;
		pte = pte_pfn_to_mfn(pte);
	}

	return native_make_pte(pte);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pte);

static pgd_t xen_make_pgd(pgdval_t pgd)
{
	pgd = pte_pfn_to_mfn(pgd);
	return native_make_pgd(pgd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pgd);

static pmdval_t xen_pmd_val(pmd_t pmd)
{
	return pte_mfn_to_pfn(pmd.pmd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pmd_val);

static void xen_set_pud_hyper(pud_t *ptr, pud_t val)
{
	struct mmu_update u;

	preempt_disable();

	xen_mc_batch();

	/* ptr may be ioremapped for 64-bit pagetable setup */
	u.ptr = arbitrary_virt_to_machine(ptr).maddr;
	u.val = pud_val_ma(val);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_set_pud(pud_t *ptr, pud_t val)
{
	trace_xen_mmu_set_pud(ptr, val);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		return;
	}

	xen_set_pud_hyper(ptr, val);
}

#ifdef CONFIG_X86_PAE
static void xen_set_pte_atomic(pte_t *ptep, pte_t pte)
{
	trace_xen_mmu_set_pte_atomic(ptep, pte);
	set_64bit((u64 *)ptep, native_pte_val(pte));
}

static void xen_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
{
	trace_xen_mmu_pte_clear(mm, addr, ptep);
	if (!xen_batched_set_pte(ptep, native_make_pte(0)))
		native_pte_clear(mm, addr, ptep);
}

static void xen_pmd_clear(pmd_t *pmdp)
{
	trace_xen_mmu_pmd_clear(pmdp);
	set_pmd(pmdp, __pmd(0));
}
#endif	/* CONFIG_X86_PAE */

static pmd_t xen_make_pmd(pmdval_t pmd)
{
	pmd = pte_pfn_to_mfn(pmd);
	return native_make_pmd(pmd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pmd);

#if PAGETABLE_LEVELS == 4
static pudval_t xen_pud_val(pud_t pud)
{
	return pte_mfn_to_pfn(pud.pud);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pud_val);

static pud_t xen_make_pud(pudval_t pud)
{
	pud = pte_pfn_to_mfn(pud);

	return native_make_pud(pud);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pud);

static pgd_t *xen_get_user_pgd(pgd_t *pgd)
{
	pgd_t *pgd_page = (pgd_t *)(((unsigned long)pgd) & PAGE_MASK);
	unsigned offset = pgd - pgd_page;
	pgd_t *user_ptr = NULL;

	if (offset < pgd_index(USER_LIMIT)) {
		struct page *page = virt_to_page(pgd_page);
		user_ptr = (pgd_t *)page->private;
		if (user_ptr)
			user_ptr += offset;
	}

	return user_ptr;
}

static void __xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
	struct mmu_update u;

	u.ptr = virt_to_machine(ptr).maddr;
	u.val = pgd_val_ma(val);
	xen_extend_mmu_update(&u);
}

/*
 * Raw hypercall-based set_pgd, intended for in early boot before
 * there's a page structure.  This implies:
 *  1. The only existing pagetable is the kernel's
 *  2. It is always pinned
 *  3. It has no user pagetable attached to it
 */
static void __init xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
	preempt_disable();

	xen_mc_batch();

	__xen_set_pgd_hyper(ptr, val);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_set_pgd(pgd_t *ptr, pgd_t val)
{
	pgd_t *user_ptr = xen_get_user_pgd(ptr);

	trace_xen_mmu_set_pgd(ptr, user_ptr, val);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		if (user_ptr) {
			WARN_ON(xen_page_pinned(user_ptr));
			*user_ptr = val;
		}
		return;
	}

	/* If it's pinned, then we can at least batch the kernel and
	   user updates together. */
	xen_mc_batch();

	__xen_set_pgd_hyper(ptr, val);
	if (user_ptr)
		__xen_set_pgd_hyper(user_ptr, val);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}
#endif	/* PAGETABLE_LEVELS == 4 */

/*
 * (Yet another) pagetable walker.  This one is intended for pinning a
 * pagetable.  This means that it walks a pagetable and calls the
 * callback function on each page it finds making up the page table,
 * at every level.  It walks the entire pagetable, but it only bothers
 * pinning pte pages which are below limit.  In the normal case this
 * will be STACK_TOP_MAX, but at boot we need to pin up to
 * FIXADDR_TOP.
 *
 * For 32-bit the important bit is that we don't pin beyond there,
 * because then we start getting into Xen's ptes.
 *
 * For 64-bit, we must skip the Xen hole in the middle of the address
 * space, just after the big x86-64 virtual hole.
 */
static int __xen_pgd_walk(struct mm_struct *mm, pgd_t *pgd,
			  int (*func)(struct mm_struct *mm, struct page *,
				      enum pt_level),
			  unsigned long limit)
{
	int flush = 0;
	unsigned hole_low, hole_high;
	unsigned pgdidx_limit, pudidx_limit, pmdidx_limit;
	unsigned pgdidx, pudidx, pmdidx;

	/* The limit is the last byte to be touched */
	limit--;
	BUG_ON(limit >= FIXADDR_TOP);

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	/*
	 * 64-bit has a great big hole in the middle of the address
	 * space, which contains the Xen mappings.  On 32-bit these
	 * will end up making a zero-sized hole and so is a no-op.
	 */
	hole_low = pgd_index(USER_LIMIT);
	hole_high = pgd_index(PAGE_OFFSET);

	pgdidx_limit = pgd_index(limit);
#if PTRS_PER_PUD > 1
	pudidx_limit = pud_index(limit);
#else
	pudidx_limit = 0;
#endif
#if PTRS_PER_PMD > 1
	pmdidx_limit = pmd_index(limit);
#else
	pmdidx_limit = 0;
#endif

	for (pgdidx = 0; pgdidx <= pgdidx_limit; pgdidx++) {
		pud_t *pud;

		if (pgdidx >= hole_low && pgdidx < hole_high)
			continue;

		if (!pgd_val(pgd[pgdidx]))
			continue;

		pud = pud_offset(&pgd[pgdidx], 0);

		if (PTRS_PER_PUD > 1) /* not folded */
			flush |= (*func)(mm, virt_to_page(pud), PT_PUD);

		for (pudidx = 0; pudidx < PTRS_PER_PUD; pudidx++) {
			pmd_t *pmd;

			if (pgdidx == pgdidx_limit &&
			    pudidx > pudidx_limit)
				goto out;

			if (pud_none(pud[pudidx]))
				continue;

			pmd = pmd_offset(&pud[pudidx], 0);

			if (PTRS_PER_PMD > 1) /* not folded */
				flush |= (*func)(mm, virt_to_page(pmd), PT_PMD);

			for (pmdidx = 0; pmdidx < PTRS_PER_PMD; pmdidx++) {
				struct page *pte;

				if (pgdidx == pgdidx_limit &&
				    pudidx == pudidx_limit &&
				    pmdidx > pmdidx_limit)
					goto out;

				if (pmd_none(pmd[pmdidx]))
					continue;

				pte = pmd_page(pmd[pmdidx]);
				flush |= (*func)(mm, pte, PT_PTE);
			}
		}
	}

out:
	/* Do the top level last, so that the callbacks can use it as
	   a cue to do final things like tlb flushes. */
	flush |= (*func)(mm, virt_to_page(pgd), PT_PGD);

	return flush;
}

static int xen_pgd_walk(struct mm_struct *mm,
			int (*func)(struct mm_struct *mm, struct page *,
				    enum pt_level),
			unsigned long limit)
{
	return __xen_pgd_walk(mm, mm->pgd, func, limit);
}

/* If we're using split pte locks, then take the page's lock and
   return a pointer to it.  Otherwise return NULL. */
static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm)
{
	spinlock_t *ptl = NULL;

#if USE_SPLIT_PTLOCKS
	ptl = __pte_lockptr(page);
	spin_lock_nest_lock(ptl, &mm->page_table_lock);
#endif

	return ptl;
}

static void xen_pte_unlock(void *v)
{
	spinlock_t *ptl = v;
	spin_unlock(ptl);
}

static void xen_do_pin(unsigned level, unsigned long pfn)
{
	struct mmuext_op op;

	op.cmd = level;
	op.arg1.mfn = pfn_to_mfn(pfn);

	xen_extend_mmuext_op(&op);
}

static int xen_pin_page(struct mm_struct *mm, struct page *page,
			enum pt_level level)
{
	unsigned pgfl = TestSetPagePinned(page);
	int flush;

	if (pgfl)
		flush = 0;		/* already pinned */
	else if (PageHighMem(page))
		/* kmaps need flushing if we found an unpinned
		   highpage */
		flush = 1;
	else {
		void *pt = lowmem_page_address(page);
		unsigned long pfn = page_to_pfn(page);
		struct multicall_space mcs = __xen_mc_entry(0);
		spinlock_t *ptl;

		flush = 0;

		/*
		 * We need to hold the pagetable lock between the time
		 * we make the pagetable RO and when we actually pin
		 * it.  If we don't, then other users may come in and
		 * attempt to update the pagetable by writing it,
		 * which will fail because the memory is RO but not
		 * pinned, so Xen won't do the trap'n'emulate.
		 *
		 * If we're using split pte locks, we can't hold the
		 * entire pagetable's worth of locks during the
		 * traverse, because we may wrap the preempt count (8
		 * bits).  The solution is to mark RO and pin each PTE
		 * page while holding the lock.  This means the number
		 * of locks we end up holding is never more than a
		 * batch size (~32 entries, at present).
		 *
		 * If we're not using split pte locks, we needn't pin
		 * the PTE pages independently, because we're
		 * protected by the overall pagetable lock.
		 */
		ptl = NULL;
		if (level == PT_PTE)
			ptl = xen_pte_lock(page, mm);

		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
					pfn_pte(pfn, PAGE_KERNEL_RO),
					level == PT_PGD ? UVMF_TLB_FLUSH : 0);

		if (ptl) {
			xen_do_pin(MMUEXT_PIN_L1_TABLE, pfn);

			/* Queue a deferred unlock for when this batch
			   is completed. */
			xen_mc_callback(xen_pte_unlock, ptl);
		}
	}

	return flush;
}

/* This is called just after a mm has been created, but it has not
   been used yet.  We need to make sure that its pagetable is all
   read-only, and can be pinned. */
static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd)
{
	trace_xen_mmu_pgd_pin(mm, pgd);

	xen_mc_batch();

	if (__xen_pgd_walk(mm, pgd, xen_pin_page, USER_LIMIT)) {
		/* re-enable interrupts for flushing */
		xen_mc_issue(0);

		kmap_flush_unused();

		xen_mc_batch();
	}

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(pgd);

		xen_do_pin(MMUEXT_PIN_L4_TABLE, PFN_DOWN(__pa(pgd)));

		if (user_pgd) {
			xen_pin_page(mm, virt_to_page(user_pgd), PT_PGD);
			xen_do_pin(MMUEXT_PIN_L4_TABLE,
				   PFN_DOWN(__pa(user_pgd)));
		}
	}
#else /* CONFIG_X86_32 */
#ifdef CONFIG_X86_PAE
	/* Need to make sure unshared kernel PMD is pinnable */
	xen_pin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]),
		     PT_PMD);
#endif
	xen_do_pin(MMUEXT_PIN_L3_TABLE, PFN_DOWN(__pa(pgd)));
#endif /* CONFIG_X86_64 */
	xen_mc_issue(0);
}

static void xen_pgd_pin(struct mm_struct *mm)
{
	__xen_pgd_pin(mm, mm->pgd);
}

/*
 * On save, we need to pin all pagetables to make sure they get their
 * mfns turned into pfns.  Search the list for any unpinned pgds and pin
 * them (unpinned pgds are not currently in use, probably because the
 * process is under construction or destruction).
 *
 * Expected to be called in stop_machine() ("equivalent to taking
 * every spinlock in the system"), so the locking doesn't really
 * matter all that much.
 */
void xen_mm_pin_all(void)
{
	struct page *page;

	spin_lock(&pgd_lock);

	list_for_each_entry(page, &pgd_list, lru) {
		if (!PagePinned(page)) {
			__xen_pgd_pin(&init_mm, (pgd_t *)page_address(page));
			SetPageSavePinned(page);
		}
	}

	spin_unlock(&pgd_lock);
}

/*
 * The init_mm pagetable is really pinned as soon as its created, but
 * that's before we have page structures to store the bits.  So do all
 * the book-keeping now.
 */
static int __init xen_mark_pinned(struct mm_struct *mm, struct page *page,
				  enum pt_level level)
{
	SetPagePinned(page);
	return 0;
}

static void __init xen_mark_init_mm_pinned(void)
{
	xen_pgd_walk(&init_mm, xen_mark_pinned, FIXADDR_TOP);
}

static int xen_unpin_page(struct mm_struct *mm, struct page *page,
			  enum pt_level level)
{
	unsigned pgfl = TestClearPagePinned(page);

	if (pgfl && !PageHighMem(page)) {
		void *pt = lowmem_page_address(page);
		unsigned long pfn = page_to_pfn(page);
		spinlock_t *ptl = NULL;
		struct multicall_space mcs;

		/*
		 * Do the converse to pin_page.  If we're using split
		 * pte locks, we must be holding the lock for while
		 * the pte page is unpinned but still RO to prevent
		 * concurrent updates from seeing it in this
		 * partially-pinned state.
		 */
		if (level == PT_PTE) {
			ptl = xen_pte_lock(page, mm);

			if (ptl)
				xen_do_pin(MMUEXT_UNPIN_TABLE, pfn);
		}

		mcs = __xen_mc_entry(0);

		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
					pfn_pte(pfn, PAGE_KERNEL),
					level == PT_PGD ? UVMF_TLB_FLUSH : 0);

		if (ptl) {
			/* unlock when batch completed */
			xen_mc_callback(xen_pte_unlock, ptl);
		}
	}

	return 0;		/* never need to flush on unpin */
}

/* Release a pagetables pages back as normal RW */
static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd)
{
	trace_xen_mmu_pgd_unpin(mm, pgd);

	xen_mc_batch();

	xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(pgd);

		if (user_pgd) {
			xen_do_pin(MMUEXT_UNPIN_TABLE,
				   PFN_DOWN(__pa(user_pgd)));
			xen_unpin_page(mm, virt_to_page(user_pgd), PT_PGD);
		}
	}
#endif

#ifdef CONFIG_X86_PAE
	/* Need to make sure unshared kernel PMD is unpinned */
	xen_unpin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]),
		       PT_PMD);
#endif

	__xen_pgd_walk(mm, pgd, xen_unpin_page, USER_LIMIT);

	xen_mc_issue(0);
}

static void xen_pgd_unpin(struct mm_struct *mm)
{
	__xen_pgd_unpin(mm, mm->pgd);
}

/*
 * On resume, undo any pinning done at save, so that the rest of the
 * kernel doesn't see any unexpected pinned pagetables.
 */
void xen_mm_unpin_all(void)
{
	struct page *page;

	spin_lock(&pgd_lock);

	list_for_each_entry(page, &pgd_list, lru) {
		if (PageSavePinned(page)) {
			BUG_ON(!PagePinned(page));
			__xen_pgd_unpin(&init_mm, (pgd_t *)page_address(page));
			ClearPageSavePinned(page);
		}
	}

	spin_unlock(&pgd_lock);
}

static void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)
{
	spin_lock(&next->page_table_lock);
	xen_pgd_pin(next);
	spin_unlock(&next->page_table_lock);
}

static void xen_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
{
	spin_lock(&mm->page_table_lock);
	xen_pgd_pin(mm);
	spin_unlock(&mm->page_table_lock);
}


#ifdef CONFIG_SMP
/* Another cpu may still have their %cr3 pointing at the pagetable, so
   we need to repoint it somewhere else before we can unpin it. */
static void drop_other_mm_ref(void *info)
{
	struct mm_struct *mm = info;
	struct mm_struct *active_mm;

	active_mm = this_cpu_read(cpu_tlbstate.active_mm);

	if (active_mm == mm && this_cpu_read(cpu_tlbstate.state) != TLBSTATE_OK)
		leave_mm(smp_processor_id());

	/* If this cpu still has a stale cr3 reference, then make sure
	   it has been flushed. */
	if (this_cpu_read(xen_current_cr3) == __pa(mm->pgd))
		load_cr3(swapper_pg_dir);
}

static void xen_drop_mm_ref(struct mm_struct *mm)
{
	cpumask_var_t mask;
	unsigned cpu;

	if (current->active_mm == mm) {
		if (current->mm == mm)
			load_cr3(swapper_pg_dir);
		else
			leave_mm(smp_processor_id());
	}

	/* Get the "official" set of cpus referring to our pagetable. */
	if (!alloc_cpumask_var(&mask, GFP_ATOMIC)) {
		for_each_online_cpu(cpu) {
			if (!cpumask_test_cpu(cpu, mm_cpumask(mm))
			    && per_cpu(xen_current_cr3, cpu) != __pa(mm->pgd))
				continue;
			smp_call_function_single(cpu, drop_other_mm_ref, mm, 1);
		}
		return;
	}
	cpumask_copy(mask, mm_cpumask(mm));

	/* It's possible that a vcpu may have a stale reference to our
	   cr3, because its in lazy mode, and it hasn't yet flushed
	   its set of pending hypercalls yet.  In this case, we can
	   look at its actual current cr3 value, and force it to flush
	   if needed. */
	for_each_online_cpu(cpu) {
		if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd))
			cpumask_set_cpu(cpu, mask);
	}

	if (!cpumask_empty(mask))
		smp_call_function_many(mask, drop_other_mm_ref, mm, 1);
	free_cpumask_var(mask);
}
#else
static void xen_drop_mm_ref(struct mm_struct *mm)
{
	if (current->active_mm == mm)
		load_cr3(swapper_pg_dir);
}
#endif

/*
 * While a process runs, Xen pins its pagetables, which means that the
 * hypervisor forces it to be read-only, and it controls all updates
 * to it.  This means that all pagetable updates have to go via the
 * hypervisor, which is moderately expensive.
 *
 * Since we're pulling the pagetable down, we switch to use init_mm,
 * unpin old process pagetable and mark it all read-write, which
 * allows further operations on it to be simple memory accesses.
 *
 * The only subtle point is that another CPU may be still using the
 * pagetable because of lazy tlb flushing.  This means we need need to
 * switch all CPUs off this pagetable before we can unpin it.
 */
static void xen_exit_mmap(struct mm_struct *mm)
{
	get_cpu();		/* make sure we don't move around */
	xen_drop_mm_ref(mm);
	put_cpu();

	spin_lock(&mm->page_table_lock);

	/* pgd may not be pinned in the error exit path of execve */
	if (xen_page_pinned(mm->pgd))
		xen_pgd_unpin(mm);

	spin_unlock(&mm->page_table_lock);
}

static void __init xen_pagetable_setup_start(pgd_t *base)
{
}

static __init void xen_mapping_pagetable_reserve(u64 start, u64 end)
{
	/* reserve the range used */
	native_pagetable_reserve(start, end);

	/* set as RW the rest */
	printk(KERN_DEBUG "xen: setting RW the range %llx - %llx\n", end,
			PFN_PHYS(pgt_buf_top));
	while (end < PFN_PHYS(pgt_buf_top)) {
		make_lowmem_page_readwrite(__va(end));
		end += PAGE_SIZE;
	}
}

static void xen_post_allocator_init(void);

static void __init xen_pagetable_setup_done(pgd_t *base)
{
	xen_setup_shared_info();

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	xen_post_allocator_init();
}

static void xen_write_cr2(unsigned long cr2)
{
	this_cpu_read(xen_vcpu)->arch.cr2 = cr2;
}

static unsigned long xen_read_cr2(void)
{
	return this_cpu_read(xen_vcpu)->arch.cr2;
}

unsigned long xen_read_cr2_direct(void)
{
	return this_cpu_read(xen_vcpu_info.arch.cr2);
}

static void xen_flush_tlb(void)
{
	struct mmuext_op *op;
	struct multicall_space mcs;

	trace_xen_mmu_flush_tlb(0);

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*op));

	op = mcs.args;
	op->cmd = MMUEXT_TLB_FLUSH_LOCAL;
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_flush_tlb_single(unsigned long addr)
{
	struct mmuext_op *op;
	struct multicall_space mcs;

	trace_xen_mmu_flush_tlb_single(addr);

