[RFC v2 PATCH 05/13] KVM: Implement fd-based memory using new memfd interfaces

[Chao Peng]

This patch pairs a fd-based memslot to a memory backing store. Two sides
handshake to exchange callbacks that will be called later.

KVM->memfd:
  - get_pfn: get or allocate(when alloc is true) page at specified
    offset in the fd, the page will be locked
  - put_pfn: put and unlock the pfn

memfd->KVM:
  - invalidate_page_range: called when userspace punch hole on the fd,
    KVM should unmap related pages in the second MMU
  - fallocate: called when userspace fallocate space on the fd, KVM
    can map related pages in the second MMU

Currently tmpfs behind memfd interface is supported.

Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com>
Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
---
 arch/x86/kvm/Makefile    |   3 +-
 include/linux/kvm_host.h |   6 +++
 virt/kvm/memfd.c         | 101 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 109 insertions(+), 1 deletion(-)
 create mode 100644 virt/kvm/memfd.c

diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile
index f919df73e5e3..5d7f289b1ca0 100644
--- a/arch/x86/kvm/Makefile
+++ b/arch/x86/kvm/Makefile
@@ -11,7 +11,8 @@ KVM := ../../../virt/kvm
 
 kvm-y                  += $(KVM)/kvm_main.o $(KVM)/coalesced_mmio.o \
                                $(KVM)/eventfd.o $(KVM)/irqchip.o $(KVM)/vfio.o 
\
-                               $(KVM)/dirty_ring.o $(KVM)/binary_stats.o
+                               $(KVM)/dirty_ring.o $(KVM)/binary_stats.o \
+                               $(KVM)/memfd.o
 kvm-$(CONFIG_KVM_ASYNC_PF)     += $(KVM)/async_pf.o
 
 kvm-y                  += x86.o emulate.o i8259.o irq.o lapic.o \
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 1d4ac0c9b63b..e8646103356b 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -769,6 +769,12 @@ static inline void kvm_irqfd_exit(void)
 {
 }
 #endif
+
+int kvm_memfd_register(struct kvm *kvm,
+                      const struct kvm_userspace_memory_region_ext *mem,
+                      struct kvm_memory_slot *slot);
+void kvm_memfd_unregister(struct kvm *kvm, struct kvm_memory_slot *slot);
+
 int kvm_init(void *opaque, unsigned vcpu_size, unsigned vcpu_align,
                  struct module *module);
 void kvm_exit(void);
diff --git a/virt/kvm/memfd.c b/virt/kvm/memfd.c
new file mode 100644
index 000000000000..bd930dcb455f
--- /dev/null
+++ b/virt/kvm/memfd.c
@@ -0,0 +1,101 @@
+
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * memfd.c: routines for fd based guest memory backing store
+ * Copyright (c) 2021, Intel Corporation.
+ *
+ * Author:
+ *     Chao Peng <chao.p.peng@linux.intel.com>
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/memfd.h>
+const static struct guest_mem_ops *memfd_ops;
+
+static void memfd_invalidate_page_range(struct inode *inode, void *owner,
+                                       pgoff_t start, pgoff_t end)
+{
+       //!!!We can get here after the owner no longer exists
+}
+
+static void memfd_fallocate(struct inode *inode, void *owner,
+                           pgoff_t start, pgoff_t end)
+{
+       //!!!We can get here after the owner no longer exists
+}
+
+static const struct guest_ops memfd_notifier = {
+       .invalidate_page_range = memfd_invalidate_page_range,
+       .fallocate = memfd_fallocate,
+};
+
+static kvm_pfn_t kvm_memfd_get_pfn(struct kvm_memory_slot *slot,
+                                  struct file *file, gfn_t gfn,
+                                  bool alloc, int *order)
+{
+       pgoff_t index = gfn - slot->base_gfn +
+                       (slot->userspace_addr >> PAGE_SHIFT);
+
+       return memfd_ops->get_lock_pfn(file->f_inode, index, alloc, order);
+}
+
+static void kvm_memfd_put_pfn(kvm_pfn_t pfn)
+{
+       memfd_ops->put_unlock_pfn(pfn);
+}
+
+static struct kvm_memfd_ops kvm_memfd_ops = {
+       .get_pfn = kvm_memfd_get_pfn,
+       .put_pfn = kvm_memfd_put_pfn,
+};
+
+int kvm_memfd_register(struct kvm *kvm,
+                      const struct kvm_userspace_memory_region_ext *mem,
+                      struct kvm_memory_slot *slot)
+{
+       int ret;
+       struct fd fd = fdget(mem->fd);
+
+       if (!fd.file)
+               return -EINVAL;
+
+       ret = memfd_register_guest(fd.file->f_inode, kvm,
+                                  &memfd_notifier, &memfd_ops);
+       if (ret)
+               return ret;
+       slot->file = fd.file;
+
+       if (mem->private_fd >= 0) {
+               fd = fdget(mem->private_fd);
+               if (!fd.file) {
+                       ret = -EINVAL;
+                       goto err;
+               }
+
+               ret = memfd_register_guest(fd.file->f_inode, kvm,
+                                          &memfd_notifier, &memfd_ops);
+               if (ret)
+                       goto err;
+               slot->priv_file = fd.file;
+       }
+
+       slot->memfd_ops = &kvm_memfd_ops;
+       return 0;
+err:
+       kvm_memfd_unregister(kvm, slot);
+       return ret;
+}
+
+void kvm_memfd_unregister(struct kvm *kvm, struct kvm_memory_slot *slot)
+{
+       if (slot->file) {
+               fput(slot->file);
+               slot->file = NULL;
+       }
+
+       if (slot->priv_file) {
+               fput(slot->priv_file);
+               slot->priv_file = NULL;
+       }
+       slot->memfd_ops = NULL;
+}
-- 
2.17.1