[PATCH v2] arm/arm64: KVM: map MMIO regions at creation time

[ard.biesheuvel@linaro.org (Ard Biesheuvel)]

There is really no point in faulting in memory regions page by page
if they are not backed by demand paged system RAM but by a linear
passthrough mapping of a host MMIO region.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---

I have omitted the other 5 patches of the series of which this was #6, as
Christoffer had indicated they could be merged separately.

Changes since v1:
- move this logic to kvm_arch_prepare_memory_region() so it can be invoked
  when moving memory regions as well as when creating memory regions
- as we are reasoning about memory regions now instead of memslots, all data
  is retrieved from the 'mem' argument which points to a struct
  kvm_userspace_memory_region
- minor tweaks to the logic flow

Again, compile tested only, due to lack of test cases.

 arch/arm/kvm/mmu.c | 52 +++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 51 insertions(+), 1 deletion(-)

diff --git a/arch/arm/kvm/mmu.c b/arch/arm/kvm/mmu.c
index fe53c3a30383..1403d9dc1190 100644
--- a/arch/arm/kvm/mmu.c
+++ b/arch/arm/kvm/mmu.c
@@ -1151,7 +1151,57 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
 				   struct kvm_userspace_memory_region *mem,
 				   enum kvm_mr_change change)
 {
-	return 0;
+	hva_t hva = mem->userspace_addr;
+	hva_t reg_end = hva + mem->memory_size;
+	phys_addr_t gpa = mem->guest_phys_addr;
+	int ret = 0;
+
+	if (change != KVM_MR_CREATE && change != KVM_MR_MOVE)
+		return 0;
+
+	/*
+	 * A memory region could potentially cover multiple VMAs, so iterate
+	 * over all of them to find out if we can map any of them right now.
+	 *
+	 *     +--------------------------------------------+
+	 * +---+---------+-------------------+--------------+----+
+	 * |   : VMA 1   |       VMA 2       |       VMA 3  :    |
+	 * +---+---------+-------------------+--------------+----+
+	 *     |               memory region                |
+	 *     +--------------------------------------------+
+	 */
+	do {
+		struct vm_area_struct *vma = find_vma(current->mm, hva);
+		hva_t vm_end;
+
+		if (!vma || vma->vm_start > hva) {
+			ret = -EFAULT;
+			break;
+		}
+
+		vm_end = min(reg_end, vma->vm_end);
+
+		if (vma->vm_flags & VM_PFNMAP) {
+			phys_addr_t pa = (vma->vm_pgoff << PAGE_SHIFT) + hva -
+					 vma->vm_start;
+			bool writable = (vma->vm_flags & VM_WRITE) &&
+					!(mem->flags & KVM_MEM_READONLY);
+
+			ret = kvm_phys_addr_ioremap(kvm, gpa, pa, vm_end - hva,
+						    writable);
+			if (ret)
+				break;
+		}
+		gpa += vm_end - hva;
+		hva = vm_end;
+	} while (hva < reg_end);
+
+	if (ret) {
+		spin_lock(&kvm->mmu_lock);
+		unmap_stage2_range(kvm, mem->guest_phys_addr, mem->memory_size);
+		spin_unlock(&kvm->mmu_lock);
+	}
+	return ret;
 }
 
 void kvm_arch_free_memslot(struct kvm *kvm, struct kvm_memory_slot *free,
-- 
1.8.3.2