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*op));
	op = mcs.args;
	op->cmd = MMUEXT_INVLPG_LOCAL;
	op->arg1.linear_addr = addr & PAGE_MASK;
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_flush_tlb_others(const struct cpumask *cpus,
				 struct mm_struct *mm, unsigned long start,
				 unsigned long end)
{
	struct {
		struct mmuext_op op;
#ifdef CONFIG_SMP
		DECLARE_BITMAP(mask, num_processors);
#else
		DECLARE_BITMAP(mask, NR_CPUS);
#endif
	} *args;
	struct multicall_space mcs;

	trace_xen_mmu_flush_tlb_others(cpus, mm, start, end);

	if (cpumask_empty(cpus))
		return;		/* nothing to do */

	mcs = xen_mc_entry(sizeof(*args));
	args = mcs.args;
	args->op.arg2.vcpumask = to_cpumask(args->mask);

	/* Remove us, and any offline CPUS. */
	cpumask_and(to_cpumask(args->mask), cpus, cpu_online_mask);
	cpumask_clear_cpu(smp_processor_id(), to_cpumask(args->mask));

	args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
	if (start != TLB_FLUSH_ALL && (end - start) <= PAGE_SIZE) {
		args->op.cmd = MMUEXT_INVLPG_MULTI;
		args->op.arg1.linear_addr = start;
	}

	MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}

static unsigned long xen_read_cr3(void)
{
	return this_cpu_read(xen_cr3);
}

static void set_current_cr3(void *v)
{
	this_cpu_write(xen_current_cr3, (unsigned long)v);
}

static void __xen_write_cr3(bool kernel, unsigned long cr3)
{
	struct mmuext_op op;
	unsigned long mfn;

	trace_xen_mmu_write_cr3(kernel, cr3);

	if (cr3)
		mfn = pfn_to_mfn(PFN_DOWN(cr3));
	else
		mfn = 0;

	WARN_ON(mfn == 0 && kernel);

	op.cmd = kernel ? MMUEXT_NEW_BASEPTR : MMUEXT_NEW_USER_BASEPTR;
	op.arg1.mfn = mfn;

	xen_extend_mmuext_op(&op);

	if (kernel) {
		this_cpu_write(xen_cr3, cr3);

		/* Update xen_current_cr3 once the batch has actually
		   been submitted. */
		xen_mc_callback(set_current_cr3, (void *)cr3);
	}
}

static void xen_write_cr3(unsigned long cr3)
{
	BUG_ON(preemptible());

	xen_mc_batch();  /* disables interrupts */

	/* Update while interrupts are disabled, so its atomic with
	   respect to ipis */
	this_cpu_write(xen_cr3, cr3);

	__xen_write_cr3(true, cr3);

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(__va(cr3));
		if (user_pgd)
			__xen_write_cr3(false, __pa(user_pgd));
		else
			__xen_write_cr3(false, 0);
	}
#endif

	xen_mc_issue(PARAVIRT_LAZY_CPU);  /* interrupts restored */
}

static int xen_pgd_alloc(struct mm_struct *mm)
{
	pgd_t *pgd = mm->pgd;
	int ret = 0;

	BUG_ON(PagePinned(virt_to_page(pgd)));

#ifdef CONFIG_X86_64
	{
		struct page *page = virt_to_page(pgd);
		pgd_t *user_pgd;

		BUG_ON(page->private != 0);

		ret = -ENOMEM;

		user_pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
		page->private = (unsigned long)user_pgd;

		if (user_pgd != NULL) {
			user_pgd[pgd_index(VSYSCALL_START)] =
				__pgd(__pa(level3_user_vsyscall) | _PAGE_TABLE);
			ret = 0;
		}

		BUG_ON(PagePinned(virt_to_page(xen_get_user_pgd(pgd))));
	}
#endif

	return ret;
}

static void xen_pgd_free(struct mm_struct *mm, pgd_t *pgd)
{
#ifdef CONFIG_X86_64
	pgd_t *user_pgd = xen_get_user_pgd(pgd);

	if (user_pgd)
		free_page((unsigned long)user_pgd);
#endif
}

#ifdef CONFIG_X86_32
static pte_t __init mask_rw_pte(pte_t *ptep, pte_t pte)
{
	/* If there's an existing pte, then don't allow _PAGE_RW to be set */
	if (pte_val_ma(*ptep) & _PAGE_PRESENT)
		pte = __pte_ma(((pte_val_ma(*ptep) & _PAGE_RW) | ~_PAGE_RW) &
			       pte_val_ma(pte));

	return pte;
}
#else /* CONFIG_X86_64 */
static pte_t __init mask_rw_pte(pte_t *ptep, pte_t pte)
{
	unsigned long pfn = pte_pfn(pte);

	/*
	 * If the new pfn is within the range of the newly allocated
	 * kernel pagetable, and it isn't being mapped into an
	 * early_ioremap fixmap slot as a freshly allocated page, make sure
	 * it is RO.
	 */
	if (((!is_early_ioremap_ptep(ptep) &&
			pfn >= pgt_buf_start && pfn < pgt_buf_top)) ||
			(is_early_ioremap_ptep(ptep) && pfn != (pgt_buf_end - 1)))
		pte = pte_wrprotect(pte);

	return pte;
}
#endif /* CONFIG_X86_64 */

/*
 * Init-time set_pte while constructing initial pagetables, which
 * doesn't allow RO page table pages to be remapped RW.
 *
 * If there is no MFN for this PFN then this page is initially
 * ballooned out so clear the PTE (as in decrease_reservation() in
 * drivers/xen/balloon.c).
 *
 * Many of these PTE updates are done on unpinned and writable pages
 * and doing a hypercall for these is unnecessary and expensive.  At
 * this point it is not possible to tell if a page is pinned or not,
 * so always write the PTE directly and rely on Xen trapping and
 * emulating any updates as necessary.
 */
static void __init xen_set_pte_init(pte_t *ptep, pte_t pte)
{
	if (pte_mfn(pte) != INVALID_P2M_ENTRY)
		pte = mask_rw_pte(ptep, pte);
	else
		pte = __pte_ma(0);

	native_set_pte(ptep, pte);
}

static noinline void pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
{
	struct mmuext_op op;

	if (xen_feature(XENFEAT_writable_page_tables))
		return;

	op.cmd = cmd;
	op.arg1.mfn = pfn_to_mfn(pfn);
	if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
		BUG();
}

/* Early in boot, while setting up the initial pagetable, assume
   everything is pinned. */
static void __init xen_alloc_pte_init(struct mm_struct *mm, unsigned long pfn)
{
#ifdef CONFIG_FLATMEM
	BUG_ON(mem_map);	/* should only be used early */
#endif
	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
	pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
}

/* Used for pmd and pud */
static void __init xen_alloc_pmd_init(struct mm_struct *mm, unsigned long pfn)
{
#ifdef CONFIG_FLATMEM
	BUG_ON(mem_map);	/* should only be used early */
#endif
	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
}

/* Early release_pte assumes that all pts are pinned, since there's
   only init_mm and anything attached to that is pinned. */
static void __init xen_release_pte_init(unsigned long pfn)
{
	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}

static void __init xen_release_pmd_init(unsigned long pfn)
{
	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}

static inline void __pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
{
	struct multicall_space mcs;
	struct mmuext_op *op;

	mcs = __xen_mc_entry(sizeof(*op));
	op = mcs.args;
	op->cmd = cmd;
	op->arg1.mfn = pfn_to_mfn(pfn);

	MULTI_mmuext_op(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
}

static inline void __set_pfn_prot(unsigned long pfn, pgprot_t prot)
{
	struct multicall_space mcs;
	unsigned long addr = (unsigned long)__va(pfn << PAGE_SHIFT);

	mcs = __xen_mc_entry(0);
	MULTI_update_va_mapping(mcs.mc, (unsigned long)addr,
				pfn_pte(pfn, prot), 0);
}

/* This needs to make sure the new pte page is pinned iff its being
   attached to a pinned pagetable. */
static inline void xen_alloc_ptpage(struct mm_struct *mm, unsigned long pfn,
				    unsigned level)
{
	bool pinned = PagePinned(virt_to_page(mm->pgd));

	trace_xen_mmu_alloc_ptpage(mm, pfn, level, pinned);

	if (pinned) {
		struct page *page = pfn_to_page(pfn);

		SetPagePinned(page);

		if (!PageHighMem(page)) {
			xen_mc_batch();

			__set_pfn_prot(pfn, PAGE_KERNEL_RO);

			if (level == PT_PTE && USE_SPLIT_PTLOCKS)
				__pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);

			xen_mc_issue(PARAVIRT_LAZY_MMU);
		} else {
			/* make sure there are no stray mappings of
			   this page */
			kmap_flush_unused();
		}
	}
}

static void xen_alloc_pte(struct mm_struct *mm, unsigned long pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PTE);
}

static void xen_alloc_pmd(struct mm_struct *mm, unsigned long pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PMD);
}

/* This should never happen until we're OK to use struct page */
static inline void xen_release_ptpage(unsigned long pfn, unsigned level)
{
	struct page *page = pfn_to_page(pfn);
	bool pinned = PagePinned(page);

	trace_xen_mmu_release_ptpage(pfn, level, pinned);

	if (pinned) {
		if (!PageHighMem(page)) {
			xen_mc_batch();

			if (level == PT_PTE && USE_SPLIT_PTLOCKS)
				__pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);

			__set_pfn_prot(pfn, PAGE_KERNEL);

			xen_mc_issue(PARAVIRT_LAZY_MMU);
		}
		ClearPagePinned(page);
	}
}

static void xen_release_pte(unsigned long pfn)
{
	xen_release_ptpage(pfn, PT_PTE);
}

static void xen_release_pmd(unsigned long pfn)
{
	xen_release_ptpage(pfn, PT_PMD);
}

#if PAGETABLE_LEVELS == 4
static void xen_alloc_pud(struct mm_struct *mm, unsigned long pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PUD);
}

static void xen_release_pud(unsigned long pfn)
{
	xen_release_ptpage(pfn, PT_PUD);
}
#endif

void __init xen_reserve_top(void)
{
#ifdef CONFIG_X86_32
	unsigned long top = HYPERVISOR_VIRT_START;
	struct xen_platform_parameters pp;

	if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
		top = pp.virt_start;

	reserve_top_address(-top);
#endif	/* CONFIG_X86_32 */
}

/*
 * Like __va(), but returns address in the kernel mapping (which is
 * all we have until the physical memory mapping has been set up.
 */
static void *__ka(phys_addr_t paddr)
{
#ifdef CONFIG_X86_64
	return (void *)(paddr + __START_KERNEL_map);
#else
	return __va(paddr);
#endif
}

/* Convert a machine address to physical address */
static unsigned long m2p(phys_addr_t maddr)
{
	phys_addr_t paddr;

	maddr &= PTE_PFN_MASK;
	paddr = mfn_to_pfn(maddr >> PAGE_SHIFT) << PAGE_SHIFT;

	return paddr;
}

/* Convert a machine address to kernel virtual */
static void *m2v(phys_addr_t maddr)
{
	return __ka(m2p(maddr));
}

/* Set the page permissions on an identity-mapped pages */
static void set_page_prot(void *addr, pgprot_t prot)
{
	unsigned long pfn = __pa(addr) >> PAGE_SHIFT;
	pte_t pte = pfn_pte(pfn, prot);

	/* recall for PVH, page tables are native. */
	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	if (HYPERVISOR_update_va_mapping((unsigned long)addr, pte, 0))
		BUG();
}

static void __init xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
{
	unsigned pmdidx, pteidx;
	unsigned ident_pte;
	unsigned long pfn;

	level1_ident_pgt = extend_brk(sizeof(pte_t) * LEVEL1_IDENT_ENTRIES,
				      PAGE_SIZE);

	ident_pte = 0;
	pfn = 0;
	for (pmdidx = 0; pmdidx < PTRS_PER_PMD && pfn < max_pfn; pmdidx++) {
		pte_t *pte_page;

		/* Reuse or allocate a page of ptes */
		if (pmd_present(pmd[pmdidx]))
			pte_page = m2v(pmd[pmdidx].pmd);
		else {
			/* Check for free pte pages */
			if (ident_pte == LEVEL1_IDENT_ENTRIES)
				break;

			pte_page = &level1_ident_pgt[ident_pte];
			ident_pte += PTRS_PER_PTE;

			pmd[pmdidx] = __pmd(__pa(pte_page) | _PAGE_TABLE);
		}

		/* Install mappings */
		for (pteidx = 0; pteidx < PTRS_PER_PTE; pteidx++, pfn++) {
			pte_t pte;

#ifdef CONFIG_X86_32
			if (pfn > max_pfn_mapped)
				max_pfn_mapped = pfn;
#endif

			if (!pte_none(pte_page[pteidx]))
				continue;

			pte = pfn_pte(pfn, PAGE_KERNEL_EXEC);
			pte_page[pteidx] = pte;
		}
	}

	for (pteidx = 0; pteidx < ident_pte; pteidx += PTRS_PER_PTE)
		set_page_prot(&level1_ident_pgt[pteidx], PAGE_KERNEL_RO);

	set_page_prot(pmd, PAGE_KERNEL_RO);
}

void __init xen_setup_machphys_mapping(void)
{
	struct xen_machphys_mapping mapping;

	if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
		machine_to_phys_mapping = (unsigned long *)mapping.v_start;
		machine_to_phys_nr = mapping.max_mfn + 1;
	} else {
		machine_to_phys_nr = MACH2PHYS_NR_ENTRIES;
	}
#ifdef CONFIG_X86_32
	WARN_ON((machine_to_phys_mapping + (machine_to_phys_nr - 1))
		< machine_to_phys_mapping);
#endif
}

#ifdef CONFIG_X86_64
static void convert_pfn_mfn(void *v)
{
	pte_t *pte = v;
	int i;

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	/* All levels are converted the same way, so just treat them
	   as ptes. */
	for (i = 0; i < PTRS_PER_PTE; i++)
		pte[i] = xen_make_pte(pte[i].pte);
}

/*
 * Set up the initial kernel pagetable.
 *
 * We can construct this by grafting the Xen provided pagetable into
 * head_64.S's preconstructed pagetables.  We copy the Xen L2's into
 * level2_ident_pgt, level2_kernel_pgt and level2_fixmap_pgt.  This
 * means that only the kernel has a physical mapping to start with -
 * but that's enough to get __va working.  We need to fill in the rest
 * of the physical mapping once some sort of allocator has been set
 * up.
 * NOTE: for PVH, the page tables are native.
 */
pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
					 unsigned long max_pfn)
{
	pud_t *l3;
	pmd_t *l2;

	/* max_pfn_mapped is the last pfn mapped in the initial memory
	 * mappings. Considering that on Xen after the kernel mappings we
	 * have the mappings of some pages that don't exist in pfn space, we
	 * set max_pfn_mapped to the last real pfn mapped. */
	max_pfn_mapped = PFN_DOWN(__pa(xen_start_info->mfn_list));

	/* Zap identity mapping */
	init_level4_pgt[0] = __pgd(0);

	/* Pre-constructed entries are in pfn, so convert to mfn */
	convert_pfn_mfn(init_level4_pgt);
	convert_pfn_mfn(level3_ident_pgt);
	convert_pfn_mfn(level3_kernel_pgt);

	l3 = m2v(pgd[pgd_index(__START_KERNEL_map)].pgd);
	l2 = m2v(l3[pud_index(__START_KERNEL_map)].pud);

	memcpy(level2_ident_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
	memcpy(level2_kernel_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);

	l3 = m2v(pgd[pgd_index(__START_KERNEL_map + PMD_SIZE)].pgd);
	l2 = m2v(l3[pud_index(__START_KERNEL_map + PMD_SIZE)].pud);
	memcpy(level2_fixmap_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);

	/* Set up identity map */
	xen_map_identity_early(level2_ident_pgt, max_pfn);

	/* Make pagetable pieces RO */
	set_page_prot(init_level4_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_ident_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_kernel_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_user_vsyscall, PAGE_KERNEL_RO);
	set_page_prot(level2_kernel_pgt, PAGE_KERNEL_RO);
	set_page_prot(level2_fixmap_pgt, PAGE_KERNEL_RO);

	/* Pin down new L4 */
	pin_pagetable_pfn(MMUEXT_PIN_L4_TABLE,
		  	PFN_DOWN(__pa_symbol(init_level4_pgt)));

	/* Unpin Xen-provided one */
	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	/* Switch over */
	pgd = init_level4_pgt;

	/*
	 * At this stage there can be no user pgd, and no page
	 * structure to attach it to, so make sure we just set kernel
	 * pgd.
	 */
	if (xen_feature(XENFEAT_writable_page_tables)) {
		native_write_cr3(__pa(pgd));
	} else {
		xen_mc_batch();
		__xen_write_cr3(true, __pa(pgd));
		xen_mc_issue(PARAVIRT_LAZY_CPU);
	}

	memblock_reserve(__pa(xen_start_info->pt_base),
			 xen_start_info->nr_pt_frames * PAGE_SIZE);

	return pgd;
}
#else	/* !CONFIG_X86_64 */
static RESERVE_BRK_ARRAY(pmd_t, initial_kernel_pmd, PTRS_PER_PMD);
static RESERVE_BRK_ARRAY(pmd_t, swapper_kernel_pmd, PTRS_PER_PMD);

static void __init xen_write_cr3_init(unsigned long cr3)
{
	unsigned long pfn = PFN_DOWN(__pa(swapper_pg_dir));

	BUG_ON(read_cr3() != __pa(initial_page_table));
	BUG_ON(cr3 != __pa(swapper_pg_dir));

	/*
	 * We are switching to swapper_pg_dir for the first time (from
	 * initial_page_table) and therefore need to mark that page
	 * read-only and then pin it.
	 *
	 * Xen disallows sharing of kernel PMDs for PAE
	 * guests. Therefore we must copy the kernel PMD from
	 * initial_page_table into a new kernel PMD to be used in
	 * swapper_pg_dir.
	 */
	swapper_kernel_pmd =
		extend_brk(sizeof(pmd_t) * PTRS_PER_PMD, PAGE_SIZE);
	memcpy(swapper_kernel_pmd, initial_kernel_pmd,
	       sizeof(pmd_t) * PTRS_PER_PMD);
	swapper_pg_dir[KERNEL_PGD_BOUNDARY] =
		__pgd(__pa(swapper_kernel_pmd) | _PAGE_PRESENT);
	set_page_prot(swapper_kernel_pmd, PAGE_KERNEL_RO);

	set_page_prot(swapper_pg_dir, PAGE_KERNEL_RO);
	xen_write_cr3(cr3);
	pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE, pfn);

	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE,
			  PFN_DOWN(__pa(initial_page_table)));
	set_page_prot(initial_page_table, PAGE_KERNEL);
	set_page_prot(initial_kernel_pmd, PAGE_KERNEL);

	pv_mmu_ops.write_cr3 = &xen_write_cr3;
}

pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
					 unsigned long max_pfn)
{
	pmd_t *kernel_pmd;

	initial_kernel_pmd =
		extend_brk(sizeof(pmd_t) * PTRS_PER_PMD, PAGE_SIZE);

	max_pfn_mapped = PFN_DOWN(__pa(xen_start_info->pt_base) +
				  xen_start_info->nr_pt_frames * PAGE_SIZE +
				  512*1024);

	kernel_pmd = m2v(pgd[KERNEL_PGD_BOUNDARY].pgd);
	memcpy(initial_kernel_pmd, kernel_pmd, sizeof(pmd_t) * PTRS_PER_PMD);

	xen_map_identity_early(initial_kernel_pmd, max_pfn);

	memcpy(initial_page_table, pgd, sizeof(pgd_t) * PTRS_PER_PGD);
	initial_page_table[KERNEL_PGD_BOUNDARY] =
		__pgd(__pa(initial_kernel_pmd) | _PAGE_PRESENT);

	set_page_prot(initial_kernel_pmd, PAGE_KERNEL_RO);
	set_page_prot(initial_page_table, PAGE_KERNEL_RO);
	set_page_prot(empty_zero_page, PAGE_KERNEL_RO);

