Merge the following changes from projects/bhyve_npt_pmap:

- add fields to 'struct pmap' that are required to manage nested page tables.
- add a parameter to 'vmspace_alloc()' that can be used to override the
  default pmap initialization routine 'pmap_pinit()'.

These changes are pushed ahead of the remaining changes in 'bhyve_npt_pmap'
in anticipation of the upcoming KBI freeze for 10.0.

Reviewed by:	kib@, alc@
Approved by:	re (glebius)

3 files changed, 24 insertions, 7 deletions

diff --git a/sys/amd64/include/pmap.h b/sys/amd64/include/pmap.h
index b570cb7..11e1cf2 100644
--- a/sys/amd64/include/pmap.h
+++ b/sys/amd64/include/pmap.h
@@ -231,6 +231,12 @@ struct md_page {
 	int			pat_mode;
 };
 
+enum pmap_type {
+	PT_X86,			/* regular x86 page tables */
+	PT_EPT,			/* Intel's nested page tables */
+	PT_RVI,			/* AMD's nested page tables */
+};
+
 /*
  * The kernel virtual address (KVA) of the level 4 page table page is always
  * within the direct map (DMAP) region.
@@ -243,9 +249,11 @@ struct pmap {
 	cpuset_t		pm_active;	/* active on cpus */
 	cpuset_t		pm_save;	/* Context valid on cpus mask */
 	int			pm_pcid;	/* context id */
-	/* spare u_int here due to padding */
+	enum pmap_type		pm_type;	/* regular or nested tables */
 	struct pmap_statistics	pm_stats;	/* pmap statistics */
 	struct vm_radix		pm_root;	/* spare page table pages */
+	long			pm_eptgen;	/* EPT pmap generation id */
+	int			pm_flags;
 };
 
 typedef struct pmap	*pmap_t;
diff --git a/sys/vm/vm_extern.h b/sys/vm/vm_extern.h
index fee55d0..39a1a21 100644
--- a/sys/vm/vm_extern.h
+++ b/sys/vm/vm_extern.h
@@ -33,6 +33,7 @@
 #ifndef _VM_EXTERN_H_
 #define	_VM_EXTERN_H_
 
+struct pmap;
 struct proc;
 struct vmspace;
 struct vnode;
@@ -88,7 +89,8 @@ int vm_mmap(vm_map_t, vm_offset_t *, vm_size_t, vm_prot_t, vm_prot_t, int, objty
 int vm_mmap_to_errno(int rv);
 void vm_set_page_size(void);
 void vm_sync_icache(vm_map_t, vm_offset_t, vm_size_t);
-struct vmspace *vmspace_alloc(vm_offset_t, vm_offset_t);
+typedef int (*pmap_pinit_t)(struct pmap *pmap);
+struct vmspace *vmspace_alloc(vm_offset_t, vm_offset_t, pmap_pinit_t);
 struct vmspace *vmspace_fork(struct vmspace *, vm_ooffset_t *);
 int vmspace_exec(struct proc *, vm_offset_t, vm_offset_t);
 int vmspace_unshare(struct proc *);
diff --git a/sys/vm/vm_map.c b/sys/vm/vm_map.c
index 5990630..45171da 100644
--- a/sys/vm/vm_map.c
+++ b/sys/vm/vm_map.c
@@ -280,15 +280,22 @@ vm_map_zdtor(void *mem, int size, void *arg)
 /*
  * Allocate a vmspace structure, including a vm_map and pmap,
  * and initialize those structures.  The refcnt is set to 1.
+ *
+ * If 'pinit' is NULL then the embedded pmap is initialized via pmap_pinit().
  */
 struct vmspace *
-vmspace_alloc(min, max)
-	vm_offset_t min, max;
+vmspace_alloc(vm_offset_t min, vm_offset_t max, pmap_pinit_t pinit)
 {
 	struct vmspace *vm;
 
 	vm = uma_zalloc(vmspace_zone, M_WAITOK);
-	if (vm->vm_map.pmap == NULL && !pmap_pinit(vmspace_pmap(vm))) {
+
+	KASSERT(vm->vm_map.pmap == NULL, ("vm_map.pmap must be NULL"));
+
+	if (pinit == NULL)
+		pinit = &pmap_pinit;
+
+	if (!pinit(vmspace_pmap(vm))) {
 		uma_zfree(vmspace_zone, vm);
 		return (NULL);
 	}
@@ -3157,7 +3164,7 @@ vmspace_fork(struct vmspace *vm1, vm_ooffset_t *fork_charge)
 
 	old_map = &vm1->vm_map;
 	/* Copy immutable fields of vm1 to vm2. */
-	vm2 = vmspace_alloc(old_map->min_offset, old_map->max_offset);
+	vm2 = vmspace_alloc(old_map->min_offset, old_map->max_offset, NULL);
 	if (vm2 == NULL)
 		return (NULL);
 	vm2->vm_taddr = vm1->vm_taddr;
@@ -3739,7 +3746,7 @@ vmspace_exec(struct proc *p, vm_offset_t minuser, vm_offset_t maxuser)
 	struct vmspace *oldvmspace = p->p_vmspace;
 	struct vmspace *newvmspace;
 
-	newvmspace = vmspace_alloc(minuser, maxuser);
+	newvmspace = vmspace_alloc(minuser, maxuser, NULL);
 	if (newvmspace == NULL)
 		return (ENOMEM);
 	newvmspace->vm_swrss = oldvmspace->vm_swrss;