fuse: move ioctl to separate source file

Next patch will expand ioctl code and fuse/file.c is large enough as it is.

Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>

4 files changed, 394 insertions, 381 deletions

diff --git a/fs/fuse/Makefile b/fs/fuse/Makefile
index 8c7021fb2cd4..0c48b35c058d 100644
--- a/fs/fuse/Makefile
+++ b/fs/fuse/Makefile
@@ -7,7 +7,7 @@ obj-$(CONFIG_FUSE_FS) += fuse.o
 obj-$(CONFIG_CUSE) += cuse.o
 obj-$(CONFIG_VIRTIO_FS) += virtiofs.o
 
-fuse-y := dev.o dir.o file.o inode.o control.o xattr.o acl.o readdir.o
+fuse-y := dev.o dir.o file.o inode.o control.o xattr.o acl.o readdir.o ioctl.o
 fuse-$(CONFIG_FUSE_DAX) += dax.o
 
 virtiofs-y := virtio_fs.o
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index 8cccecb55fb8..c549850089bb 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -14,24 +14,11 @@
 #include <linux/sched.h>
 #include <linux/sched/signal.h>
 #include <linux/module.h>
-#include <linux/compat.h>
 #include <linux/swap.h>
 #include <linux/falloc.h>
 #include <linux/uio.h>
 #include <linux/fs.h>
 
-static struct page **fuse_pages_alloc(unsigned int npages, gfp_t flags,
-				      struct fuse_page_desc **desc)
-{
-	struct page **pages;
-
-	pages = kzalloc(npages * (sizeof(struct page *) +
-				  sizeof(struct fuse_page_desc)), flags);
-	*desc = (void *) (pages + npages);
-
-	return pages;
-}
-
 static int fuse_send_open(struct fuse_mount *fm, u64 nodeid, struct file *file,
 			  int opcode, struct fuse_open_out *outargp)
 {
@@ -1346,16 +1333,6 @@ out:
 	return written ? written : err;
 }
 
-static inline void fuse_page_descs_length_init(struct fuse_page_desc *descs,
-					       unsigned int index,
-					       unsigned int nr_pages)
-{
-	int i;
-
-	for (i = index; i < index + nr_pages; i++)
-		descs[i].length = PAGE_SIZE - descs[i].offset;
-}
-
 static inline unsigned long fuse_get_user_addr(const struct iov_iter *ii)
 {
 	return (unsigned long)ii->iov->iov_base + ii->iov_offset;
@@ -2637,363 +2614,6 @@ static loff_t fuse_file_llseek(struct file *file, loff_t offset, int whence)
 }
 