	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE,
			  PFN_DOWN(__pa(initial_page_table)));
	xen_write_cr3(__pa(initial_page_table));

	memblock_reserve(__pa(xen_start_info->pt_base),
			 xen_start_info->nr_pt_frames * PAGE_SIZE);

	return initial_page_table;
}
#endif	/* CONFIG_X86_64 */

static unsigned char dummy_mapping[PAGE_SIZE] __page_aligned_bss;

static void xen_set_fixmap(unsigned idx, phys_addr_t phys, pgprot_t prot)
{
	pte_t pte;

	phys >>= PAGE_SHIFT;

	switch (idx) {
	case FIX_BTMAP_END ... FIX_BTMAP_BEGIN:
#ifdef CONFIG_X86_F00F_BUG
	case FIX_F00F_IDT:
#endif
#ifdef CONFIG_X86_32
	case FIX_WP_TEST:
	case FIX_VDSO:
# ifdef CONFIG_HIGHMEM
	case FIX_KMAP_BEGIN ... FIX_KMAP_END:
# endif
#else
	case VSYSCALL_LAST_PAGE ... VSYSCALL_FIRST_PAGE:
	case VVAR_PAGE:
#endif
	case FIX_TEXT_POKE0:
	case FIX_TEXT_POKE1:
		/* All local page mappings */
		pte = pfn_pte(phys, prot);
		break;

#ifdef CONFIG_X86_LOCAL_APIC
	case FIX_APIC_BASE:	/* maps dummy local APIC */
		pte = pfn_pte(PFN_DOWN(__pa(dummy_mapping)), PAGE_KERNEL);
		break;
#endif

#ifdef CONFIG_X86_IO_APIC
	case FIX_IO_APIC_BASE_0 ... FIX_IO_APIC_BASE_END:
		/*
		 * We just don't map the IO APIC - all access is via
		 * hypercalls.  Keep the address in the pte for reference.
		 */
		pte = pfn_pte(PFN_DOWN(__pa(dummy_mapping)), PAGE_KERNEL);
		break;
#endif

	case FIX_PARAVIRT_BOOTMAP:
		/* This is an MFN, but it isn't an IO mapping from the
		   IO domain */
		pte = mfn_pte(phys, prot);
		break;

	default:
		/* By default, set_fixmap is used for hardware mappings */
		pte = mfn_pte(phys, __pgprot(pgprot_val(prot) | _PAGE_IOMAP));
		break;
	}

	__native_set_fixmap(idx, pte);

#ifdef CONFIG_X86_64
	/* Replicate changes to map the vsyscall page into the user
	   pagetable vsyscall mapping. */
	if ((idx >= VSYSCALL_LAST_PAGE && idx <= VSYSCALL_FIRST_PAGE) ||
	    idx == VVAR_PAGE) {
		unsigned long vaddr = __fix_to_virt(idx);
		set_pte_vaddr_pud(level3_user_vsyscall, vaddr, pte);
	}
#endif
}

static void __init xen_post_allocator_init(void)
{
	pv_mmu_ops.set_pte = xen_set_pte;
	pv_mmu_ops.set_pmd = xen_set_pmd;
	pv_mmu_ops.set_pud = xen_set_pud;
#if PAGETABLE_LEVELS == 4
	pv_mmu_ops.set_pgd = xen_set_pgd;
#endif

	/* This will work as long as patching hasn't happened yet
	   (which it hasn't) */
	pv_mmu_ops.alloc_pte = xen_alloc_pte;
	pv_mmu_ops.alloc_pmd = xen_alloc_pmd;
	pv_mmu_ops.release_pte = xen_release_pte;
	pv_mmu_ops.release_pmd = xen_release_pmd;
#if PAGETABLE_LEVELS == 4
	pv_mmu_ops.alloc_pud = xen_alloc_pud;
	pv_mmu_ops.release_pud = xen_release_pud;
#endif

#ifdef CONFIG_X86_64
	SetPagePinned(virt_to_page(level3_user_vsyscall));
#endif
	xen_mark_init_mm_pinned();
}

static void xen_leave_lazy_mmu(void)
{
	preempt_disable();
	xen_mc_flush();
	paravirt_leave_lazy_mmu();
	preempt_enable();
}

static const struct pv_mmu_ops xen_mmu_ops __initconst = {
	.read_cr2 = xen_read_cr2,
	.write_cr2 = xen_write_cr2,

	.read_cr3 = xen_read_cr3,
#ifdef CONFIG_X86_32
	.write_cr3 = xen_write_cr3_init,
#else
	.write_cr3 = xen_write_cr3,
#endif

	.flush_tlb_user = xen_flush_tlb,
	.flush_tlb_kernel = xen_flush_tlb,
	.flush_tlb_single = xen_flush_tlb_single,
	.flush_tlb_others = xen_flush_tlb_others,

	.pte_update = paravirt_nop,
	.pte_update_defer = paravirt_nop,

	.pgd_alloc = xen_pgd_alloc,
	.pgd_free = xen_pgd_free,

	.alloc_pte = xen_alloc_pte_init,
	.release_pte = xen_release_pte_init,
	.alloc_pmd = xen_alloc_pmd_init,
	.release_pmd = xen_release_pmd_init,

	.set_pte = xen_set_pte_init,
	.set_pte_at = xen_set_pte_at,
	.set_pmd = xen_set_pmd_hyper,

	.ptep_modify_prot_start = __ptep_modify_prot_start,
	.ptep_modify_prot_commit = __ptep_modify_prot_commit,

	.pte_val = PV_CALLEE_SAVE(xen_pte_val),
	.pgd_val = PV_CALLEE_SAVE(xen_pgd_val),

	.make_pte = PV_CALLEE_SAVE(xen_make_pte),
	.make_pgd = PV_CALLEE_SAVE(xen_make_pgd),

#ifdef CONFIG_X86_PAE
	.set_pte_atomic = xen_set_pte_atomic,
	.pte_clear = xen_pte_clear,
	.pmd_clear = xen_pmd_clear,
#endif	/* CONFIG_X86_PAE */
	.set_pud = xen_set_pud_hyper,

	.make_pmd = PV_CALLEE_SAVE(xen_make_pmd),
	.pmd_val = PV_CALLEE_SAVE(xen_pmd_val),

#if PAGETABLE_LEVELS == 4
	.pud_val = PV_CALLEE_SAVE(xen_pud_val),
	.make_pud = PV_CALLEE_SAVE(xen_make_pud),
	.set_pgd = xen_set_pgd_hyper,

	.alloc_pud = xen_alloc_pmd_init,
	.release_pud = xen_release_pmd_init,
#endif	/* PAGETABLE_LEVELS == 4 */

	.activate_mm = xen_activate_mm,
	.dup_mmap = xen_dup_mmap,
	.exit_mmap = xen_exit_mmap,

	.lazy_mode = {
		.enter = paravirt_enter_lazy_mmu,
		.leave = xen_leave_lazy_mmu,
	},

	.set_fixmap = xen_set_fixmap,
};

void __init xen_init_mmu_ops(void)
{
	x86_init.paging.pagetable_setup_done = xen_pagetable_setup_done;

	if (xen_feature(XENFEAT_auto_translated_physmap)) {
		pv_mmu_ops.flush_tlb_others = xen_flush_tlb_others;

		/* set_pte* for PCI devices to map iomem. */
		if (xen_initial_domain()) {
			pv_mmu_ops.set_pte = native_set_pte;
			pv_mmu_ops.set_pte_at = xen_dom0pvh_set_pte_at;
		}
		return;
	}

	x86_init.mapping.pagetable_reserve = xen_mapping_pagetable_reserve;
	x86_init.paging.pagetable_setup_start = xen_pagetable_setup_start;
	pv_mmu_ops = xen_mmu_ops;

	memset(dummy_mapping, 0xff, PAGE_SIZE);
}

/* Protected by xen_reservation_lock. */
#define MAX_CONTIG_ORDER 9 /* 2MB */
static unsigned long discontig_frames[1<<MAX_CONTIG_ORDER];

#define VOID_PTE (mfn_pte(0, __pgprot(0)))
static void xen_zap_pfn_range(unsigned long vaddr, unsigned int order,
				unsigned long *in_frames,
				unsigned long *out_frames)
{
	int i;
	struct multicall_space mcs;

	xen_mc_batch();
	for (i = 0; i < (1UL<<order); i++, vaddr += PAGE_SIZE) {
		mcs = __xen_mc_entry(0);

		if (in_frames)
			in_frames[i] = virt_to_mfn(vaddr);

		MULTI_update_va_mapping(mcs.mc, vaddr, VOID_PTE, 0);
		__set_phys_to_machine(virt_to_pfn(vaddr), INVALID_P2M_ENTRY);

		if (out_frames)
			out_frames[i] = virt_to_pfn(vaddr);
	}
	xen_mc_issue(0);
}

/*
 * Update the pfn-to-mfn mappings for a virtual address range, either to
 * point to an array of mfns, or contiguously from a single starting
 * mfn.
 */
static void xen_remap_exchanged_ptes(unsigned long vaddr, int order,
				     unsigned long *mfns,
				     unsigned long first_mfn)
{
	unsigned i, limit;
	unsigned long mfn;

	xen_mc_batch();

	limit = 1u << order;
	for (i = 0; i < limit; i++, vaddr += PAGE_SIZE) {
		struct multicall_space mcs;
		unsigned flags;

		mcs = __xen_mc_entry(0);
		if (mfns)
			mfn = mfns[i];
		else
			mfn = first_mfn + i;

		if (i < (limit - 1))
			flags = 0;
		else {
			if (order == 0)
				flags = UVMF_INVLPG | UVMF_ALL;
			else
				flags = UVMF_TLB_FLUSH | UVMF_ALL;
		}

		MULTI_update_va_mapping(mcs.mc, vaddr,
				mfn_pte(mfn, PAGE_KERNEL), flags);

		set_phys_to_machine(virt_to_pfn(vaddr), mfn);
	}

	xen_mc_issue(0);
}

/*
 * Perform the hypercall to exchange a region of our pfns to point to
 * memory with the required contiguous alignment.  Takes the pfns as
 * input, and populates mfns as output.
 *
 * Returns a success code indicating whether the hypervisor was able to
 * satisfy the request or not.
 */
static int xen_exchange_memory(unsigned long extents_in, unsigned int order_in,
			       unsigned long *pfns_in,
			       unsigned long extents_out,
			       unsigned int order_out,
			       unsigned long *mfns_out,
			       unsigned int address_bits)
{
	long rc;
	int success;

	struct xen_memory_exchange exchange = {
		.in = {
			.nr_extents   = extents_in,
			.extent_order = order_in,
			.extent_start = pfns_in,
			.domid        = DOMID_SELF
		},
		.out = {
			.nr_extents   = extents_out,
			.extent_order = order_out,
			.extent_start = mfns_out,
			.address_bits = address_bits,
			.domid        = DOMID_SELF
		}
	};

	BUG_ON(extents_in << order_in != extents_out << order_out);

	rc = HYPERVISOR_memory_op(XENMEM_exchange, &exchange);
	success = (exchange.nr_exchanged == extents_in);

	BUG_ON(!success && ((exchange.nr_exchanged != 0) || (rc == 0)));
	BUG_ON(success && (rc != 0));

	return success;
}

int xen_create_contiguous_region(unsigned long vstart, unsigned int order,
				 unsigned int address_bits)
{
	unsigned long *in_frames = discontig_frames, out_frame;
	unsigned long  flags;
	int            success;

	/*
	 * Currently an auto-translated guest will not perform I/O, nor will
	 * it require PAE page directories below 4GB. Therefore any calls to
	 * this function are redundant and can be ignored.
	 */

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	if (unlikely(order > MAX_CONTIG_ORDER))
		return -ENOMEM;

	memset((void *) vstart, 0, PAGE_SIZE << order);

	spin_lock_irqsave(&xen_reservation_lock, flags);

	/* 1. Zap current PTEs, remembering MFNs. */
	xen_zap_pfn_range(vstart, order, in_frames, NULL);

	/* 2. Get a new contiguous memory extent. */
	out_frame = virt_to_pfn(vstart);
	success = xen_exchange_memory(1UL << order, 0, in_frames,
				      1, order, &out_frame,
				      address_bits);

	/* 3. Map the new extent in place of old pages. */
	if (success)
		xen_remap_exchanged_ptes(vstart, order, NULL, out_frame);
	else
		xen_remap_exchanged_ptes(vstart, order, in_frames, 0);

	spin_unlock_irqrestore(&xen_reservation_lock, flags);

	return success ? 0 : -ENOMEM;
}
EXPORT_SYMBOL_GPL(xen_create_contiguous_region);

void xen_destroy_contiguous_region(unsigned long vstart, unsigned int order)
{
	unsigned long *out_frames = discontig_frames, in_frame;
	unsigned long  flags;
	int success;

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	if (unlikely(order > MAX_CONTIG_ORDER))
		return;

	memset((void *) vstart, 0, PAGE_SIZE << order);

	spin_lock_irqsave(&xen_reservation_lock, flags);

	/* 1. Find start MFN of contiguous extent. */
	in_frame = virt_to_mfn(vstart);

	/* 2. Zap current PTEs. */
	xen_zap_pfn_range(vstart, order, NULL, out_frames);

	/* 3. Do the exchange for non-contiguous MFNs. */
	success = xen_exchange_memory(1, order, &in_frame, 1UL << order,
					0, out_frames, 0);

	/* 4. Map new pages in place of old pages. */
	if (success)
		xen_remap_exchanged_ptes(vstart, order, out_frames, 0);
	else
		xen_remap_exchanged_ptes(vstart, order, NULL, in_frame);

	spin_unlock_irqrestore(&xen_reservation_lock, flags);
}
EXPORT_SYMBOL_GPL(xen_destroy_contiguous_region);

#ifdef CONFIG_XEN_PVHVM
static void xen_hvm_exit_mmap(struct mm_struct *mm)
{
	struct xen_hvm_pagetable_dying a;
	int rc;

	a.domid = DOMID_SELF;
	a.gpa = __pa(mm->pgd);
	rc = HYPERVISOR_hvm_op(HVMOP_pagetable_dying, &a);
	WARN_ON_ONCE(rc < 0);
}

static int is_pagetable_dying_supported(void)
{
	struct xen_hvm_pagetable_dying a;
	int rc = 0;

	a.domid = DOMID_SELF;
	a.gpa = 0x00;
	rc = HYPERVISOR_hvm_op(HVMOP_pagetable_dying, &a);
	if (rc < 0) {
		printk(KERN_DEBUG "HVMOP_pagetable_dying not supported\n");
		return 0;
	}
	return 1;
}

void __init xen_hvm_init_mmu_ops(void)
{
	if (is_pagetable_dying_supported())
		pv_mmu_ops.exit_mmap = xen_hvm_exit_mmap;
}
#endif

/* Map foreign gmfn, fgmfn, to local pfn, lpfn. This for the user space 
 * creating new guest on PVH dom0 and needs to map domU pages. Called from
 * exported function, so no need to export this.
 */
static noinline int pvh_add_to_xen_p2m(unsigned long lpfn, unsigned long fgmfn,
			      unsigned int domid)
{
	int rc;
	struct xen_add_to_physmap xatp = { .foreign_domid = domid };

	xatp.gpfn = lpfn;
	xatp.idx = fgmfn;
	xatp.space = XENMAPSPACE_gmfn_foreign;
	rc = HYPERVISOR_memory_op(XENMEM_add_to_physmap, &xatp);
	if (rc)
		pr_warn("d0: Failed to map pfn to mfn rc:%d pfn:%lx mfn:%lx\n", 
			rc, lpfn, fgmfn); 
	return rc;
}

int pvh_rem_xen_p2m(unsigned long spfn, int count)
{
	struct xen_remove_from_physmap xrp;
	int i, rc;

	for (i=0; i < count; i++) {
		xrp.domid = DOMID_SELF;
		xrp.gpfn = spfn+i;
		rc = HYPERVISOR_memory_op(XENMEM_remove_from_physmap, &xrp);
		if (rc) {
			pr_warn("Failed to unmap pfn:%lx rc:%d done:%d\n",
				spfn+i, rc, i);
			return 1;
		}
	}
	return 0;
}
EXPORT_SYMBOL_GPL(pvh_rem_xen_p2m);

struct pvh_remap_data {
	unsigned long fgmfn;		/* foreign domain's gmfn */
	pgprot_t prot;
	domid_t  domid;
	struct vm_area_struct *vma;
};

static noinline int pvh_map_pte_fn(pte_t *ptep, pgtable_t token, unsigned long addr, 
			void *data)
{
	int rc;
	struct pvh_remap_data *pvhp = data;
	struct xen_pvh_sav_pfn_info *savp = pvhp->vma->vm_private_data;
	unsigned long pfn = page_to_pfn(savp->sp_paga[savp->sp_next_todo++]);
	pte_t pteval = pte_mkspecial(pfn_pte(pfn, pvhp->prot));

	if ((rc=pvh_add_to_xen_p2m(pfn, pvhp->fgmfn, pvhp->domid)))
		return rc;
	native_set_pte(ptep, pteval);
	savp->sp_addr = addr;

	return 0;
}

/* The only caller at moment passes one gmfn at a time.
 * PVH TBD/FIXME: expand this in future to honor batch requests.
 */
static noinline int pvh_remap_gmfn_range(struct vm_area_struct *vma,
				unsigned long addr, unsigned long mfn, int nr,
				pgprot_t prot, unsigned domid)
{
	int err;
	struct pvh_remap_data pvhdata;

	if (nr > 1)
		return -EINVAL;

	pvhdata.fgmfn = mfn;
	pvhdata.prot = prot;
	pvhdata.domid = domid;
	pvhdata.vma = vma;
	err = apply_to_page_range(vma->vm_mm, addr, nr << PAGE_SHIFT,
				  pvh_map_pte_fn, &pvhdata);
	flush_tlb_all();
	return err;
}

#define REMAP_BATCH_SIZE 16

struct remap_data {
	unsigned long mfn;
	pgprot_t prot;
	struct mmu_update *mmu_update;
};

static int remap_area_mfn_pte_fn(pte_t *ptep, pgtable_t token,
				 unsigned long addr, void *data)
{
	struct remap_data *rmd = data;
	pte_t pte = pte_mkspecial(pfn_pte(rmd->mfn++, rmd->prot));

	rmd->mmu_update->ptr = virt_to_machine(ptep).maddr;
	rmd->mmu_update->val = pte_val_ma(pte);
	rmd->mmu_update++;

	return 0;
}

int xen_remap_domain_mfn_range(struct vm_area_struct *vma,
			       unsigned long addr,
			       unsigned long mfn, int nr,
			       pgprot_t prot, unsigned domid)
{
	struct remap_data rmd;
	struct mmu_update mmu_update[REMAP_BATCH_SIZE];
	int batch;
	unsigned long range;
	int err = 0;

	prot = __pgprot(pgprot_val(prot) | _PAGE_IOMAP);

	BUG_ON(!((vma->vm_flags & (VM_PFNMAP | VM_RESERVED | VM_IO)) ==
				(VM_PFNMAP | VM_RESERVED | VM_IO)));

	if (xen_feature(XENFEAT_auto_translated_physmap)) {
		/* We need to update the local page tables and the xen HAP */
		return pvh_remap_gmfn_range(vma, addr, mfn, nr, prot, domid);
	}

	rmd.mfn = mfn;
	rmd.prot = prot;

	while (nr) {
		batch = min(REMAP_BATCH_SIZE, nr);
		range = (unsigned long)batch << PAGE_SHIFT;

		rmd.mmu_update = mmu_update;
		err = apply_to_page_range(vma->vm_mm, addr, range,
					  remap_area_mfn_pte_fn, &rmd);
		if (err)
			goto out;

		err = -EFAULT;
		if (HYPERVISOR_mmu_update(mmu_update, batch, NULL, domid) < 0)
			goto out;

		nr -= batch;
		addr += range;
	}

	err = 0;
out:

	flush_tlb_all();

	return err;
}
EXPORT_SYMBOL_GPL(xen_remap_domain_mfn_range);

int xen_unmap_domain_mfn_range(struct vm_area_struct *vma)
{
	struct xen_pvh_sav_pfn_info *savp = vma ? vma->vm_private_data : NULL;

	if (!savp || !xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	while (savp->sp_next_todo--) {
		unsigned long pfn;
#if 0
		unsigned int level;
		pte_t *ptep = lookup_address(savp->sp_addr, &level);

		set_pte(ptep, __pte(0));
#endif
		pfn = page_to_pfn(savp->sp_paga[savp->sp_next_todo]);
		pvh_rem_xen_p2m(pfn, 1);

		free_xenballooned_pages(1, &savp->sp_paga[savp->sp_next_todo]);
	}

	flush_tlb_all();
	return 0;
}
EXPORT_SYMBOL_GPL(xen_unmap_domain_mfn_range);

[-- Attachment #4: Type: text/plain, Size: 126 bytes --]

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

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Ian Campbell]

On Thu, 2012-09-13 at 02:19 +0100, Mukesh Rathor wrote:
> On Tue, 11 Sep 2012 15:10:23 +0100
> Ian Campbell <Ian.Campbell@citrix.com> wrote:
> 
> > I think you can't rely on the implicit teardown here since you need to
> > unmap before you hand the page back to the balloon. The reason this
> > doesn't look necessary now is that you don't give the page back.
> > Also not ordering the stage 1 and stage 2 teardown correctly is
> > dangerous, depending on the eventual ordering you potentially turn an
> > erroneous access to a virtual address, which should result in a guest
> > OS level page fault (and e.g. a seg fault to the offending process)
> > into a hypervisor shoots the guest due to an unexpected stage 2 fault
> > type failure, which is somewhat undesirable ;-)
> > 
> > With that in mind I think you do in the end need to add
> > xen_unmap_domain_mfn_range which does the unmap from both stage 1 and
> > stage 2 -- that balances out the interface (making pvh_rem_xen_p2m
> > internal) too, which is nice. This function may turn out to be a nop
> > on !pvh, but that's ok (although maybe there would be no harm in doing
> > explicit unmaps, for consistency?).
>  
> Ok, I added xen_unmap_domain_mfn_range(). Take a look.

