[RFC PATCH 02/11] KVM: guest_mem: Add ioctl KVM_LINK_GUEST_MEMFD

[Ackerley Tng]

KVM_LINK_GUEST_MEMFD will link a gmem fd's underlying inode to a new
file (and fd).

Signed-off-by: Ackerley Tng <ackerleytng@xxxxxxxxxx>
---
 include/uapi/linux/kvm.h |  8 +++++
 virt/kvm/guest_mem.c     | 73 ++++++++++++++++++++++++++++++++++++++++
 virt/kvm/kvm_main.c      | 10 ++++++
 virt/kvm/kvm_mm.h        |  7 ++++
 4 files changed, 98 insertions(+)

diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index eb900344a054..d0e2a2ce0df2 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -2299,4 +2299,12 @@ struct kvm_create_guest_memfd {
 	__u64 reserved[6];
 };
 
+#define KVM_LINK_GUEST_MEMFD	_IOWR(KVMIO,  0xd5, struct kvm_link_guest_memfd)
+
+struct kvm_link_guest_memfd {
+	__u64 fd;
+	__u64 flags;
+	__u64 reserved[6];
+};
+
 #endif /* __LINUX_KVM_H */
diff --git a/virt/kvm/guest_mem.c b/virt/kvm/guest_mem.c
index 30d0ab8745ee..1b3df273f785 100644
--- a/virt/kvm/guest_mem.c
+++ b/virt/kvm/guest_mem.c
@@ -477,6 +477,79 @@ int kvm_gmem_create(struct kvm *kvm, struct kvm_create_guest_memfd *args)
 	return __kvm_gmem_create(kvm, size, flags, kvm_gmem_mnt);
 }
 
+static inline void __kvm_gmem_do_link(struct inode *inode)
+{
+	/* Refer to simple_link() */
+
+	inode->i_ctime = current_time(inode);
+	inc_nlink(inode);
+
+	/*
+	 * ihold() to add additional reference to inode for reference in dentry,
+	 * created in kvm_gmem_alloc_file() -> alloc_file_pseudo(). This is not
+	 * necessary when creating a new file because alloc_inode() creates
+	 * inodes with i_count = 1, which is the refcount for the dentry in the
+	 * file.
+	 */
+	ihold(inode);
+
+	/*
+	 * dget() and d_instantiate() complete the setup of a dentry, but those
+	 * have already been done in kvm_gmem_alloc_file() ->
+	 * alloc_file_pseudo()
+	 */
+}
+
+int kvm_gmem_link(struct kvm *kvm, struct kvm_link_guest_memfd *args)
+{
+	int ret;
+	int fd;
+	struct fd f;
+	struct kvm_gmem *gmem;
+	u64 flags = args->flags;
+	u64 valid_flags = 0;
+	struct inode *inode;
+	struct file *dst_file;
+
+	if (flags & ~valid_flags)
+		return -EINVAL;
+
+	f = fdget(args->fd);
+	if (!f.file)
+		return -EINVAL;
+
+	ret = -EINVAL;
+	if (f.file->f_op != &kvm_gmem_fops)
+		goto out;
+
+	/* Cannot link a gmem file with the same vm again */
+	gmem = f.file->private_data;
+	if (gmem->kvm == kvm)
+		goto out;
+
+	ret = fd = get_unused_fd_flags(0);
+	if (fd < 0)
+		goto out;
+
+	inode = file_inode(f.file);
+	dst_file = kvm_gmem_alloc_file(inode, kvm_gmem_mnt);
+	if (IS_ERR(dst_file)) {
+		ret = PTR_ERR(dst_file);
+		goto out_fd;
+	}
+
+	__kvm_gmem_do_link(inode);
+
+	fd_install(fd, dst_file);
+	return fd;
+
+out_fd:
+	put_unused_fd(fd);
+out:
+	fdput(f);
+	return ret;
+}
+
 int kvm_gmem_bind(struct kvm *kvm, struct kvm_memory_slot *slot,
 		  unsigned int fd, loff_t offset)
 {
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index ee331cf8ba54..51cc8b80ebe0 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -5177,6 +5177,16 @@ static long kvm_vm_ioctl(struct file *filp,
 		r = kvm_gmem_create(kvm, &guest_memfd);
 		break;
 	}
+	case KVM_LINK_GUEST_MEMFD: {
+		struct kvm_link_guest_memfd params;
+
+		r = -EFAULT;
+		if (copy_from_user(&params, argp, sizeof(params)))
+			goto out;
+
+		r = kvm_gmem_link(kvm, &params);
+		break;
+	}
 	default:
 		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
 	}
diff --git a/virt/kvm/kvm_mm.h b/virt/kvm/kvm_mm.h
index 798f20d612bb..f85f452133b3 100644
--- a/virt/kvm/kvm_mm.h
+++ b/virt/kvm/kvm_mm.h
@@ -41,6 +41,7 @@ static inline void gfn_to_pfn_cache_invalidate_start(struct kvm *kvm,
 int kvm_gmem_init(void);
 void kvm_gmem_exit(void);
 int kvm_gmem_create(struct kvm *kvm, struct kvm_create_guest_memfd *args);
+int kvm_gmem_link(struct kvm *kvm, struct kvm_link_guest_memfd *args);
 int kvm_gmem_bind(struct kvm *kvm, struct kvm_memory_slot *slot,
 		  unsigned int fd, loff_t offset);
 void kvm_gmem_unbind(struct kvm_memory_slot *slot);
@@ -61,6 +62,12 @@ static inline int kvm_gmem_create(struct kvm *kvm,
 	return -EOPNOTSUPP;
 }
 
+static inline int kvm_gmem_link(struct kvm *kvm,
+				struct kvm_link_guest_memfd *args)
+{
+	return -EOPNOTSUPP;
+}
+
 static inline int kvm_gmem_bind(struct kvm *kvm,
 					 struct kvm_memory_slot *slot,
 					 unsigned int fd, loff_t offset)
-- 
2.41.0.640.ga95def55d0-goog