 /*
- * CUSE servers compiled on 32bit broke on 64bit kernels because the
- * ABI was defined to be 'struct iovec' which is different on 32bit
- * and 64bit.  Fortunately we can determine which structure the server
- * used from the size of the reply.
- */
-static int fuse_copy_ioctl_iovec_old(struct iovec *dst, void *src,
-				     size_t transferred, unsigned count,
-				     bool is_compat)
-{
-#ifdef CONFIG_COMPAT
-	if (count * sizeof(struct compat_iovec) == transferred) {
-		struct compat_iovec *ciov = src;
-		unsigned i;
-
-		/*
-		 * With this interface a 32bit server cannot support
-		 * non-compat (i.e. ones coming from 64bit apps) ioctl
-		 * requests
-		 */
-		if (!is_compat)
-			return -EINVAL;
-
-		for (i = 0; i < count; i++) {
-			dst[i].iov_base = compat_ptr(ciov[i].iov_base);
-			dst[i].iov_len = ciov[i].iov_len;
-		}
-		return 0;
-	}
-#endif
-
-	if (count * sizeof(struct iovec) != transferred)
-		return -EIO;
-
-	memcpy(dst, src, transferred);
-	return 0;
-}
-
-/* Make sure iov_length() won't overflow */
-static int fuse_verify_ioctl_iov(struct fuse_conn *fc, struct iovec *iov,
-				 size_t count)
-{
-	size_t n;
-	u32 max = fc->max_pages << PAGE_SHIFT;
-
-	for (n = 0; n < count; n++, iov++) {
-		if (iov->iov_len > (size_t) max)
-			return -ENOMEM;
-		max -= iov->iov_len;
-	}
-	return 0;
-}
-
-static int fuse_copy_ioctl_iovec(struct fuse_conn *fc, struct iovec *dst,
-				 void *src, size_t transferred, unsigned count,
-				 bool is_compat)
-{
-	unsigned i;
-	struct fuse_ioctl_iovec *fiov = src;
-
-	if (fc->minor < 16) {
-		return fuse_copy_ioctl_iovec_old(dst, src, transferred,
-						 count, is_compat);
-	}
-
-	if (count * sizeof(struct fuse_ioctl_iovec) != transferred)
-		return -EIO;
-
-	for (i = 0; i < count; i++) {
-		/* Did the server supply an inappropriate value? */
-		if (fiov[i].base != (unsigned long) fiov[i].base ||
-		    fiov[i].len != (unsigned long) fiov[i].len)
-			return -EIO;
-
-		dst[i].iov_base = (void __user *) (unsigned long) fiov[i].base;
-		dst[i].iov_len = (size_t) fiov[i].len;
-
-#ifdef CONFIG_COMPAT
-		if (is_compat &&
-		    (ptr_to_compat(dst[i].iov_base) != fiov[i].base ||
-		     (compat_size_t) dst[i].iov_len != fiov[i].len))
-			return -EIO;
-#endif
-	}
-
-	return 0;
-}
-
-
-/*
- * For ioctls, there is no generic way to determine how much memory
- * needs to be read and/or written.  Furthermore, ioctls are allowed
- * to dereference the passed pointer, so the parameter requires deep
- * copying but FUSE has no idea whatsoever about what to copy in or
- * out.
- *
- * This is solved by allowing FUSE server to retry ioctl with
- * necessary in/out iovecs.  Let's assume the ioctl implementation
- * needs to read in the following structure.
- *
- * struct a {
- *	char	*buf;
- *	size_t	buflen;
- * }
- *
- * On the first callout to FUSE server, inarg->in_size and
- * inarg->out_size will be NULL; then, the server completes the ioctl
- * with FUSE_IOCTL_RETRY set in out->flags, out->in_iovs set to 1 and
- * the actual iov array to
- *
- * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a) } }
- *
- * which tells FUSE to copy in the requested area and retry the ioctl.
- * On the second round, the server has access to the structure and
- * from that it can tell what to look for next, so on the invocation,
- * it sets FUSE_IOCTL_RETRY, out->in_iovs to 2 and iov array to
- *
- * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a)	},
- *   { .iov_base = a.buf,	.iov_len = a.buflen		} }
- *
- * FUSE will copy both struct a and the pointed buffer from the
- * process doing the ioctl and retry ioctl with both struct a and the
- * buffer.
- *
- * This time, FUSE server has everything it needs and completes ioctl
- * without FUSE_IOCTL_RETRY which finishes the ioctl call.
- *
- * Copying data out works the same way.
- *
- * Note that if FUSE_IOCTL_UNRESTRICTED is clear, the kernel
- * automatically initializes in and out iovs by decoding @cmd with
- * _IOC_* macros and the server is not allowed to request RETRY.  This
- * limits ioctl data transfers to well-formed ioctls and is the forced
- * behavior for all FUSE servers.
- */
-long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
-		   unsigned int flags)
-{
-	struct fuse_file *ff = file->private_data;
-	struct fuse_mount *fm = ff->fm;
-	struct fuse_ioctl_in inarg = {
-		.fh = ff->fh,
-		.cmd = cmd,
-		.arg = arg,
-		.flags = flags
-	};
-	struct fuse_ioctl_out outarg;
-	struct iovec *iov_page = NULL;
-	struct iovec *in_iov = NULL, *out_iov = NULL;
-	unsigned int in_iovs = 0, out_iovs = 0, max_pages;
-	size_t in_size, out_size, c;
-	ssize_t transferred;
-	int err, i;
-	struct iov_iter ii;
-	struct fuse_args_pages ap = {};
-
-#if BITS_PER_LONG == 32
-	inarg.flags |= FUSE_IOCTL_32BIT;
-#else
-	if (flags & FUSE_IOCTL_COMPAT) {
-		inarg.flags |= FUSE_IOCTL_32BIT;
-#ifdef CONFIG_X86_X32
-		if (in_x32_syscall())
-			inarg.flags |= FUSE_IOCTL_COMPAT_X32;
-#endif
-	}
-#endif
-
-	/* assume all the iovs returned by client always fits in a page */
-	BUILD_BUG_ON(sizeof(struct fuse_ioctl_iovec) * FUSE_IOCTL_MAX_IOV > PAGE_SIZE);
-
-	err = -ENOMEM;