Thanks, I'll take a gander once I get ballooning working on ARM.

>  It appears that
> by the time privcmd_close() is called, the kernel has already freed 
> process resources and lookup_address() returns NULL. Now I am wondering
> if the kernel/mmu does anything to the page while shooting the pte
> entry. Well the page was orig from balloon, so the cleanup hopefully
> leaves it alone.

I suppose it depends on whether the core "takes over" the reference
which you hold. I think it doesn't, so this is just a leak, rather than
putting a ballooned page back into the general allocation pool (things
would be crashing left & right if it was doing this I reckon)

> 
> I had looked for other hooks initially when I did this, but 
> vm_operations_struct->close was the only one to pan out.
> 
> I can't really move pvh_privcmd_resv_pfns to mmu.c because the 
> xen_remap_domain_mfn_range is called one page at a time, and I need
> to allocate the array first. I'd have to change it to linked list, worth
> it? Or I'd have to move and export it.

Another alternative would be to add the page array as a parameter to the
map/unmap function, rather than relying on it propagating via
vma_private.

The other option I can see is for privcmd to use traverse_pages to hook
all the struct pages in at the right virtual address and then have
remap_mfn_range do a walk for each page. That's O(N) walks for each
mapping though, unless perhaps apply_to_page_range gives the callback
something which can be turned back into a struct pag or a pfn?

> > WRT passing data between interfaces in vma->vm_private, which is
> > pretty subtle, can we push that whole thing down into
> > xen_{remap,unmap}_domain_mfn_range too? This would make the
> > requirement on the caller be simple "never use vm_private", as
> > opposed to now where the requirement is "sometimes you might have to
> > allocate some stuff and stick it in here". The downside is that it
> > pushes code which could be generic down into per-arch stuff, although
> > with suitable generic helper functions this isn't so bad (whatever
> > happened to {alloc,free}_empty_pages_and_pagevec from the classic
> > kernels? Those did exactly what we want here, I think)
> 
> Well, it has to hang off of vma->vm_private. The alternative would be to
> have a hash table by process id or something, again not sure if worth it. 

I think using vm_private within a subsystem/layer is ok, what I think we
should avoid is having layers pass data back and forth in that field.

> 
> Take a look at my latest files attached. Now the alloc_balloon and free
> are split between privcmd and mmu.c. The alternative would be to call 
> xen_unmap_domain_mfn_range one pfn at a time and have it call 
> pvh_rem_xen_p2m(), and move free_xenballooned_pages to privcmd. But
> that would be same as just changing the name of pvh_rem_xen_p2m to
> xen_unmap_domain_mfn_range(). Also, remap and unmap won't be much
> symmetric then.
> 
> Not sure if there's a lot we could do here to be honest. LMK what you 
> think.
> 
> thanks,
> Mukesh
> 

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

> I suppose it depends on whether the core "takes over" the reference
> which you hold. I think it doesn't, so this is just a leak, rather
> than putting a ballooned page back into the general allocation pool
> (things would be crashing left & right if it was doing this I reckon)
> 
> > 
> > I had looked for other hooks initially when I did this, but 
> > vm_operations_struct->close was the only one to pan out.
> > 
> > I can't really move pvh_privcmd_resv_pfns to mmu.c because the 
> > xen_remap_domain_mfn_range is called one page at a time, and I need
> > to allocate the array first. I'd have to change it to linked list,
> > worth it? Or I'd have to move and export it.
> 
> Another alternative would be to add the page array as a parameter to
> the map/unmap function, rather than relying on it propagating via
> vma_private.

I thought it was a no-no to change an exported API. Konrad, is it OK
to change an exported API like xen_remap_domain_mfn_range, I mean, are
there any guidelines?

thanks,
Mukesh

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Ian Campbell]

On Thu, 2012-09-13 at 19:27 +0100, Mukesh Rathor wrote:
> > I suppose it depends on whether the core "takes over" the reference
> > which you hold. I think it doesn't, so this is just a leak, rather
> > than putting a ballooned page back into the general allocation pool
> > (things would be crashing left & right if it was doing this I reckon)
> > 
> > > 
> > > I had looked for other hooks initially when I did this, but 
> > > vm_operations_struct->close was the only one to pan out.
> > > 
> > > I can't really move pvh_privcmd_resv_pfns to mmu.c because the 
> > > xen_remap_domain_mfn_range is called one page at a time, and I need
> > > to allocate the array first. I'd have to change it to linked list,
> > > worth it? Or I'd have to move and export it.
> > 
> > Another alternative would be to add the page array as a parameter to
> > the map/unmap function, rather than relying on it propagating via
> > vma_private.
> 
> I thought it was a no-no to change an exported API. Konrad, is it OK
> to change an exported API like xen_remap_domain_mfn_range, I mean, are
> there any guidelines?

You are thinking of system calls, which cannot be changed.

Kernel internal APIs, including those exported to modules are fair game
to change. See Documentation/stable_api_nonsense.txt.

Ian.

> 
> thanks,
> Mukesh
> 
> 

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Konrad Rzeszutek Wilk]

On Thu, Sep 13, 2012 at 11:27:53AM -0700, Mukesh Rathor wrote:
> > I suppose it depends on whether the core "takes over" the reference
> > which you hold. I think it doesn't, so this is just a leak, rather
> > than putting a ballooned page back into the general allocation pool
> > (things would be crashing left & right if it was doing this I reckon)
> > 
> > > 
> > > I had looked for other hooks initially when I did this, but 
> > > vm_operations_struct->close was the only one to pan out.
> > > 
> > > I can't really move pvh_privcmd_resv_pfns to mmu.c because the 
> > > xen_remap_domain_mfn_range is called one page at a time, and I need
> > > to allocate the array first. I'd have to change it to linked list,
> > > worth it? Or I'd have to move and export it.
> > 
> > Another alternative would be to add the page array as a parameter to
> > the map/unmap function, rather than relying on it propagating via
> > vma_private.
> 
> I thought it was a no-no to change an exported API. Konrad, is it OK
> to change an exported API like xen_remap_domain_mfn_range, I mean, are
> there any guidelines?

Make it work right. We (upstream) don't care about APIs stability. Only
distros need to do it.
> 
> thanks,
> Mukesh
> 
> 

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

On Thu, 13 Sep 2012 12:37:46 +0100
Ian Campbell <Ian.Campbell@citrix.com> wrote:

> On Thu, 2012-09-13 at 02:19 +0100, Mukesh Rathor wrote:
> > On Tue, 11 Sep 2012 15:10:23 +0100
> > Ian Campbell <Ian.Campbell@citrix.com> wrote:
> > 
> > 
> > Well, it has to hang off of vma->vm_private. The alternative would
> > be to have a hash table by process id or something, again not sure
> > if worth it. 
> 
> I think using vm_private within a subsystem/layer is ok, what I think
> we should avoid is having layers pass data back and forth in that
> field.

Ah I see your point. Ok, let me play around a bit see what I can do.

thanks
Mukesh

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Mukesh Rathor]

[-- Attachment #1: Type: text/plain, Size: 798 bytes --]

On Thu, 13 Sep 2012 17:34:20 -0700
Mukesh Rathor <mukesh.rathor@oracle.com> wrote:

> On Thu, 13 Sep 2012 12:37:46 +0100
> Ian Campbell <Ian.Campbell@citrix.com> wrote:
> 
> > On Thu, 2012-09-13 at 02:19 +0100, Mukesh Rathor wrote:
> > > On Tue, 11 Sep 2012 15:10:23 +0100
> > > 
> > I think using vm_private within a subsystem/layer is ok, what I
> > think we should avoid is having layers pass data back and forth in
> > that field.
> 
> Ah I see your point. Ok, let me play around a bit see what I can do.

Hey Ian,

I played around a bit with privcmd, but not a whole lot can be done. I
did change things around a bit to make the APIs more symmetric and hide
vm_private_data from mmu.c. Please take a look. Unless major objections,
I'd like to resubmit all linux patches asap.

thanks,
Mukesh


[-- Attachment #2: privcmd.c --]
[-- Type: text/x-c++src, Size: 10775 bytes --]

/******************************************************************************
 * privcmd.c
 *
 * Interface to privileged domain-0 commands.
 *
 * Copyright (c) 2002-2004, K A Fraser, B Dragovic
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include <linux/uaccess.h>
#include <linux/swap.h>
#include <linux/highmem.h>
#include <linux/pagemap.h>
#include <linux/seq_file.h>
#include <linux/miscdevice.h>

#include <asm/pgalloc.h>
#include <asm/pgtable.h>
#include <asm/tlb.h>
#include <asm/xen/hypervisor.h>
#include <asm/xen/hypercall.h>

#include <xen/xen.h>
#include <xen/privcmd.h>
#include <xen/interface/xen.h>
#include <xen/features.h>
#include <xen/page.h>
#include <xen/xen-ops.h>
#include <xen/balloon.h>

#include "privcmd.h"

MODULE_LICENSE("GPL");

#ifndef HAVE_ARCH_PRIVCMD_MMAP
static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma);
#endif

static long privcmd_ioctl_hypercall(void __user *udata)
{
	struct privcmd_hypercall hypercall;
	long ret;

	if (copy_from_user(&hypercall, udata, sizeof(hypercall)))
		return -EFAULT;

	ret = privcmd_call(hypercall.op,
			   hypercall.arg[0], hypercall.arg[1],
			   hypercall.arg[2], hypercall.arg[3],
			   hypercall.arg[4]);

	return ret;
}

static void free_page_list(struct list_head *pages)
{
	struct page *p, *n;

	list_for_each_entry_safe(p, n, pages, lru)
		__free_page(p);

	INIT_LIST_HEAD(pages);
}

/*
 * Given an array of items in userspace, return a list of pages
 * containing the data.  If copying fails, either because of memory
 * allocation failure or a problem reading user memory, return an
 * error code; its up to the caller to dispose of any partial list.
 */
static int gather_array(struct list_head *pagelist,
			unsigned nelem, size_t size,
			void __user *data)
{
	unsigned pageidx;
	void *pagedata;
	int ret;

	if (size > PAGE_SIZE)
		return 0;

	pageidx = PAGE_SIZE;
	pagedata = NULL;	/* quiet, gcc */
	while (nelem--) {
		if (pageidx > PAGE_SIZE-size) {
			struct page *page = alloc_page(GFP_KERNEL);

			ret = -ENOMEM;
			if (page == NULL)
				goto fail;

			pagedata = page_address(page);

			list_add_tail(&page->lru, pagelist);
			pageidx = 0;
		}

		ret = -EFAULT;
		if (copy_from_user(pagedata + pageidx, data, size))
			goto fail;

		data += size;
		pageidx += size;
	}

	ret = 0;

fail:
	return ret;
}

/*
 * Call function "fn" on each element of the array fragmented
 * over a list of pages.
 */
static int traverse_pages(unsigned nelem, size_t size,
			  struct list_head *pos,
			  int (*fn)(void *data, void *state),
			  void *state)
{
	void *pagedata;
	unsigned pageidx;
	int ret = 0;

	BUG_ON(size > PAGE_SIZE);

	pageidx = PAGE_SIZE;
	pagedata = NULL;	/* hush, gcc */

	while (nelem--) {
		if (pageidx > PAGE_SIZE-size) {
			struct page *page;
			pos = pos->next;
			page = list_entry(pos, struct page, lru);
			pagedata = page_address(page);
			pageidx = 0;
		}

		ret = (*fn)(pagedata + pageidx, state);
		if (ret)
			break;
		pageidx += size;
	}

	return ret;
}

struct mmap_mfn_state {
	unsigned long va;
	struct vm_area_struct *vma;
	domid_t domain;
};

static int mmap_mfn_range(void *data, void *state)
{
	struct privcmd_mmap_entry *msg = data;
	struct mmap_mfn_state *st = state;
	struct vm_area_struct *vma = st->vma;
	int rc;

	/* Do not allow range to wrap the address space. */
	if ((msg->npages > (LONG_MAX >> PAGE_SHIFT)) ||
	    ((unsigned long)(msg->npages << PAGE_SHIFT) >= -st->va))
		return -EINVAL;

	/* Range chunks must be contiguous in va space. */
	if ((msg->va != st->va) ||
	    ((msg->va+(msg->npages<<PAGE_SHIFT)) > vma->vm_end))
		return -EINVAL;

	rc = xen_remap_domain_mfn_range(vma,
					msg->va & PAGE_MASK,
					msg->mfn, msg->npages,
					vma->vm_page_prot,
					st->domain, NULL);
	if (rc < 0)
		return rc;

	st->va += msg->npages << PAGE_SHIFT;

	return 0;
}

static long privcmd_ioctl_mmap(void __user *udata)
{
	struct privcmd_mmap mmapcmd;
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	int rc;
	LIST_HEAD(pagelist);
	struct mmap_mfn_state state;

	if (!xen_initial_domain())
		return -EPERM;

	/* PVH: TBD/FIXME. For now we only support privcmd_ioctl_mmap_batch */
	if (xen_pv_domain() && xen_feature(XENFEAT_auto_translated_physmap))
		return -ENOSYS;

	if (copy_from_user(&mmapcmd, udata, sizeof(mmapcmd)))
		return -EFAULT;

	rc = gather_array(&pagelist,
			  mmapcmd.num, sizeof(struct privcmd_mmap_entry),
			  mmapcmd.entry);

	if (rc || list_empty(&pagelist))
		goto out;

	down_write(&mm->mmap_sem);

	{
		struct page *page = list_first_entry(&pagelist,
						     struct page, lru);
		struct privcmd_mmap_entry *msg = page_address(page);

		vma = find_vma(mm, msg->va);
		rc = -EINVAL;

		if (!vma || (msg->va != vma->vm_start) ||
		    !privcmd_enforce_singleshot_mapping(vma))
			goto out_up;
	}

	state.va = vma->vm_start;
	state.vma = vma;
	state.domain = mmapcmd.dom;

	rc = traverse_pages(mmapcmd.num, sizeof(struct privcmd_mmap_entry),
			    &pagelist,
			    mmap_mfn_range, &state);


out_up:
	up_write(&mm->mmap_sem);

out:
	free_page_list(&pagelist);

	return rc;
}

struct mmap_batch_state {
	domid_t domain;
	unsigned long va;
	struct vm_area_struct *vma;
	int err;

	xen_pfn_t __user *user;
};

/* PVH dom0 fyi: if domU being created is PV, then mfn is mfn(addr on bus). If 
 * it's PVH then mfn is pfn (input to HAP). */
static int mmap_batch_fn(void *data, void *state)
{
	xen_pfn_t *mfnp = data;
	struct mmap_batch_state *st = state;
	struct xen_pvh_pfn_info *pvhp = st->vma ? st->vma->vm_private_data 
						: NULL;

	if (xen_remap_domain_mfn_range(st->vma, st->va & PAGE_MASK, *mfnp, 1,
				       st->vma->vm_page_prot, st->domain, 
				       pvhp) < 0) {
		*mfnp |= 0xf0000000U;
		st->err++;
	}
	st->va += PAGE_SIZE;

	return 0;
}

static int mmap_return_errors(void *data, void *state)
{
	xen_pfn_t *mfnp = data;
	struct mmap_batch_state *st = state;

	return put_user(*mfnp, st->user++);
}

/* Allocate pfns that are then mapped with gmfns from foreign domid. Update
 * the vma with the page info to use later.
 * Returns: 0 if success, otherwise -errno
 */ 
static int pvh_privcmd_resv_pfns(struct vm_area_struct *vma, int numpgs)
{
	int rc;
	struct xen_pvh_pfn_info *pvhp;

	pvhp = kzalloc(sizeof(struct xen_pvh_pfn_info), GFP_KERNEL);
	if (pvhp == NULL)
		return -ENOMEM;

	pvhp->pi_paga = kcalloc(numpgs, sizeof(pvhp->pi_paga[0]), GFP_KERNEL);
	if (pvhp->pi_paga == NULL) {
		kfree(pvhp);
		return -ENOMEM;
	}

	rc = alloc_xenballooned_pages(numpgs, pvhp->pi_paga, 0);
	if (rc != 0) {
		pr_warn("%s Could not alloc %d pfns rc:%d\n", __FUNCTION__, 
			numpgs, rc);
		kfree(pvhp->pi_paga);
		kfree(pvhp);
		return -ENOMEM;
	}
	pvhp->pi_num_pgs = numpgs;
	BUG_ON(vma->vm_private_data != (void *)1);
	vma->vm_private_data = pvhp;

	return 0;
}

static struct vm_operations_struct privcmd_vm_ops;

static long privcmd_ioctl_mmap_batch(void __user *udata)
{
	int ret;
	struct privcmd_mmapbatch m;
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	unsigned long nr_pages;
	LIST_HEAD(pagelist);
	struct mmap_batch_state state;

	if (!xen_initial_domain())
		return -EPERM;

	if (copy_from_user(&m, udata, sizeof(m)))
		return -EFAULT;

	nr_pages = m.num;
	if ((m.num <= 0) || (nr_pages > (LONG_MAX >> PAGE_SHIFT)))
		return -EINVAL;

	ret = gather_array(&pagelist, m.num, sizeof(xen_pfn_t),
			   m.arr);

	if (ret || list_empty(&pagelist))
		goto out;

	down_write(&mm->mmap_sem);

	vma = find_vma(mm, m.addr);
	ret = -EINVAL;
	if (!vma ||
	    vma->vm_ops != &privcmd_vm_ops ||
	    (m.addr != vma->vm_start) ||
	    ((m.addr + (nr_pages << PAGE_SHIFT)) != vma->vm_end) ||
	    !privcmd_enforce_singleshot_mapping(vma)) {
		up_write(&mm->mmap_sem);
		goto out;
	}

	if (xen_pv_domain() && xen_feature(XENFEAT_auto_translated_physmap)) {
		if ((ret=pvh_privcmd_resv_pfns(vma, m.num))) {
			up_write(&mm->mmap_sem);
			goto out;
		}
	}
	state.domain = m.dom;
	state.vma = vma;
	state.va = m.addr;
	state.err = 0;

	ret = traverse_pages(m.num, sizeof(xen_pfn_t),
			     &pagelist, mmap_batch_fn, &state);

	up_write(&mm->mmap_sem);

	if (state.err > 0) {
		state.user = m.arr;
		ret = traverse_pages(m.num, sizeof(xen_pfn_t),
			       &pagelist,
			       mmap_return_errors, &state);
	}

out:
	free_page_list(&pagelist);

	return ret;
}

static long privcmd_ioctl(struct file *file,
			  unsigned int cmd, unsigned long data)
{
	int ret = -ENOSYS;
	void __user *udata = (void __user *) data;

	switch (cmd) {
	case IOCTL_PRIVCMD_HYPERCALL:
		ret = privcmd_ioctl_hypercall(udata);
		break;

	case IOCTL_PRIVCMD_MMAP:
		ret = privcmd_ioctl_mmap(udata);
		break;

	case IOCTL_PRIVCMD_MMAPBATCH:
		ret = privcmd_ioctl_mmap_batch(udata);
		break;

	default:
		ret = -EINVAL;
		break;
	}

	return ret;
}

static void privcmd_close(struct vm_area_struct *vma)
{
	int count;
	struct xen_pvh_pfn_info *pvhp = vma ? vma->vm_private_data : NULL;

	if (!xen_pv_domain()  || !pvhp ||
	    !xen_feature(XENFEAT_auto_translated_physmap))
		return;

	count = xen_unmap_domain_mfn_range(vma, pvhp);
	while (count--)
		free_xenballooned_pages(1, &pvhp->pi_paga[count]);
	kfree(pvhp->pi_paga);
	kfree(pvhp);
}

static int privcmd_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
	printk(KERN_DEBUG "privcmd_fault: vma=%p %lx-%lx, pgoff=%lx, uv=%p\n",
	       vma, vma->vm_start, vma->vm_end,
	       vmf->pgoff, vmf->virtual_address);

	return VM_FAULT_SIGBUS;
}

static struct vm_operations_struct privcmd_vm_ops = {
	.close = privcmd_close,
	.fault = privcmd_fault
};

static int privcmd_mmap(struct file *file, struct vm_area_struct *vma)
{
	/* DONTCOPY is essential for Xen because copy_page_range doesn't know
	 * how to recreate these mappings */
	vma->vm_flags |= VM_RESERVED | VM_IO | VM_DONTCOPY | VM_PFNMAP;
	vma->vm_ops = &privcmd_vm_ops;
	vma->vm_private_data = NULL;

	return 0;
}

static int privcmd_enforce_singleshot_mapping(struct vm_area_struct *vma)
{
	return (xchg(&vma->vm_private_data, (void *)1) == NULL);
}

const struct file_operations xen_privcmd_fops = {
	.owner = THIS_MODULE,
	.unlocked_ioctl = privcmd_ioctl,
	.mmap = privcmd_mmap,
};
EXPORT_SYMBOL_GPL(xen_privcmd_fops);

static struct miscdevice privcmd_dev = {
	.minor = MISC_DYNAMIC_MINOR,
	.name = "xen/privcmd",
	.fops = &xen_privcmd_fops,
};

static int __init privcmd_init(void)
{
	int err;

	if (!xen_domain())
		return -ENODEV;

	err = misc_register(&privcmd_dev);
	if (err != 0) {
		printk(KERN_ERR "Could not register Xen privcmd device\n");
		return err;
	}
	return 0;
}

static void __exit privcmd_exit(void)
{
	misc_deregister(&privcmd_dev);
}

module_init(privcmd_init);
module_exit(privcmd_exit);