-	ap.pages = fuse_pages_alloc(fm->fc->max_pages, GFP_KERNEL, &ap.descs);
-	iov_page = (struct iovec *) __get_free_page(GFP_KERNEL);
-	if (!ap.pages || !iov_page)
-		goto out;
-
-	fuse_page_descs_length_init(ap.descs, 0, fm->fc->max_pages);
-
-	/*
-	 * If restricted, initialize IO parameters as encoded in @cmd.
-	 * RETRY from server is not allowed.
-	 */
-	if (!(flags & FUSE_IOCTL_UNRESTRICTED)) {
-		struct iovec *iov = iov_page;
-
-		iov->iov_base = (void __user *)arg;
-
-		switch (cmd) {
-		case FS_IOC_GETFLAGS:
-		case FS_IOC_SETFLAGS:
-			iov->iov_len = sizeof(int);
-			break;
-		default:
-			iov->iov_len = _IOC_SIZE(cmd);
-			break;
-		}
-
-		if (_IOC_DIR(cmd) & _IOC_WRITE) {
-			in_iov = iov;
-			in_iovs = 1;
-		}
-
-		if (_IOC_DIR(cmd) & _IOC_READ) {
-			out_iov = iov;
-			out_iovs = 1;
-		}
-	}
-
- retry:
-	inarg.in_size = in_size = iov_length(in_iov, in_iovs);
-	inarg.out_size = out_size = iov_length(out_iov, out_iovs);
-
-	/*
-	 * Out data can be used either for actual out data or iovs,
-	 * make sure there always is at least one page.
-	 */
-	out_size = max_t(size_t, out_size, PAGE_SIZE);
-	max_pages = DIV_ROUND_UP(max(in_size, out_size), PAGE_SIZE);
-
-	/* make sure there are enough buffer pages and init request with them */
-	err = -ENOMEM;
-	if (max_pages > fm->fc->max_pages)
-		goto out;
-	while (ap.num_pages < max_pages) {
-		ap.pages[ap.num_pages] = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
-		if (!ap.pages[ap.num_pages])
-			goto out;
-		ap.num_pages++;
-	}
-
-
-	/* okay, let's send it to the client */
-	ap.args.opcode = FUSE_IOCTL;
-	ap.args.nodeid = ff->nodeid;
-	ap.args.in_numargs = 1;
-	ap.args.in_args[0].size = sizeof(inarg);
-	ap.args.in_args[0].value = &inarg;
-	if (in_size) {
-		ap.args.in_numargs++;
-		ap.args.in_args[1].size = in_size;
-		ap.args.in_pages = true;
-
-		err = -EFAULT;
-		iov_iter_init(&ii, WRITE, in_iov, in_iovs, in_size);
-		for (i = 0; iov_iter_count(&ii) && !WARN_ON(i >= ap.num_pages); i++) {
-			c = copy_page_from_iter(ap.pages[i], 0, PAGE_SIZE, &ii);
-			if (c != PAGE_SIZE && iov_iter_count(&ii))
-				goto out;
-		}
-	}
-
-	ap.args.out_numargs = 2;
-	ap.args.out_args[0].size = sizeof(outarg);
-	ap.args.out_args[0].value = &outarg;
-	ap.args.out_args[1].size = out_size;
-	ap.args.out_pages = true;
-	ap.args.out_argvar = true;
-
-	transferred = fuse_simple_request(fm, &ap.args);
-	err = transferred;
-	if (transferred < 0)
-		goto out;
-
-	/* did it ask for retry? */
-	if (outarg.flags & FUSE_IOCTL_RETRY) {
-		void *vaddr;
-
-		/* no retry if in restricted mode */
-		err = -EIO;
-		if (!(flags & FUSE_IOCTL_UNRESTRICTED))
-			goto out;
-
-		in_iovs = outarg.in_iovs;
-		out_iovs = outarg.out_iovs;
-
-		/*
-		 * Make sure things are in boundary, separate checks
-		 * are to protect against overflow.
-		 */
-		err = -ENOMEM;
-		if (in_iovs > FUSE_IOCTL_MAX_IOV ||
-		    out_iovs > FUSE_IOCTL_MAX_IOV ||
-		    in_iovs + out_iovs > FUSE_IOCTL_MAX_IOV)
-			goto out;
-
-		vaddr = kmap_atomic(ap.pages[0]);
-		err = fuse_copy_ioctl_iovec(fm->fc, iov_page, vaddr,
-					    transferred, in_iovs + out_iovs,
-					    (flags & FUSE_IOCTL_COMPAT) != 0);
-		kunmap_atomic(vaddr);
-		if (err)
-			goto out;
-
-		in_iov = iov_page;
-		out_iov = in_iov + in_iovs;
-
-		err = fuse_verify_ioctl_iov(fm->fc, in_iov, in_iovs);
-		if (err)
-			goto out;
-
-		err = fuse_verify_ioctl_iov(fm->fc, out_iov, out_iovs);
-		if (err)
-			goto out;
-
-		goto retry;
-	}
-
-	err = -EIO;
-	if (transferred > inarg.out_size)
-		goto out;
-
-	err = -EFAULT;
-	iov_iter_init(&ii, READ, out_iov, out_iovs, transferred);
-	for (i = 0; iov_iter_count(&ii) && !WARN_ON(i >= ap.num_pages); i++) {
-		c = copy_page_to_iter(ap.pages[i], 0, PAGE_SIZE, &ii);
-		if (c != PAGE_SIZE && iov_iter_count(&ii))
-			goto out;
-	}
-	err = 0;
- out:
-	free_page((unsigned long) iov_page);
-	while (ap.num_pages)
-		__free_page(ap.pages[--ap.num_pages]);
-	kfree(ap.pages);
-
-	return err ? err : outarg.result;
-}
-EXPORT_SYMBOL_GPL(fuse_do_ioctl);
-
-long fuse_ioctl_common(struct file *file, unsigned int cmd,
-		       unsigned long arg, unsigned int flags)
-{
-	struct inode *inode = file_inode(file);
-	struct fuse_conn *fc = get_fuse_conn(inode);
-
-	if (!fuse_allow_current_process(fc))
-		return -EACCES;
-
-	if (fuse_is_bad(inode))
-		return -EIO;
-
-	return fuse_do_ioctl(file, cmd, arg, flags);
-}
-
-static long fuse_file_ioctl(struct file *file, unsigned int cmd,
-			    unsigned long arg)
-{
-	return fuse_ioctl_common(file, cmd, arg, 0);
-}
-
-static long fuse_file_compat_ioctl(struct file *file, unsigned int cmd,
-				   unsigned long arg)
-{
-	return fuse_ioctl_common(file, cmd, arg, FUSE_IOCTL_COMPAT);
-}
-
-/*
  * All files which have been polled are linked to RB tree
  * fuse_conn->polled_files which is indexed by kh.  Walk the tree and
  * find the matching one.
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index 63d97a15ffde..ad7e629f3903 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -872,6 +872,28 @@ static inline bool fuse_is_bad(struct inode *inode)
 	return unlikely(test_bit(FUSE_I_BAD, &get_fuse_inode(inode)->state));
 }
 