[-- Attachment #3: mmu.c --]
[-- Type: text/x-c++src, Size: 63836 bytes --]

/*
 * Xen mmu operations
 *
 * This file contains the various mmu fetch and update operations.
 * The most important job they must perform is the mapping between the
 * domain's pfn and the overall machine mfns.
 *
 * Xen allows guests to directly update the pagetable, in a controlled
 * fashion.  In other words, the guest modifies the same pagetable
 * that the CPU actually uses, which eliminates the overhead of having
 * a separate shadow pagetable.
 *
 * In order to allow this, it falls on the guest domain to map its
 * notion of a "physical" pfn - which is just a domain-local linear
 * address - into a real "machine address" which the CPU's MMU can
 * use.
 *
 * A pgd_t/pmd_t/pte_t will typically contain an mfn, and so can be
 * inserted directly into the pagetable.  When creating a new
 * pte/pmd/pgd, it converts the passed pfn into an mfn.  Conversely,
 * when reading the content back with __(pgd|pmd|pte)_val, it converts
 * the mfn back into a pfn.
 *
 * The other constraint is that all pages which make up a pagetable
 * must be mapped read-only in the guest.  This prevents uncontrolled
 * guest updates to the pagetable.  Xen strictly enforces this, and
 * will disallow any pagetable update which will end up mapping a
 * pagetable page RW, and will disallow using any writable page as a
 * pagetable.
 *
 * Naively, when loading %cr3 with the base of a new pagetable, Xen
 * would need to validate the whole pagetable before going on.
 * Naturally, this is quite slow.  The solution is to "pin" a
 * pagetable, which enforces all the constraints on the pagetable even
 * when it is not actively in use.  This menas that Xen can be assured
 * that it is still valid when you do load it into %cr3, and doesn't
 * need to revalidate it.
 *
 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
 */
#include <linux/sched.h>
#include <linux/highmem.h>
#include <linux/debugfs.h>
#include <linux/bug.h>
#include <linux/vmalloc.h>
#include <linux/module.h>
#include <linux/gfp.h>
#include <linux/memblock.h>
#include <linux/seq_file.h>

#include <trace/events/xen.h>

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/fixmap.h>
#include <asm/mmu_context.h>
#include <asm/setup.h>
#include <asm/paravirt.h>
#include <asm/e820.h>
#include <asm/linkage.h>
#include <asm/page.h>
#include <asm/init.h>
#include <asm/pat.h>
#include <asm/smp.h>

#include <asm/xen/hypercall.h>
#include <asm/xen/hypervisor.h>

#include <xen/xen.h>
#include <xen/page.h>
#include <xen/interface/xen.h>
#include <xen/interface/hvm/hvm_op.h>
#include <xen/interface/version.h>
#include <xen/interface/memory.h>
#include <xen/hvc-console.h>
#include <xen/balloon.h>

#include "multicalls.h"
#include "mmu.h"
#include "debugfs.h"

/*
 * Protects atomic reservation decrease/increase against concurrent increases.
 * Also protects non-atomic updates of current_pages and balloon lists.
 */
DEFINE_SPINLOCK(xen_reservation_lock);

/*
 * Identity map, in addition to plain kernel map.  This needs to be
 * large enough to allocate page table pages to allocate the rest.
 * Each page can map 2MB.
 */
#define LEVEL1_IDENT_ENTRIES	(PTRS_PER_PTE * 4)
static RESERVE_BRK_ARRAY(pte_t, level1_ident_pgt, LEVEL1_IDENT_ENTRIES);

#ifdef CONFIG_X86_64
/* l3 pud for userspace vsyscall mapping */
static pud_t level3_user_vsyscall[PTRS_PER_PUD] __page_aligned_bss;
#endif /* CONFIG_X86_64 */

/*
 * Note about cr3 (pagetable base) values:
 *
 * xen_cr3 contains the current logical cr3 value; it contains the
 * last set cr3.  This may not be the current effective cr3, because
 * its update may be being lazily deferred.  However, a vcpu looking
 * at its own cr3 can use this value knowing that it everything will
 * be self-consistent.
 *
 * xen_current_cr3 contains the actual vcpu cr3; it is set once the
 * hypercall to set the vcpu cr3 is complete (so it may be a little
 * out of date, but it will never be set early).  If one vcpu is
 * looking at another vcpu's cr3 value, it should use this variable.
 */
DEFINE_PER_CPU(unsigned long, xen_cr3);	 /* cr3 stored as physaddr */
DEFINE_PER_CPU(unsigned long, xen_current_cr3);	 /* actual vcpu cr3 */


/*
 * Just beyond the highest usermode address.  STACK_TOP_MAX has a
 * redzone above it, so round it up to a PGD boundary.
 */
#define USER_LIMIT	((STACK_TOP_MAX + PGDIR_SIZE - 1) & PGDIR_MASK)

unsigned long arbitrary_virt_to_mfn(void *vaddr)
{
	xmaddr_t maddr = arbitrary_virt_to_machine(vaddr);

	return PFN_DOWN(maddr.maddr);
}

xmaddr_t arbitrary_virt_to_machine(void *vaddr)
{
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;
	pte_t *pte;
	unsigned offset;

	/*
	 * if the PFN is in the linear mapped vaddr range, we can just use
	 * the (quick) virt_to_machine() p2m lookup
	 */
	if (virt_addr_valid(vaddr))
		return virt_to_machine(vaddr);

	/* otherwise we have to do a (slower) full page-table walk */

	pte = lookup_address(address, &level);
	BUG_ON(pte == NULL);
	offset = address & ~PAGE_MASK;
	return XMADDR(((phys_addr_t)pte_mfn(*pte) << PAGE_SHIFT) + offset);
}
EXPORT_SYMBOL_GPL(arbitrary_virt_to_machine);

void make_lowmem_page_readonly(void *vaddr)
{
	pte_t *pte, ptev;
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;

	pte = lookup_address(address, &level);
	if (pte == NULL)
		return;		/* vaddr missing */

	ptev = pte_wrprotect(*pte);

	if (HYPERVISOR_update_va_mapping(address, ptev, 0))
		BUG();
}

void make_lowmem_page_readwrite(void *vaddr)
{
	pte_t *pte, ptev;
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;

	pte = lookup_address(address, &level);
	if (pte == NULL)
		return;		/* vaddr missing */

	ptev = pte_mkwrite(*pte);

	if (HYPERVISOR_update_va_mapping(address, ptev, 0))
		BUG();
}


static bool xen_page_pinned(void *ptr)
{
	struct page *page = virt_to_page(ptr);

	return PagePinned(page);
}

void xen_set_domain_pte(pte_t *ptep, pte_t pteval, unsigned domid)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	trace_xen_mmu_set_domain_pte(ptep, pteval, domid);

	mcs = xen_mc_entry(sizeof(*u));
	u = mcs.args;

	/* ptep might be kmapped when using 32-bit HIGHPTE */
	u->ptr = virt_to_machine(ptep).maddr;
	u->val = pte_val_ma(pteval);

	MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, domid);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}
EXPORT_SYMBOL_GPL(xen_set_domain_pte);

static void xen_extend_mmu_update(const struct mmu_update *update)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	mcs = xen_mc_extend_args(__HYPERVISOR_mmu_update, sizeof(*u));

	if (mcs.mc != NULL) {
		mcs.mc->args[1]++;
	} else {
		mcs = __xen_mc_entry(sizeof(*u));
		MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
	}

	u = mcs.args;
	*u = *update;
}

static void xen_extend_mmuext_op(const struct mmuext_op *op)
{
	struct multicall_space mcs;
	struct mmuext_op *u;

	mcs = xen_mc_extend_args(__HYPERVISOR_mmuext_op, sizeof(*u));

	if (mcs.mc != NULL) {
		mcs.mc->args[1]++;
	} else {
		mcs = __xen_mc_entry(sizeof(*u));
		MULTI_mmuext_op(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
	}

	u = mcs.args;
	*u = *op;
}

static void xen_set_pmd_hyper(pmd_t *ptr, pmd_t val)
{
	struct mmu_update u;

	preempt_disable();

	xen_mc_batch();

	/* ptr may be ioremapped for 64-bit pagetable setup */
	u.ptr = arbitrary_virt_to_machine(ptr).maddr;
	u.val = pmd_val_ma(val);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_set_pmd(pmd_t *ptr, pmd_t val)
{
	trace_xen_mmu_set_pmd(ptr, val);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		return;
	}

	xen_set_pmd_hyper(ptr, val);
}

/*
 * Associate a virtual page frame with a given physical page frame
 * and protection flags for that frame.
 */
void set_pte_mfn(unsigned long vaddr, unsigned long mfn, pgprot_t flags)
{
	set_pte_vaddr(vaddr, mfn_pte(mfn, flags));
}

static bool xen_batched_set_pte(pte_t *ptep, pte_t pteval)
{
	struct mmu_update u;

	if (paravirt_get_lazy_mode() != PARAVIRT_LAZY_MMU)
		return false;

	xen_mc_batch();

	u.ptr = virt_to_machine(ptep).maddr | MMU_NORMAL_PT_UPDATE;
	u.val = pte_val_ma(pteval);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	return true;
}

static inline void __xen_set_pte(pte_t *ptep, pte_t pteval)
{
	if (!xen_batched_set_pte(ptep, pteval)) {
		/*
		 * Could call native_set_pte() here and trap and
		 * emulate the PTE write but with 32-bit guests this
		 * needs two traps (one for each of the two 32-bit
		 * words in the PTE) so do one hypercall directly
		 * instead.
		 */
		struct mmu_update u;

		u.ptr = virt_to_machine(ptep).maddr | MMU_NORMAL_PT_UPDATE;
		u.val = pte_val_ma(pteval);
		HYPERVISOR_mmu_update(&u, 1, NULL, DOMID_SELF);
	}
}

static void xen_set_pte(pte_t *ptep, pte_t pteval)
{
	trace_xen_mmu_set_pte(ptep, pteval);
	__xen_set_pte(ptep, pteval);
}

void xen_set_clr_mmio_pvh_pte(unsigned long pfn, unsigned long mfn, 
			      int nr_mfns, int add_mapping)
{
	struct physdev_map_iomem iomem;

	iomem.first_gfn = pfn;
	iomem.first_mfn = mfn;
	iomem.nr_mfns = nr_mfns;
	iomem.add_mapping = add_mapping;

	if (HYPERVISOR_physdev_op(PHYSDEVOP_pvh_map_iomem, &iomem))
		BUG();
}

static void xen_dom0pvh_set_pte_at(struct mm_struct *mm, unsigned long addr,
		    		   pte_t *ptep, pte_t pteval)
{
	native_set_pte(ptep, pteval);
}

static void xen_set_pte_at(struct mm_struct *mm, unsigned long addr,
		    pte_t *ptep, pte_t pteval)
{
	trace_xen_mmu_set_pte_at(mm, addr, ptep, pteval);
	__xen_set_pte(ptep, pteval);
}

pte_t xen_ptep_modify_prot_start(struct mm_struct *mm,
				 unsigned long addr, pte_t *ptep)
{
	/* Just return the pte as-is.  We preserve the bits on commit */
	trace_xen_mmu_ptep_modify_prot_start(mm, addr, ptep, *ptep);
	return *ptep;
}

void xen_ptep_modify_prot_commit(struct mm_struct *mm, unsigned long addr,
				 pte_t *ptep, pte_t pte)
{
	struct mmu_update u;

	trace_xen_mmu_ptep_modify_prot_commit(mm, addr, ptep, pte);
	xen_mc_batch();

	u.ptr = virt_to_machine(ptep).maddr | MMU_PT_UPDATE_PRESERVE_AD;
	u.val = pte_val_ma(pte);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}

/* Assume pteval_t is equivalent to all the other *val_t types. */
static pteval_t pte_mfn_to_pfn(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long mfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		unsigned long pfn = mfn_to_pfn(mfn);

		pteval_t flags = val & PTE_FLAGS_MASK;
		if (unlikely(pfn == ~0))
			val = flags & ~_PAGE_PRESENT;
		else
			val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t pte_pfn_to_mfn(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		pteval_t flags = val & PTE_FLAGS_MASK;
		unsigned long mfn;

		if (!xen_feature(XENFEAT_auto_translated_physmap))
			mfn = get_phys_to_machine(pfn);
		else
			mfn = pfn;
		/*
		 * If there's no mfn for the pfn, then just create an
		 * empty non-present pte.  Unfortunately this loses
		 * information about the original pfn, so
		 * pte_mfn_to_pfn is asymmetric.
		 */
		if (unlikely(mfn == INVALID_P2M_ENTRY)) {
			mfn = 0;
			flags = 0;
		} else {
			/*
			 * Paramount to do this test _after_ the
			 * INVALID_P2M_ENTRY as INVALID_P2M_ENTRY &
			 * IDENTITY_FRAME_BIT resolves to true.
			 */
			mfn &= ~FOREIGN_FRAME_BIT;
			if (mfn & IDENTITY_FRAME_BIT) {
				mfn &= ~IDENTITY_FRAME_BIT;
				flags |= _PAGE_IOMAP;
			}
		}
		val = ((pteval_t)mfn << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t iomap_pte(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		pteval_t flags = val & PTE_FLAGS_MASK;

		/* We assume the pte frame number is a MFN, so
		   just use it as-is. */
		val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t xen_pte_val(pte_t pte)
{
	pteval_t pteval = pte.pte;
#if 0
	/* If this is a WC pte, convert back from Xen WC to Linux WC */
	if ((pteval & (_PAGE_PAT | _PAGE_PCD | _PAGE_PWT)) == _PAGE_PAT) {
		WARN_ON(!pat_enabled);
		pteval = (pteval & ~_PAGE_PAT) | _PAGE_PWT;
	}
#endif
	if (xen_initial_domain() && (pteval & _PAGE_IOMAP))
		return pteval;

	return pte_mfn_to_pfn(pteval);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pte_val);

static pgdval_t xen_pgd_val(pgd_t pgd)
{
	return pte_mfn_to_pfn(pgd.pgd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pgd_val);

/*
 * Xen's PAT setup is part of its ABI, though I assume entries 6 & 7
 * are reserved for now, to correspond to the Intel-reserved PAT
 * types.
 *
 * We expect Linux's PAT set as follows:
 *
 * Idx  PTE flags        Linux    Xen    Default
 * 0                     WB       WB     WB
 * 1            PWT      WC       WT     WT
 * 2        PCD          UC-      UC-    UC-
 * 3        PCD PWT      UC       UC     UC
 * 4    PAT              WB       WC     WB
 * 5    PAT     PWT      WC       WP     WT
 * 6    PAT PCD          UC-      UC     UC-
 * 7    PAT PCD PWT      UC       UC     UC
 */

void xen_set_pat(u64 pat)
{
	/* We expect Linux to use a PAT setting of
	 * UC UC- WC WB (ignoring the PAT flag) */
	WARN_ON(pat != 0x0007010600070106ull);
}

static pte_t xen_make_pte(pteval_t pte)
{
	phys_addr_t addr = (pte & PTE_PFN_MASK);
#if 0
	/* If Linux is trying to set a WC pte, then map to the Xen WC.
	 * If _PAGE_PAT is set, then it probably means it is really
	 * _PAGE_PSE, so avoid fiddling with the PAT mapping and hope
	 * things work out OK...
	 *
	 * (We should never see kernel mappings with _PAGE_PSE set,
	 * but we could see hugetlbfs mappings, I think.).
	 */
	if (pat_enabled && !WARN_ON(pte & _PAGE_PAT)) {
		if ((pte & (_PAGE_PCD | _PAGE_PWT)) == _PAGE_PWT)
			pte = (pte & ~(_PAGE_PCD | _PAGE_PWT)) | _PAGE_PAT;
	}
#endif
	/*
	 * Unprivileged domains are allowed to do IOMAPpings for
	 * PCI passthrough, but not map ISA space.  The ISA
	 * mappings are just dummy local mappings to keep other
	 * parts of the kernel happy.
	 */
	if (unlikely(pte & _PAGE_IOMAP) &&
	    (xen_initial_domain() || addr >= ISA_END_ADDRESS)) {
		pte = iomap_pte(pte);
	} else {
		pte &= ~_PAGE_IOMAP;
		pte = pte_pfn_to_mfn(pte);
	}

	return native_make_pte(pte);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pte);

static pgd_t xen_make_pgd(pgdval_t pgd)
{
	pgd = pte_pfn_to_mfn(pgd);
	return native_make_pgd(pgd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pgd);

static pmdval_t xen_pmd_val(pmd_t pmd)
{
	return pte_mfn_to_pfn(pmd.pmd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pmd_val);

static void xen_set_pud_hyper(pud_t *ptr, pud_t val)
{
	struct mmu_update u;

	preempt_disable();

	xen_mc_batch();

	/* ptr may be ioremapped for 64-bit pagetable setup */
	u.ptr = arbitrary_virt_to_machine(ptr).maddr;
	u.val = pud_val_ma(val);
	xen_extend_mmu_update(&u);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_set_pud(pud_t *ptr, pud_t val)
{
	trace_xen_mmu_set_pud(ptr, val);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		return;
	}

	xen_set_pud_hyper(ptr, val);
}

#ifdef CONFIG_X86_PAE
static void xen_set_pte_atomic(pte_t *ptep, pte_t pte)
{
	trace_xen_mmu_set_pte_atomic(ptep, pte);
	set_64bit((u64 *)ptep, native_pte_val(pte));
}

static void xen_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
{
	trace_xen_mmu_pte_clear(mm, addr, ptep);
	if (!xen_batched_set_pte(ptep, native_make_pte(0)))
		native_pte_clear(mm, addr, ptep);
}

static void xen_pmd_clear(pmd_t *pmdp)
{
	trace_xen_mmu_pmd_clear(pmdp);
	set_pmd(pmdp, __pmd(0));
}
#endif	/* CONFIG_X86_PAE */

static pmd_t xen_make_pmd(pmdval_t pmd)
{
	pmd = pte_pfn_to_mfn(pmd);
	return native_make_pmd(pmd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pmd);

#if PAGETABLE_LEVELS == 4
static pudval_t xen_pud_val(pud_t pud)
{
	return pte_mfn_to_pfn(pud.pud);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pud_val);

static pud_t xen_make_pud(pudval_t pud)
{
	pud = pte_pfn_to_mfn(pud);

	return native_make_pud(pud);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pud);

static pgd_t *xen_get_user_pgd(pgd_t *pgd)
{
	pgd_t *pgd_page = (pgd_t *)(((unsigned long)pgd) & PAGE_MASK);
	unsigned offset = pgd - pgd_page;
	pgd_t *user_ptr = NULL;

	if (offset < pgd_index(USER_LIMIT)) {
		struct page *page = virt_to_page(pgd_page);
		user_ptr = (pgd_t *)page->private;
		if (user_ptr)
			user_ptr += offset;
	}

	return user_ptr;
}

static void __xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
	struct mmu_update u;

	u.ptr = virt_to_machine(ptr).maddr;
	u.val = pgd_val_ma(val);
	xen_extend_mmu_update(&u);
}

/*
 * Raw hypercall-based set_pgd, intended for in early boot before
 * there's a page structure.  This implies:
 *  1. The only existing pagetable is the kernel's
 *  2. It is always pinned
 *  3. It has no user pagetable attached to it
 */
static void __init xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
	preempt_disable();

	xen_mc_batch();

	__xen_set_pgd_hyper(ptr, val);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_set_pgd(pgd_t *ptr, pgd_t val)
{
	pgd_t *user_ptr = xen_get_user_pgd(ptr);

	trace_xen_mmu_set_pgd(ptr, user_ptr, val);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		if (user_ptr) {
			WARN_ON(xen_page_pinned(user_ptr));
			*user_ptr = val;
		}
		return;
	}

	/* If it's pinned, then we can at least batch the kernel and
	   user updates together. */
	xen_mc_batch();

	__xen_set_pgd_hyper(ptr, val);
	if (user_ptr)
		__xen_set_pgd_hyper(user_ptr, val);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}
#endif	/* PAGETABLE_LEVELS == 4 */

/*
 * (Yet another) pagetable walker.  This one is intended for pinning a
 * pagetable.  This means that it walks a pagetable and calls the
 * callback function on each page it finds making up the page table,
 * at every level.  It walks the entire pagetable, but it only bothers
 * pinning pte pages which are below limit.  In the normal case this
 * will be STACK_TOP_MAX, but at boot we need to pin up to
 * FIXADDR_TOP.
 *
 * For 32-bit the important bit is that we don't pin beyond there,
 * because then we start getting into Xen's ptes.
 *
 * For 64-bit, we must skip the Xen hole in the middle of the address
 * space, just after the big x86-64 virtual hole.
 */
static int __xen_pgd_walk(struct mm_struct *mm, pgd_t *pgd,
			  int (*func)(struct mm_struct *mm, struct page *,
				      enum pt_level),
			  unsigned long limit)
{
	int flush = 0;
	unsigned hole_low, hole_high;
	unsigned pgdidx_limit, pudidx_limit, pmdidx_limit;
	unsigned pgdidx, pudidx, pmdidx;

	/* The limit is the last byte to be touched */
	limit--;
	BUG_ON(limit >= FIXADDR_TOP);

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	/*
	 * 64-bit has a great big hole in the middle of the address
	 * space, which contains the Xen mappings.  On 32-bit these
	 * will end up making a zero-sized hole and so is a no-op.
	 */
	hole_low = pgd_index(USER_LIMIT);
	hole_high = pgd_index(PAGE_OFFSET);

	pgdidx_limit = pgd_index(limit);
#if PTRS_PER_PUD > 1
	pudidx_limit = pud_index(limit);
#else
	pudidx_limit = 0;
#endif
#if PTRS_PER_PMD > 1
	pmdidx_limit = pmd_index(limit);
#else
	pmdidx_limit = 0;
#endif

	for (pgdidx = 0; pgdidx <= pgdidx_limit; pgdidx++) {
		pud_t *pud;

		if (pgdidx >= hole_low && pgdidx < hole_high)
			continue;

		if (!pgd_val(pgd[pgdidx]))
			continue;

		pud = pud_offset(&pgd[pgdidx], 0);

		if (PTRS_PER_PUD > 1) /* not folded */
			flush |= (*func)(mm, virt_to_page(pud), PT_PUD);

		for (pudidx = 0; pudidx < PTRS_PER_PUD; pudidx++) {
			pmd_t *pmd;

			if (pgdidx == pgdidx_limit &&
			    pudidx > pudidx_limit)
				goto out;

			if (pud_none(pud[pudidx]))
				continue;

			pmd = pmd_offset(&pud[pudidx], 0);

			if (PTRS_PER_PMD > 1) /* not folded */
				flush |= (*func)(mm, virt_to_page(pmd), PT_PMD);

			for (pmdidx = 0; pmdidx < PTRS_PER_PMD; pmdidx++) {
				struct page *pte;

				if (pgdidx == pgdidx_limit &&
				    pudidx == pudidx_limit &&
				    pmdidx > pmdidx_limit)
					goto out;

				if (pmd_none(pmd[pmdidx]))
					continue;

				pte = pmd_page(pmd[pmdidx]);
				flush |= (*func)(mm, pte, PT_PTE);
			}
		}
	}

out:
	/* Do the top level last, so that the callbacks can use it as
	   a cue to do final things like tlb flushes. */
	flush |= (*func)(mm, virt_to_page(pgd), PT_PGD);

	return flush;
}

static int xen_pgd_walk(struct mm_struct *mm,
			int (*func)(struct mm_struct *mm, struct page *,
				    enum pt_level),
			unsigned long limit)
{
	return __xen_pgd_walk(mm, mm->pgd, func, limit);
}

/* If we're using split pte locks, then take the page's lock and
   return a pointer to it.  Otherwise return NULL. */
static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm)
{
	spinlock_t *ptl = NULL;

#if USE_SPLIT_PTLOCKS
	ptl = __pte_lockptr(page);
	spin_lock_nest_lock(ptl, &mm->page_table_lock);
#endif

	return ptl;
}

static void xen_pte_unlock(void *v)
{
	spinlock_t *ptl = v;
	spin_unlock(ptl);
}

static void xen_do_pin(unsigned level, unsigned long pfn)
{
	struct mmuext_op op;

	op.cmd = level;
	op.arg1.mfn = pfn_to_mfn(pfn);

	xen_extend_mmuext_op(&op);
}

static int xen_pin_page(struct mm_struct *mm, struct page *page,
			enum pt_level level)
{
	unsigned pgfl = TestSetPagePinned(page);
	int flush;

	if (pgfl)
		flush = 0;		/* already pinned */
	else if (PageHighMem(page))
		/* kmaps need flushing if we found an unpinned
		   highpage */
		flush = 1;
	else {
		void *pt = lowmem_page_address(page);
		unsigned long pfn = page_to_pfn(page);
		struct multicall_space mcs = __xen_mc_entry(0);
		spinlock_t *ptl;

		flush = 0;

		/*
		 * We need to hold the pagetable lock between the time
		 * we make the pagetable RO and when we actually pin
		 * it.  If we don't, then other users may come in and
		 * attempt to update the pagetable by writing it,
		 * which will fail because the memory is RO but not
		 * pinned, so Xen won't do the trap'n'emulate.
		 *
		 * If we're using split pte locks, we can't hold the
		 * entire pagetable's worth of locks during the
		 * traverse, because we may wrap the preempt count (8
		 * bits).  The solution is to mark RO and pin each PTE
		 * page while holding the lock.  This means the number
		 * of locks we end up holding is never more than a
		 * batch size (~32 entries, at present).
		 *
		 * If we're not using split pte locks, we needn't pin
		 * the PTE pages independently, because we're
		 * protected by the overall pagetable lock.
		 */
		ptl = NULL;
		if (level == PT_PTE)
			ptl = xen_pte_lock(page, mm);

		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
					pfn_pte(pfn, PAGE_KERNEL_RO),
					level == PT_PGD ? UVMF_TLB_FLUSH : 0);

		if (ptl) {
			xen_do_pin(MMUEXT_PIN_L1_TABLE, pfn);

			/* Queue a deferred unlock for when this batch
			   is completed. */
			xen_mc_callback(xen_pte_unlock, ptl);
		}
	}

	return flush;
}

/* This is called just after a mm has been created, but it has not
   been used yet.  We need to make sure that its pagetable is all
   read-only, and can be pinned. */
static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd)
{
	trace_xen_mmu_pgd_pin(mm, pgd);

	xen_mc_batch();

	if (__xen_pgd_walk(mm, pgd, xen_pin_page, USER_LIMIT)) {
		/* re-enable interrupts for flushing */
		xen_mc_issue(0);

		kmap_flush_unused();

		xen_mc_batch();
	}

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(pgd);

		xen_do_pin(MMUEXT_PIN_L4_TABLE, PFN_DOWN(__pa(pgd)));

		if (user_pgd) {
			xen_pin_page(mm, virt_to_page(user_pgd), PT_PGD);
			xen_do_pin(MMUEXT_PIN_L4_TABLE,
				   PFN_DOWN(__pa(user_pgd)));
		}
	}
#else /* CONFIG_X86_32 */
#ifdef CONFIG_X86_PAE
	/* Need to make sure unshared kernel PMD is pinnable */
	xen_pin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]),
		     PT_PMD);
#endif
	xen_do_pin(MMUEXT_PIN_L3_TABLE, PFN_DOWN(__pa(pgd)));
#endif /* CONFIG_X86_64 */
	xen_mc_issue(0);
}

static void xen_pgd_pin(struct mm_struct *mm)
{
	__xen_pgd_pin(mm, mm->pgd);
}

/*
 * On save, we need to pin all pagetables to make sure they get their
 * mfns turned into pfns.  Search the list for any unpinned pgds and pin
 * them (unpinned pgds are not currently in use, probably because the
 * process is under construction or destruction).
 *
 * Expected to be called in stop_machine() ("equivalent to taking
 * every spinlock in the system"), so the locking doesn't really
 * matter all that much.
 */
void xen_mm_pin_all(void)
{
	struct page *page;

	spin_lock(&pgd_lock);

	list_for_each_entry(page, &pgd_list, lru) {
		if (!PagePinned(page)) {
			__xen_pgd_pin(&init_mm, (pgd_t *)page_address(page));
			SetPageSavePinned(page);
		}
	}

	spin_unlock(&pgd_lock);
}

/*
 * The init_mm pagetable is really pinned as soon as its created, but
 * that's before we have page structures to store the bits.  So do all
 * the book-keeping now.
 */
static int __init xen_mark_pinned(struct mm_struct *mm, struct page *page,
				  enum pt_level level)
{
	SetPagePinned(page);
	return 0;
}

static void __init xen_mark_init_mm_pinned(void)
{
	xen_pgd_walk(&init_mm, xen_mark_pinned, FIXADDR_TOP);
}

static int xen_unpin_page(struct mm_struct *mm, struct page *page,
			  enum pt_level level)
{
	unsigned pgfl = TestClearPagePinned(page);

	if (pgfl && !PageHighMem(page)) {
		void *pt = lowmem_page_address(page);
		unsigned long pfn = page_to_pfn(page);
		spinlock_t *ptl = NULL;
		struct multicall_space mcs;

		/*
		 * Do the converse to pin_page.  If we're using split
		 * pte locks, we must be holding the lock for while
		 * the pte page is unpinned but still RO to prevent
		 * concurrent updates from seeing it in this
		 * partially-pinned state.
		 */
		if (level == PT_PTE) {
			ptl = xen_pte_lock(page, mm);

			if (ptl)
				xen_do_pin(MMUEXT_UNPIN_TABLE, pfn);
		}

		mcs = __xen_mc_entry(0);

		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
					pfn_pte(pfn, PAGE_KERNEL),
					level == PT_PGD ? UVMF_TLB_FLUSH : 0);

		if (ptl) {
			/* unlock when batch completed */
			xen_mc_callback(xen_pte_unlock, ptl);
		}
	}

	return 0;		/* never need to flush on unpin */
}

/* Release a pagetables pages back as normal RW */
static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd)
{
	trace_xen_mmu_pgd_unpin(mm, pgd);

	xen_mc_batch();

	xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(pgd);

		if (user_pgd) {
			xen_do_pin(MMUEXT_UNPIN_TABLE,
				   PFN_DOWN(__pa(user_pgd)));
			xen_unpin_page(mm, virt_to_page(user_pgd), PT_PGD);
		}
	}
#endif

#ifdef CONFIG_X86_PAE
	/* Need to make sure unshared kernel PMD is unpinned */
	xen_unpin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]),
		       PT_PMD);
#endif

	__xen_pgd_walk(mm, pgd, xen_unpin_page, USER_LIMIT);

	xen_mc_issue(0);
}

static void xen_pgd_unpin(struct mm_struct *mm)
{
	__xen_pgd_unpin(mm, mm->pgd);
}

/*
 * On resume, undo any pinning done at save, so that the rest of the
 * kernel doesn't see any unexpected pinned pagetables.
 */
void xen_mm_unpin_all(void)
{
	struct page *page;

	spin_lock(&pgd_lock);

	list_for_each_entry(page, &pgd_list, lru) {
		if (PageSavePinned(page)) {
			BUG_ON(!PagePinned(page));
			__xen_pgd_unpin(&init_mm, (pgd_t *)page_address(page));
			ClearPageSavePinned(page);
		}
	}

	spin_unlock(&pgd_lock);
}

static void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)
{
	spin_lock(&next->page_table_lock);
	xen_pgd_pin(next);
	spin_unlock(&next->page_table_lock);
}

static void xen_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
{
	spin_lock(&mm->page_table_lock);
	xen_pgd_pin(mm);
	spin_unlock(&mm->page_table_lock);
}


#ifdef CONFIG_SMP
/* Another cpu may still have their %cr3 pointing at the pagetable, so
   we need to repoint it somewhere else before we can unpin it. */
static void drop_other_mm_ref(void *info)
{
	struct mm_struct *mm = info;
	struct mm_struct *active_mm;

	active_mm = this_cpu_read(cpu_tlbstate.active_mm);

	if (active_mm == mm && this_cpu_read(cpu_tlbstate.state) != TLBSTATE_OK)
		leave_mm(smp_processor_id());

	/* If this cpu still has a stale cr3 reference, then make sure
	   it has been flushed. */
	if (this_cpu_read(xen_current_cr3) == __pa(mm->pgd))
		load_cr3(swapper_pg_dir);
}

static void xen_drop_mm_ref(struct mm_struct *mm)
{
	cpumask_var_t mask;
	unsigned cpu;

	if (current->active_mm == mm) {
		if (current->mm == mm)
			load_cr3(swapper_pg_dir);
		else
			leave_mm(smp_processor_id());
	}

	/* Get the "official" set of cpus referring to our pagetable. */
	if (!alloc_cpumask_var(&mask, GFP_ATOMIC)) {
		for_each_online_cpu(cpu) {
			if (!cpumask_test_cpu(cpu, mm_cpumask(mm))
			    && per_cpu(xen_current_cr3, cpu) != __pa(mm->pgd))
				continue;
			smp_call_function_single(cpu, drop_other_mm_ref, mm, 1);
		}
		return;
	}
	cpumask_copy(mask, mm_cpumask(mm));

	/* It's possible that a vcpu may have a stale reference to our
	   cr3, because its in lazy mode, and it hasn't yet flushed
	   its set of pending hypercalls yet.  In this case, we can
	   look at its actual current cr3 value, and force it to flush
	   if needed. */
	for_each_online_cpu(cpu) {
		if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd))
			cpumask_set_cpu(cpu, mask);
	}

	if (!cpumask_empty(mask))
		smp_call_function_many(mask, drop_other_mm_ref, mm, 1);
	free_cpumask_var(mask);
}
#else
static void xen_drop_mm_ref(struct mm_struct *mm)
{
	if (current->active_mm == mm)
		load_cr3(swapper_pg_dir);
}
#endif

/*
 * While a process runs, Xen pins its pagetables, which means that the
 * hypervisor forces it to be read-only, and it controls all updates
 * to it.  This means that all pagetable updates have to go via the
 * hypervisor, which is moderately expensive.
 *
 * Since we're pulling the pagetable down, we switch to use init_mm,
 * unpin old process pagetable and mark it all read-write, which
 * allows further operations on it to be simple memory accesses.
 *
 * The only subtle point is that another CPU may be still using the
 * pagetable because of lazy tlb flushing.  This means we need need to
 * switch all CPUs off this pagetable before we can unpin it.
 */
static void xen_exit_mmap(struct mm_struct *mm)
{
	get_cpu();		/* make sure we don't move around */
	xen_drop_mm_ref(mm);
	put_cpu();

	spin_lock(&mm->page_table_lock);

	/* pgd may not be pinned in the error exit path of execve */
	if (xen_page_pinned(mm->pgd))
		xen_pgd_unpin(mm);

	spin_unlock(&mm->page_table_lock);
}

static void __init xen_pagetable_setup_start(pgd_t *base)
{
}

static __init void xen_mapping_pagetable_reserve(u64 start, u64 end)
{
	/* reserve the range used */
	native_pagetable_reserve(start, end);

	/* set as RW the rest */
	printk(KERN_DEBUG "xen: setting RW the range %llx - %llx\n", end,
			PFN_PHYS(pgt_buf_top));
	while (end < PFN_PHYS(pgt_buf_top)) {
		make_lowmem_page_readwrite(__va(end));
		end += PAGE_SIZE;
	}
}

static void xen_post_allocator_init(void);

static void __init xen_pagetable_setup_done(pgd_t *base)
{
	xen_setup_shared_info();

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	xen_post_allocator_init();
}

static void xen_write_cr2(unsigned long cr2)
{
	this_cpu_read(xen_vcpu)->arch.cr2 = cr2;
}

static unsigned long xen_read_cr2(void)
{
	return this_cpu_read(xen_vcpu)->arch.cr2;
}

unsigned long xen_read_cr2_direct(void)
{
	return this_cpu_read(xen_vcpu_info.arch.cr2);
}

static void xen_flush_tlb(void)
{
	struct mmuext_op *op;
	struct multicall_space mcs;

	trace_xen_mmu_flush_tlb(0);

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*op));

	op = mcs.args;
	op->cmd = MMUEXT_TLB_FLUSH_LOCAL;
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_flush_tlb_single(unsigned long addr)
{
	struct mmuext_op *op;
	struct multicall_space mcs;

	trace_xen_mmu_flush_tlb_single(addr);

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*op));
	op = mcs.args;
	op->cmd = MMUEXT_INVLPG_LOCAL;
	op->arg1.linear_addr = addr & PAGE_MASK;
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

static void xen_flush_tlb_others(const struct cpumask *cpus,
				 struct mm_struct *mm, unsigned long start,
				 unsigned long end)
{
	struct {
		struct mmuext_op op;
#ifdef CONFIG_SMP
		DECLARE_BITMAP(mask, num_processors);
#else
		DECLARE_BITMAP(mask, NR_CPUS);
#endif
	} *args;
	struct multicall_space mcs;

	trace_xen_mmu_flush_tlb_others(cpus, mm, start, end);

	if (cpumask_empty(cpus))
		return;		/* nothing to do */

	mcs = xen_mc_entry(sizeof(*args));
	args = mcs.args;
	args->op.arg2.vcpumask = to_cpumask(args->mask);

	/* Remove us, and any offline CPUS. */
	cpumask_and(to_cpumask(args->mask), cpus, cpu_online_mask);
	cpumask_clear_cpu(smp_processor_id(), to_cpumask(args->mask));

	args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
	if (start != TLB_FLUSH_ALL && (end - start) <= PAGE_SIZE) {
		args->op.cmd = MMUEXT_INVLPG_MULTI;
		args->op.arg1.linear_addr = start;
	}

	MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}

static unsigned long xen_read_cr3(void)
{
	return this_cpu_read(xen_cr3);
}

static void set_current_cr3(void *v)
{
	this_cpu_write(xen_current_cr3, (unsigned long)v);
}

static void __xen_write_cr3(bool kernel, unsigned long cr3)
{
	struct mmuext_op op;
	unsigned long mfn;

	trace_xen_mmu_write_cr3(kernel, cr3);

	if (cr3)
		mfn = pfn_to_mfn(PFN_DOWN(cr3));
	else
		mfn = 0;

	WARN_ON(mfn == 0 && kernel);

	op.cmd = kernel ? MMUEXT_NEW_BASEPTR : MMUEXT_NEW_USER_BASEPTR;
	op.arg1.mfn = mfn;

	xen_extend_mmuext_op(&op);

	if (kernel) {
		this_cpu_write(xen_cr3, cr3);

		/* Update xen_current_cr3 once the batch has actually
		   been submitted. */
		xen_mc_callback(set_current_cr3, (void *)cr3);
	}
}

static void xen_write_cr3(unsigned long cr3)
{
	BUG_ON(preemptible());

	xen_mc_batch();  /* disables interrupts */

	/* Update while interrupts are disabled, so its atomic with
	   respect to ipis */
	this_cpu_write(xen_cr3, cr3);

	__xen_write_cr3(true, cr3);

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(__va(cr3));
		if (user_pgd)
			__xen_write_cr3(false, __pa(user_pgd));
		else
			__xen_write_cr3(false, 0);
	}
#endif

	xen_mc_issue(PARAVIRT_LAZY_CPU);  /* interrupts restored */
}

static int xen_pgd_alloc(struct mm_struct *mm)
{
	pgd_t *pgd = mm->pgd;
	int ret = 0;

	BUG_ON(PagePinned(virt_to_page(pgd)));

#ifdef CONFIG_X86_64
	{
		struct page *page = virt_to_page(pgd);
		pgd_t *user_pgd;

		BUG_ON(page->private != 0);

		ret = -ENOMEM;

		user_pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
		page->private = (unsigned long)user_pgd;

		if (user_pgd != NULL) {
			user_pgd[pgd_index(VSYSCALL_START)] =
				__pgd(__pa(level3_user_vsyscall) | _PAGE_TABLE);
			ret = 0;
		}

		BUG_ON(PagePinned(virt_to_page(xen_get_user_pgd(pgd))));
	}
#endif

	return ret;
}

static void xen_pgd_free(struct mm_struct *mm, pgd_t *pgd)
{
#ifdef CONFIG_X86_64
	pgd_t *user_pgd = xen_get_user_pgd(pgd);

	if (user_pgd)
		free_page((unsigned long)user_pgd);
#endif
}

#ifdef CONFIG_X86_32
static pte_t __init mask_rw_pte(pte_t *ptep, pte_t pte)
{
	/* If there's an existing pte, then don't allow _PAGE_RW to be set */
	if (pte_val_ma(*ptep) & _PAGE_PRESENT)
		pte = __pte_ma(((pte_val_ma(*ptep) & _PAGE_RW) | ~_PAGE_RW) &
			       pte_val_ma(pte));

	return pte;
}
#else /* CONFIG_X86_64 */
static pte_t __init mask_rw_pte(pte_t *ptep, pte_t pte)
{
	unsigned long pfn = pte_pfn(pte);

	/*
	 * If the new pfn is within the range of the newly allocated
	 * kernel pagetable, and it isn't being mapped into an
	 * early_ioremap fixmap slot as a freshly allocated page, make sure
	 * it is RO.
	 */
	if (((!is_early_ioremap_ptep(ptep) &&
			pfn >= pgt_buf_start && pfn < pgt_buf_top)) ||
			(is_early_ioremap_ptep(ptep) && pfn != (pgt_buf_end - 1)))
		pte = pte_wrprotect(pte);

	return pte;
}
#endif /* CONFIG_X86_64 */

/*
 * Init-time set_pte while constructing initial pagetables, which
 * doesn't allow RO page table pages to be remapped RW.
 *
 * If there is no MFN for this PFN then this page is initially
 * ballooned out so clear the PTE (as in decrease_reservation() in
 * drivers/xen/balloon.c).
 *
 * Many of these PTE updates are done on unpinned and writable pages
 * and doing a hypercall for these is unnecessary and expensive.  At
 * this point it is not possible to tell if a page is pinned or not,
 * so always write the PTE directly and rely on Xen trapping and
 * emulating any updates as necessary.
 */
static void __init xen_set_pte_init(pte_t *ptep, pte_t pte)
{
	if (pte_mfn(pte) != INVALID_P2M_ENTRY)
		pte = mask_rw_pte(ptep, pte);
	else
		pte = __pte_ma(0);

	native_set_pte(ptep, pte);
}

static noinline void pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
{
	struct mmuext_op op;

	if (xen_feature(XENFEAT_writable_page_tables))
		return;

	op.cmd = cmd;
	op.arg1.mfn = pfn_to_mfn(pfn);
	if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
		BUG();
}

/* Early in boot, while setting up the initial pagetable, assume
   everything is pinned. */
static void __init xen_alloc_pte_init(struct mm_struct *mm, unsigned long pfn)
{
#ifdef CONFIG_FLATMEM
	BUG_ON(mem_map);	/* should only be used early */
#endif
	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
	pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
}

/* Used for pmd and pud */
static void __init xen_alloc_pmd_init(struct mm_struct *mm, unsigned long pfn)
{
#ifdef CONFIG_FLATMEM
	BUG_ON(mem_map);	/* should only be used early */
#endif
	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
}

/* Early release_pte assumes that all pts are pinned, since there's
   only init_mm and anything attached to that is pinned. */
static void __init xen_release_pte_init(unsigned long pfn)
{
	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}

static void __init xen_release_pmd_init(unsigned long pfn)
{
	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}

static inline void __pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
{
	struct multicall_space mcs;
	struct mmuext_op *op;

	mcs = __xen_mc_entry(sizeof(*op));
	op = mcs.args;
	op->cmd = cmd;
	op->arg1.mfn = pfn_to_mfn(pfn);

	MULTI_mmuext_op(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
}

static inline void __set_pfn_prot(unsigned long pfn, pgprot_t prot)
{
	struct multicall_space mcs;
	unsigned long addr = (unsigned long)__va(pfn << PAGE_SHIFT);

	mcs = __xen_mc_entry(0);
	MULTI_update_va_mapping(mcs.mc, (unsigned long)addr,
				pfn_pte(pfn, prot), 0);
}

/* This needs to make sure the new pte page is pinned iff its being
   attached to a pinned pagetable. */
static inline void xen_alloc_ptpage(struct mm_struct *mm, unsigned long pfn,
				    unsigned level)
{
	bool pinned = PagePinned(virt_to_page(mm->pgd));

	trace_xen_mmu_alloc_ptpage(mm, pfn, level, pinned);

	if (pinned) {
		struct page *page = pfn_to_page(pfn);

		SetPagePinned(page);

		if (!PageHighMem(page)) {
			xen_mc_batch();

			__set_pfn_prot(pfn, PAGE_KERNEL_RO);

			if (level == PT_PTE && USE_SPLIT_PTLOCKS)
				__pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);

			xen_mc_issue(PARAVIRT_LAZY_MMU);
		} else {
			/* make sure there are no stray mappings of
			   this page */
			kmap_flush_unused();
		}
	}
}

static void xen_alloc_pte(struct mm_struct *mm, unsigned long pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PTE);
}

static void xen_alloc_pmd(struct mm_struct *mm, unsigned long pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PMD);
}

/* This should never happen until we're OK to use struct page */
static inline void xen_release_ptpage(unsigned long pfn, unsigned level)
{
	struct page *page = pfn_to_page(pfn);
	bool pinned = PagePinned(page);

	trace_xen_mmu_release_ptpage(pfn, level, pinned);

	if (pinned) {
		if (!PageHighMem(page)) {
			xen_mc_batch();

			if (level == PT_PTE && USE_SPLIT_PTLOCKS)
				__pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);

			__set_pfn_prot(pfn, PAGE_KERNEL);

			xen_mc_issue(PARAVIRT_LAZY_MMU);
		}
		ClearPagePinned(page);
	}
}

static void xen_release_pte(unsigned long pfn)
{
	xen_release_ptpage(pfn, PT_PTE);
}

static void xen_release_pmd(unsigned long pfn)
{
	xen_release_ptpage(pfn, PT_PMD);
}

#if PAGETABLE_LEVELS == 4
static void xen_alloc_pud(struct mm_struct *mm, unsigned long pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PUD);
}

static void xen_release_pud(unsigned long pfn)
{
	xen_release_ptpage(pfn, PT_PUD);
}
#endif

void __init xen_reserve_top(void)
{
#ifdef CONFIG_X86_32
	unsigned long top = HYPERVISOR_VIRT_START;
	struct xen_platform_parameters pp;

	if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
		top = pp.virt_start;

	reserve_top_address(-top);
#endif	/* CONFIG_X86_32 */
}

/*
 * Like __va(), but returns address in the kernel mapping (which is
 * all we have until the physical memory mapping has been set up.
 */
static void *__ka(phys_addr_t paddr)
{
#ifdef CONFIG_X86_64
	return (void *)(paddr + __START_KERNEL_map);
#else
	return __va(paddr);
#endif
}

/* Convert a machine address to physical address */
static unsigned long m2p(phys_addr_t maddr)
{
	phys_addr_t paddr;

	maddr &= PTE_PFN_MASK;
	paddr = mfn_to_pfn(maddr >> PAGE_SHIFT) << PAGE_SHIFT;

	return paddr;
}

/* Convert a machine address to kernel virtual */
static void *m2v(phys_addr_t maddr)
{
	return __ka(m2p(maddr));
}

/* Set the page permissions on an identity-mapped pages */
static void set_page_prot(void *addr, pgprot_t prot)
{
	unsigned long pfn = __pa(addr) >> PAGE_SHIFT;
	pte_t pte = pfn_pte(pfn, prot);

	/* recall for PVH, page tables are native. */
	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	if (HYPERVISOR_update_va_mapping((unsigned long)addr, pte, 0))
		BUG();
}

static void __init xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
{
	unsigned pmdidx, pteidx;
	unsigned ident_pte;
	unsigned long pfn;

	level1_ident_pgt = extend_brk(sizeof(pte_t) * LEVEL1_IDENT_ENTRIES,
				      PAGE_SIZE);

	ident_pte = 0;
	pfn = 0;
	for (pmdidx = 0; pmdidx < PTRS_PER_PMD && pfn < max_pfn; pmdidx++) {
		pte_t *pte_page;

		/* Reuse or allocate a page of ptes */
		if (pmd_present(pmd[pmdidx]))
			pte_page = m2v(pmd[pmdidx].pmd);
		else {
			/* Check for free pte pages */
			if (ident_pte == LEVEL1_IDENT_ENTRIES)
				break;

			pte_page = &level1_ident_pgt[ident_pte];
			ident_pte += PTRS_PER_PTE;

			pmd[pmdidx] = __pmd(__pa(pte_page) | _PAGE_TABLE);
		}

		/* Install mappings */
		for (pteidx = 0; pteidx < PTRS_PER_PTE; pteidx++, pfn++) {
			pte_t pte;

#ifdef CONFIG_X86_32
			if (pfn > max_pfn_mapped)
				max_pfn_mapped = pfn;
#endif

			if (!pte_none(pte_page[pteidx]))
				continue;

			pte = pfn_pte(pfn, PAGE_KERNEL_EXEC);
			pte_page[pteidx] = pte;
		}
	}

	for (pteidx = 0; pteidx < ident_pte; pteidx += PTRS_PER_PTE)
		set_page_prot(&level1_ident_pgt[pteidx], PAGE_KERNEL_RO);

	set_page_prot(pmd, PAGE_KERNEL_RO);
}

void __init xen_setup_machphys_mapping(void)
{
	struct xen_machphys_mapping mapping;

	if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
		machine_to_phys_mapping = (unsigned long *)mapping.v_start;
		machine_to_phys_nr = mapping.max_mfn + 1;
	} else {
		machine_to_phys_nr = MACH2PHYS_NR_ENTRIES;
	}
#ifdef CONFIG_X86_32
	WARN_ON((machine_to_phys_mapping + (machine_to_phys_nr - 1))
		< machine_to_phys_mapping);
#endif
}

#ifdef CONFIG_X86_64
static void convert_pfn_mfn(void *v)
{
	pte_t *pte = v;
	int i;

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	/* All levels are converted the same way, so just treat them
	   as ptes. */
	for (i = 0; i < PTRS_PER_PTE; i++)
		pte[i] = xen_make_pte(pte[i].pte);
}

/*
 * Set up the initial kernel pagetable.
 *
 * We can construct this by grafting the Xen provided pagetable into
 * head_64.S's preconstructed pagetables.  We copy the Xen L2's into
 * level2_ident_pgt, level2_kernel_pgt and level2_fixmap_pgt.  This
 * means that only the kernel has a physical mapping to start with -
 * but that's enough to get __va working.  We need to fill in the rest
 * of the physical mapping once some sort of allocator has been set
 * up.
 * NOTE: for PVH, the page tables are native.
 */
pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
					 unsigned long max_pfn)
{
	pud_t *l3;
	pmd_t *l2;

	/* max_pfn_mapped is the last pfn mapped in the initial memory
	 * mappings. Considering that on Xen after the kernel mappings we
	 * have the mappings of some pages that don't exist in pfn space, we
	 * set max_pfn_mapped to the last real pfn mapped. */
	max_pfn_mapped = PFN_DOWN(__pa(xen_start_info->mfn_list));

	/* Zap identity mapping */
	init_level4_pgt[0] = __pgd(0);

	/* Pre-constructed entries are in pfn, so convert to mfn */
	convert_pfn_mfn(init_level4_pgt);
	convert_pfn_mfn(level3_ident_pgt);
	convert_pfn_mfn(level3_kernel_pgt);

	l3 = m2v(pgd[pgd_index(__START_KERNEL_map)].pgd);
	l2 = m2v(l3[pud_index(__START_KERNEL_map)].pud);

	memcpy(level2_ident_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
	memcpy(level2_kernel_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);

	l3 = m2v(pgd[pgd_index(__START_KERNEL_map + PMD_SIZE)].pgd);
	l2 = m2v(l3[pud_index(__START_KERNEL_map + PMD_SIZE)].pud);
	memcpy(level2_fixmap_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);

	/* Set up identity map */
	xen_map_identity_early(level2_ident_pgt, max_pfn);

	/* Make pagetable pieces RO */
	set_page_prot(init_level4_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_ident_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_kernel_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_user_vsyscall, PAGE_KERNEL_RO);
	set_page_prot(level2_kernel_pgt, PAGE_KERNEL_RO);
	set_page_prot(level2_fixmap_pgt, PAGE_KERNEL_RO);

	/* Pin down new L4 */
	pin_pagetable_pfn(MMUEXT_PIN_L4_TABLE,
		  	PFN_DOWN(__pa_symbol(init_level4_pgt)));

	/* Unpin Xen-provided one */
	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	/* Switch over */
	pgd = init_level4_pgt;

	/*
	 * At this stage there can be no user pgd, and no page
	 * structure to attach it to, so make sure we just set kernel
	 * pgd.
	 */
	if (xen_feature(XENFEAT_writable_page_tables)) {
		native_write_cr3(__pa(pgd));
	} else {
		xen_mc_batch();
		__xen_write_cr3(true, __pa(pgd));
		xen_mc_issue(PARAVIRT_LAZY_CPU);
	}

	memblock_reserve(__pa(xen_start_info->pt_base),
			 xen_start_info->nr_pt_frames * PAGE_SIZE);

	return pgd;
}
#else	/* !CONFIG_X86_64 */
static RESERVE_BRK_ARRAY(pmd_t, initial_kernel_pmd, PTRS_PER_PMD);
static RESERVE_BRK_ARRAY(pmd_t, swapper_kernel_pmd, PTRS_PER_PMD);

static void __init xen_write_cr3_init(unsigned long cr3)
{
	unsigned long pfn = PFN_DOWN(__pa(swapper_pg_dir));

	BUG_ON(read_cr3() != __pa(initial_page_table));
	BUG_ON(cr3 != __pa(swapper_pg_dir));

	/*
	 * We are switching to swapper_pg_dir for the first time (from
	 * initial_page_table) and therefore need to mark that page
	 * read-only and then pin it.
	 *
	 * Xen disallows sharing of kernel PMDs for PAE
	 * guests. Therefore we must copy the kernel PMD from
	 * initial_page_table into a new kernel PMD to be used in
	 * swapper_pg_dir.
	 */
	swapper_kernel_pmd =
		extend_brk(sizeof(pmd_t) * PTRS_PER_PMD, PAGE_SIZE);
	memcpy(swapper_kernel_pmd, initial_kernel_pmd,
	       sizeof(pmd_t) * PTRS_PER_PMD);
	swapper_pg_dir[KERNEL_PGD_BOUNDARY] =
		__pgd(__pa(swapper_kernel_pmd) | _PAGE_PRESENT);
	set_page_prot(swapper_kernel_pmd, PAGE_KERNEL_RO);

	set_page_prot(swapper_pg_dir, PAGE_KERNEL_RO);
	xen_write_cr3(cr3);
	pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE, pfn);

	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE,
			  PFN_DOWN(__pa(initial_page_table)));
	set_page_prot(initial_page_table, PAGE_KERNEL);
	set_page_prot(initial_kernel_pmd, PAGE_KERNEL);

	pv_mmu_ops.write_cr3 = &xen_write_cr3;
}

pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
					 unsigned long max_pfn)
{
	pmd_t *kernel_pmd;

	initial_kernel_pmd =
		extend_brk(sizeof(pmd_t) * PTRS_PER_PMD, PAGE_SIZE);

	max_pfn_mapped = PFN_DOWN(__pa(xen_start_info->pt_base) +
				  xen_start_info->nr_pt_frames * PAGE_SIZE +
				  512*1024);

	kernel_pmd = m2v(pgd[KERNEL_PGD_BOUNDARY].pgd);
	memcpy(initial_kernel_pmd, kernel_pmd, sizeof(pmd_t) * PTRS_PER_PMD);

	xen_map_identity_early(initial_kernel_pmd, max_pfn);

	memcpy(initial_page_table, pgd, sizeof(pgd_t) * PTRS_PER_PGD);
	initial_page_table[KERNEL_PGD_BOUNDARY] =
		__pgd(__pa(initial_kernel_pmd) | _PAGE_PRESENT);

	set_page_prot(initial_kernel_pmd, PAGE_KERNEL_RO);
	set_page_prot(initial_page_table, PAGE_KERNEL_RO);
	set_page_prot(empty_zero_page, PAGE_KERNEL_RO);

	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE,
			  PFN_DOWN(__pa(initial_page_table)));
	xen_write_cr3(__pa(initial_page_table));

	memblock_reserve(__pa(xen_start_info->pt_base),
			 xen_start_info->nr_pt_frames * PAGE_SIZE);

	return initial_page_table;
}
#endif	/* CONFIG_X86_64 */

static unsigned char dummy_mapping[PAGE_SIZE] __page_aligned_bss;

static void xen_set_fixmap(unsigned idx, phys_addr_t phys, pgprot_t prot)
{
	pte_t pte;

	phys >>= PAGE_SHIFT;

	switch (idx) {
	case FIX_BTMAP_END ... FIX_BTMAP_BEGIN:
#ifdef CONFIG_X86_F00F_BUG
	case FIX_F00F_IDT:
#endif
#ifdef CONFIG_X86_32
	case FIX_WP_TEST:
	case FIX_VDSO:
# ifdef CONFIG_HIGHMEM
	case FIX_KMAP_BEGIN ... FIX_KMAP_END:
# endif
#else
	case VSYSCALL_LAST_PAGE ... VSYSCALL_FIRST_PAGE:
	case VVAR_PAGE:
#endif
	case FIX_TEXT_POKE0:
	case FIX_TEXT_POKE1:
		/* All local page mappings */
		pte = pfn_pte(phys, prot);
		break;

#ifdef CONFIG_X86_LOCAL_APIC
	case FIX_APIC_BASE:	/* maps dummy local APIC */
		pte = pfn_pte(PFN_DOWN(__pa(dummy_mapping)), PAGE_KERNEL);
		break;
#endif

#ifdef CONFIG_X86_IO_APIC
	case FIX_IO_APIC_BASE_0 ... FIX_IO_APIC_BASE_END:
		/*
		 * We just don't map the IO APIC - all access is via
		 * hypercalls.  Keep the address in the pte for reference.
		 */
		pte = pfn_pte(PFN_DOWN(__pa(dummy_mapping)), PAGE_KERNEL);
		break;
#endif

	case FIX_PARAVIRT_BOOTMAP:
		/* This is an MFN, but it isn't an IO mapping from the
		   IO domain */
		pte = mfn_pte(phys, prot);
		break;

	default:
		/* By default, set_fixmap is used for hardware mappings */
		pte = mfn_pte(phys, __pgprot(pgprot_val(prot) | _PAGE_IOMAP));
		break;
	}

	__native_set_fixmap(idx, pte);

#ifdef CONFIG_X86_64
	/* Replicate changes to map the vsyscall page into the user
	   pagetable vsyscall mapping. */
	if ((idx >= VSYSCALL_LAST_PAGE && idx <= VSYSCALL_FIRST_PAGE) ||
	    idx == VVAR_PAGE) {
		unsigned long vaddr = __fix_to_virt(idx);
		set_pte_vaddr_pud(level3_user_vsyscall, vaddr, pte);
	}
#endif
}

static void __init xen_post_allocator_init(void)
{
	pv_mmu_ops.set_pte = xen_set_pte;
	pv_mmu_ops.set_pmd = xen_set_pmd;
	pv_mmu_ops.set_pud = xen_set_pud;
#if PAGETABLE_LEVELS == 4
	pv_mmu_ops.set_pgd = xen_set_pgd;
#endif

	/* This will work as long as patching hasn't happened yet
	   (which it hasn't) */
	pv_mmu_ops.alloc_pte = xen_alloc_pte;
	pv_mmu_ops.alloc_pmd = xen_alloc_pmd;
	pv_mmu_ops.release_pte = xen_release_pte;
	pv_mmu_ops.release_pmd = xen_release_pmd;
#if PAGETABLE_LEVELS == 4
	pv_mmu_ops.alloc_pud = xen_alloc_pud;
	pv_mmu_ops.release_pud = xen_release_pud;
#endif

#ifdef CONFIG_X86_64
	SetPagePinned(virt_to_page(level3_user_vsyscall));
#endif
	xen_mark_init_mm_pinned();
}

static void xen_leave_lazy_mmu(void)
{
	preempt_disable();
	xen_mc_flush();
	paravirt_leave_lazy_mmu();
	preempt_enable();
}

static const struct pv_mmu_ops xen_mmu_ops __initconst = {
	.read_cr2 = xen_read_cr2,
	.write_cr2 = xen_write_cr2,

	.read_cr3 = xen_read_cr3,
#ifdef CONFIG_X86_32
	.write_cr3 = xen_write_cr3_init,
#else
	.write_cr3 = xen_write_cr3,
#endif

	.flush_tlb_user = xen_flush_tlb,
	.flush_tlb_kernel = xen_flush_tlb,
	.flush_tlb_single = xen_flush_tlb_single,
	.flush_tlb_others = xen_flush_tlb_others,

	.pte_update = paravirt_nop,
	.pte_update_defer = paravirt_nop,

	.pgd_alloc = xen_pgd_alloc,
	.pgd_free = xen_pgd_free,

	.alloc_pte = xen_alloc_pte_init,
	.release_pte = xen_release_pte_init,
	.alloc_pmd = xen_alloc_pmd_init,
	.release_pmd = xen_release_pmd_init,

	.set_pte = xen_set_pte_init,
	.set_pte_at = xen_set_pte_at,
	.set_pmd = xen_set_pmd_hyper,

	.ptep_modify_prot_start = __ptep_modify_prot_start,
	.ptep_modify_prot_commit = __ptep_modify_prot_commit,

	.pte_val = PV_CALLEE_SAVE(xen_pte_val),
	.pgd_val = PV_CALLEE_SAVE(xen_pgd_val),

	.make_pte = PV_CALLEE_SAVE(xen_make_pte),
	.make_pgd = PV_CALLEE_SAVE(xen_make_pgd),

#ifdef CONFIG_X86_PAE
	.set_pte_atomic = xen_set_pte_atomic,
	.pte_clear = xen_pte_clear,
	.pmd_clear = xen_pmd_clear,
#endif	/* CONFIG_X86_PAE */
	.set_pud = xen_set_pud_hyper,

	.make_pmd = PV_CALLEE_SAVE(xen_make_pmd),
	.pmd_val = PV_CALLEE_SAVE(xen_pmd_val),

#if PAGETABLE_LEVELS == 4
	.pud_val = PV_CALLEE_SAVE(xen_pud_val),
	.make_pud = PV_CALLEE_SAVE(xen_make_pud),
	.set_pgd = xen_set_pgd_hyper,

	.alloc_pud = xen_alloc_pmd_init,
	.release_pud = xen_release_pmd_init,
#endif	/* PAGETABLE_LEVELS == 4 */

	.activate_mm = xen_activate_mm,
	.dup_mmap = xen_dup_mmap,
	.exit_mmap = xen_exit_mmap,

	.lazy_mode = {
		.enter = paravirt_enter_lazy_mmu,
		.leave = xen_leave_lazy_mmu,
	},

	.set_fixmap = xen_set_fixmap,
};

void __init xen_init_mmu_ops(void)
{
	x86_init.paging.pagetable_setup_done = xen_pagetable_setup_done;

	if (xen_feature(XENFEAT_auto_translated_physmap)) {
		pv_mmu_ops.flush_tlb_others = xen_flush_tlb_others;

		/* set_pte* for PCI devices to map iomem. */
		if (xen_initial_domain()) {
			pv_mmu_ops.set_pte = native_set_pte;
			pv_mmu_ops.set_pte_at = xen_dom0pvh_set_pte_at;
		}
		return;
	}

	x86_init.mapping.pagetable_reserve = xen_mapping_pagetable_reserve;
	x86_init.paging.pagetable_setup_start = xen_pagetable_setup_start;
	pv_mmu_ops = xen_mmu_ops;

	memset(dummy_mapping, 0xff, PAGE_SIZE);
}

/* Protected by xen_reservation_lock. */
#define MAX_CONTIG_ORDER 9 /* 2MB */
static unsigned long discontig_frames[1<<MAX_CONTIG_ORDER];

#define VOID_PTE (mfn_pte(0, __pgprot(0)))
static void xen_zap_pfn_range(unsigned long vaddr, unsigned int order,
				unsigned long *in_frames,
				unsigned long *out_frames)
{
	int i;
	struct multicall_space mcs;

	xen_mc_batch();
	for (i = 0; i < (1UL<<order); i++, vaddr += PAGE_SIZE) {
		mcs = __xen_mc_entry(0);

		if (in_frames)
			in_frames[i] = virt_to_mfn(vaddr);

		MULTI_update_va_mapping(mcs.mc, vaddr, VOID_PTE, 0);
		__set_phys_to_machine(virt_to_pfn(vaddr), INVALID_P2M_ENTRY);

		if (out_frames)
			out_frames[i] = virt_to_pfn(vaddr);
	}
	xen_mc_issue(0);
}

/*
 * Update the pfn-to-mfn mappings for a virtual address range, either to
 * point to an array of mfns, or contiguously from a single starting
 * mfn.
 */
static void xen_remap_exchanged_ptes(unsigned long vaddr, int order,
				     unsigned long *mfns,
				     unsigned long first_mfn)
{
	unsigned i, limit;
	unsigned long mfn;

	xen_mc_batch();

	limit = 1u << order;
	for (i = 0; i < limit; i++, vaddr += PAGE_SIZE) {
		struct multicall_space mcs;
		unsigned flags;

		mcs = __xen_mc_entry(0);
		if (mfns)
			mfn = mfns[i];
		else
			mfn = first_mfn + i;

		if (i < (limit - 1))
			flags = 0;
		else {
			if (order == 0)
				flags = UVMF_INVLPG | UVMF_ALL;
			else
				flags = UVMF_TLB_FLUSH | UVMF_ALL;
		}

		MULTI_update_va_mapping(mcs.mc, vaddr,
				mfn_pte(mfn, PAGE_KERNEL), flags);

		set_phys_to_machine(virt_to_pfn(vaddr), mfn);
	}

	xen_mc_issue(0);
}

/*
 * Perform the hypercall to exchange a region of our pfns to point to
 * memory with the required contiguous alignment.  Takes the pfns as
 * input, and populates mfns as output.
 *
 * Returns a success code indicating whether the hypervisor was able to
 * satisfy the request or not.
 */
static int xen_exchange_memory(unsigned long extents_in, unsigned int order_in,
			       unsigned long *pfns_in,
			       unsigned long extents_out,
			       unsigned int order_out,
			       unsigned long *mfns_out,
			       unsigned int address_bits)
{
	long rc;
	int success;

	struct xen_memory_exchange exchange = {
		.in = {
			.nr_extents   = extents_in,
			.extent_order = order_in,
			.extent_start = pfns_in,
			.domid        = DOMID_SELF
		},
		.out = {
			.nr_extents   = extents_out,
			.extent_order = order_out,
			.extent_start = mfns_out,
			.address_bits = address_bits,
			.domid        = DOMID_SELF
		}
	};

	BUG_ON(extents_in << order_in != extents_out << order_out);

	rc = HYPERVISOR_memory_op(XENMEM_exchange, &exchange);
	success = (exchange.nr_exchanged == extents_in);

	BUG_ON(!success && ((exchange.nr_exchanged != 0) || (rc == 0)));
	BUG_ON(success && (rc != 0));

	return success;
}

int xen_create_contiguous_region(unsigned long vstart, unsigned int order,
				 unsigned int address_bits)
{
	unsigned long *in_frames = discontig_frames, out_frame;
	unsigned long  flags;
	int            success;

	/*
	 * Currently an auto-translated guest will not perform I/O, nor will
	 * it require PAE page directories below 4GB. Therefore any calls to
	 * this function are redundant and can be ignored.
	 */

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	if (unlikely(order > MAX_CONTIG_ORDER))
		return -ENOMEM;

	memset((void *) vstart, 0, PAGE_SIZE << order);

	spin_lock_irqsave(&xen_reservation_lock, flags);

	/* 1. Zap current PTEs, remembering MFNs. */
	xen_zap_pfn_range(vstart, order, in_frames, NULL);

	/* 2. Get a new contiguous memory extent. */
	out_frame = virt_to_pfn(vstart);
	success = xen_exchange_memory(1UL << order, 0, in_frames,
				      1, order, &out_frame,
				      address_bits);

	/* 3. Map the new extent in place of old pages. */
	if (success)
		xen_remap_exchanged_ptes(vstart, order, NULL, out_frame);
	else
		xen_remap_exchanged_ptes(vstart, order, in_frames, 0);

	spin_unlock_irqrestore(&xen_reservation_lock, flags);

	return success ? 0 : -ENOMEM;
}
EXPORT_SYMBOL_GPL(xen_create_contiguous_region);

void xen_destroy_contiguous_region(unsigned long vstart, unsigned int order)
{
	unsigned long *out_frames = discontig_frames, in_frame;
	unsigned long  flags;
	int success;

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return;

	if (unlikely(order > MAX_CONTIG_ORDER))
		return;

	memset((void *) vstart, 0, PAGE_SIZE << order);

	spin_lock_irqsave(&xen_reservation_lock, flags);

	/* 1. Find start MFN of contiguous extent. */
	in_frame = virt_to_mfn(vstart);

	/* 2. Zap current PTEs. */
	xen_zap_pfn_range(vstart, order, NULL, out_frames);

	/* 3. Do the exchange for non-contiguous MFNs. */
	success = xen_exchange_memory(1, order, &in_frame, 1UL << order,
					0, out_frames, 0);

	/* 4. Map new pages in place of old pages. */
	if (success)
		xen_remap_exchanged_ptes(vstart, order, out_frames, 0);
	else
		xen_remap_exchanged_ptes(vstart, order, NULL, in_frame);

	spin_unlock_irqrestore(&xen_reservation_lock, flags);
}
EXPORT_SYMBOL_GPL(xen_destroy_contiguous_region);

#ifdef CONFIG_XEN_PVHVM
static void xen_hvm_exit_mmap(struct mm_struct *mm)
{
	struct xen_hvm_pagetable_dying a;
	int rc;

	a.domid = DOMID_SELF;
	a.gpa = __pa(mm->pgd);
	rc = HYPERVISOR_hvm_op(HVMOP_pagetable_dying, &a);
	WARN_ON_ONCE(rc < 0);
}

static int is_pagetable_dying_supported(void)
{
	struct xen_hvm_pagetable_dying a;
	int rc = 0;

	a.domid = DOMID_SELF;
	a.gpa = 0x00;
	rc = HYPERVISOR_hvm_op(HVMOP_pagetable_dying, &a);
	if (rc < 0) {
		printk(KERN_DEBUG "HVMOP_pagetable_dying not supported\n");
		return 0;
	}
	return 1;
}

void __init xen_hvm_init_mmu_ops(void)
{
	if (is_pagetable_dying_supported())
		pv_mmu_ops.exit_mmap = xen_hvm_exit_mmap;
}
#endif

/* Map foreign gmfn, fgmfn, to local pfn, lpfn. This for the user space 
 * creating new guest on PVH dom0 and needs to map domU pages. Called from
 * exported function, so no need to export this.
 */
static noinline int pvh_add_to_xen_p2m(unsigned long lpfn, unsigned long fgmfn,
			      unsigned int domid)
{
	int rc;
	struct xen_add_to_physmap xatp = { .foreign_domid = domid };

	xatp.gpfn = lpfn;
	xatp.idx = fgmfn;
	xatp.domid = DOMID_SELF;
	xatp.space = XENMAPSPACE_gmfn_foreign;
	rc = HYPERVISOR_memory_op(XENMEM_add_to_physmap, &xatp);
	if (rc)
		pr_warn("d0: Failed to map pfn to mfn rc:%d pfn:%lx mfn:%lx\n", 
			rc, lpfn, fgmfn); 
	return rc;
}

int pvh_rem_xen_p2m(unsigned long spfn, int count)
{
	struct xen_remove_from_physmap xrp;
	int i, rc;

	for (i=0; i < count; i++) {
		xrp.domid = DOMID_SELF;
		xrp.gpfn = spfn+i;
		rc = HYPERVISOR_memory_op(XENMEM_remove_from_physmap, &xrp);
		if (rc) {
			pr_warn("Failed to unmap pfn:%lx rc:%d done:%d\n",
				spfn+i, rc, i);
			return 1;
		}
	}
	return 0;
}
EXPORT_SYMBOL_GPL(pvh_rem_xen_p2m);

struct pvh_remap_data {
	unsigned long fgmfn;		/* foreign domain's gmfn */
	pgprot_t prot;
	domid_t  domid;
	struct xen_pvh_pfn_info *pvhinfop;
};

static noinline int pvh_map_pte_fn(pte_t *ptep, pgtable_t token, unsigned long addr, 
			void *data)
{
	int rc;
	struct pvh_remap_data *remapp = data;
	struct xen_pvh_pfn_info *pvhp = remapp->pvhinfop;
	unsigned long pfn = page_to_pfn(pvhp->pi_paga[pvhp->pi_next_todo++]);
	pte_t pteval = pte_mkspecial(pfn_pte(pfn, remapp->prot));

	if ((rc=pvh_add_to_xen_p2m(pfn, remapp->fgmfn, remapp->domid)))
		return rc;
	native_set_pte(ptep, pteval);

	return 0;
}

/* The only caller at moment passes one gmfn at a time.
 * PVH TBD/FIXME: expand this in future to honor batch requests.
 */
static noinline int pvh_remap_gmfn_range(struct vm_area_struct *vma,
				unsigned long addr, unsigned long mfn, int nr,
				pgprot_t prot, unsigned domid,
				struct xen_pvh_pfn_info *pvhp)
{
	int err;
	struct pvh_remap_data pvhdata;

	if (nr > 1)
		return -EINVAL;

	pvhdata.fgmfn = mfn;
	pvhdata.prot = prot;
	pvhdata.domid = domid;
	pvhdata.pvhinfop = pvhp;
	err = apply_to_page_range(vma->vm_mm, addr, nr << PAGE_SHIFT,
				  pvh_map_pte_fn, &pvhdata);
	flush_tlb_all();
	return err;
}

#define REMAP_BATCH_SIZE 16

struct remap_data {
	unsigned long mfn;
	pgprot_t prot;
	struct mmu_update *mmu_update;
};

static int remap_area_mfn_pte_fn(pte_t *ptep, pgtable_t token,
				 unsigned long addr, void *data)
{
	struct remap_data *rmd = data;
	pte_t pte = pte_mkspecial(pfn_pte(rmd->mfn++, rmd->prot));

	rmd->mmu_update->ptr = virt_to_machine(ptep).maddr;
	rmd->mmu_update->val = pte_val_ma(pte);
	rmd->mmu_update++;

	return 0;
}

int xen_remap_domain_mfn_range(struct vm_area_struct *vma,
			       unsigned long addr,
			       unsigned long mfn, int nr,
			       pgprot_t prot, unsigned domid,
			       struct xen_pvh_pfn_info *pvhp)

{
	struct remap_data rmd;
	struct mmu_update mmu_update[REMAP_BATCH_SIZE];
	int batch;
	unsigned long range;
	int err = 0;

	prot = __pgprot(pgprot_val(prot) | _PAGE_IOMAP);

	BUG_ON(!((vma->vm_flags & (VM_PFNMAP | VM_RESERVED | VM_IO)) ==
				(VM_PFNMAP | VM_RESERVED | VM_IO)));

	if (xen_feature(XENFEAT_auto_translated_physmap)) {
		/* We need to update the local page tables and the xen HAP */
		return pvh_remap_gmfn_range(vma, addr, mfn, nr, prot, domid,
					    pvhp);
	}

	rmd.mfn = mfn;
	rmd.prot = prot;

	while (nr) {
		batch = min(REMAP_BATCH_SIZE, nr);
		range = (unsigned long)batch << PAGE_SHIFT;

		rmd.mmu_update = mmu_update;
		err = apply_to_page_range(vma->vm_mm, addr, range,
					  remap_area_mfn_pte_fn, &rmd);
		if (err)
			goto out;

		err = -EFAULT;
		if (HYPERVISOR_mmu_update(mmu_update, batch, NULL, domid) < 0)
			goto out;

		nr -= batch;
		addr += range;
	}

	err = 0;
out:

	flush_tlb_all();

	return err;
}
EXPORT_SYMBOL_GPL(xen_remap_domain_mfn_range);

/* Returns: Number of pages unmapped */
int xen_unmap_domain_mfn_range(struct vm_area_struct *vma,
			       struct xen_pvh_pfn_info *pvhp)
{
	int count = 0;

	if (!pvhp || !xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	while (pvhp->pi_next_todo--) {
		unsigned long pfn;

		/* the mmu has already cleaned up the process mmu resources at
		 * this point (lookup_address will return NULL). */
		pfn = page_to_pfn(pvhp->pi_paga[pvhp->pi_next_todo]);
		pvh_rem_xen_p2m(pfn, 1);
		count++;
	}
	flush_tlb_all();
	return count;
}
EXPORT_SYMBOL_GPL(xen_unmap_domain_mfn_range);

[-- Attachment #4: Type: text/plain, Size: 126 bytes --]

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

Re: [RFC PATCH 8/8]: PVH: privcmd changes

[Ian Campbell]

On Tue, 2012-09-18 at 00:50 +0100, Mukesh Rathor wrote:
> On Thu, 13 Sep 2012 17:34:20 -0700
> Mukesh Rathor <mukesh.rathor@oracle.com> wrote:
> 
> > On Thu, 13 Sep 2012 12:37:46 +0100
> > Ian Campbell <Ian.Campbell@citrix.com> wrote:
> > 
> > > On Thu, 2012-09-13 at 02:19 +0100, Mukesh Rathor wrote:
> > > > On Tue, 11 Sep 2012 15:10:23 +0100
> > > > 
> > > I think using vm_private within a subsystem/layer is ok, what I
> > > think we should avoid is having layers pass data back and forth in
> > > that field.
> > 
> > Ah I see your point. Ok, let me play around a bit see what I can do.
> 
> Hey Ian,
> 
> I played around a bit with privcmd, but not a whole lot can be done. I
> did change things around a bit to make the APIs more symmetric and hide
> vm_private_data from mmu.c. Please take a look. Unless major objections,
> I'd like to resubmit all linux patches asap.

It's pretty hard to review complete .c files rather than diffs. I think
you can/should just go ahead and resubmit the latest patches.

Ian.