+static inline struct page **fuse_pages_alloc(unsigned int npages, gfp_t flags,
+					     struct fuse_page_desc **desc)
+{
+	struct page **pages;
+
+	pages = kzalloc(npages * (sizeof(struct page *) +
+				  sizeof(struct fuse_page_desc)), flags);
+	*desc = (void *) (pages + npages);
+
+	return pages;
+}
+
+static inline void fuse_page_descs_length_init(struct fuse_page_desc *descs,
+					       unsigned int index,
+					       unsigned int nr_pages)
+{
+	int i;
+
+	for (i = index; i < index + nr_pages; i++)
+		descs[i].length = PAGE_SIZE - descs[i].offset;
+}
+
 /** Device operations */
 extern const struct file_operations fuse_dev_operations;
 
@@ -1214,4 +1236,9 @@ void fuse_dax_inode_cleanup(struct inode *inode);
 bool fuse_dax_check_alignment(struct fuse_conn *fc, unsigned int map_alignment);
 void fuse_dax_cancel_work(struct fuse_conn *fc);
 
+/* ioctl.c */
+long fuse_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
+long fuse_file_compat_ioctl(struct file *file, unsigned int cmd,
+			    unsigned long arg);
+
 #endif /* _FS_FUSE_I_H */
diff --git a/fs/fuse/ioctl.c b/fs/fuse/ioctl.c
new file mode 100644
index 000000000000..c2f3ba162ecb
--- /dev/null
+++ b/fs/fuse/ioctl.c
@@ -0,0 +1,366 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017 Red Hat, Inc.
+ */
+
+#include "fuse_i.h"
+
+#include <linux/uio.h>
+#include <linux/compat.h>
+#include <linux/fileattr.h>
+
+/*
+ * CUSE servers compiled on 32bit broke on 64bit kernels because the
+ * ABI was defined to be 'struct iovec' which is different on 32bit
+ * and 64bit.  Fortunately we can determine which structure the server
+ * used from the size of the reply.
+ */
+static int fuse_copy_ioctl_iovec_old(struct iovec *dst, void *src,
+				     size_t transferred, unsigned count,
+				     bool is_compat)
+{
+#ifdef CONFIG_COMPAT
+	if (count * sizeof(struct compat_iovec) == transferred) {
+		struct compat_iovec *ciov = src;
+		unsigned i;
+
+		/*
+		 * With this interface a 32bit server cannot support
+		 * non-compat (i.e. ones coming from 64bit apps) ioctl
+		 * requests
+		 */
+		if (!is_compat)
+			return -EINVAL;
+
+		for (i = 0; i < count; i++) {
+			dst[i].iov_base = compat_ptr(ciov[i].iov_base);
+			dst[i].iov_len = ciov[i].iov_len;
+		}
+		return 0;
+	}
+#endif
+
+	if (count * sizeof(struct iovec) != transferred)
+		return -EIO;
+
+	memcpy(dst, src, transferred);
+	return 0;
+}
+
+/* Make sure iov_length() won't overflow */
+static int fuse_verify_ioctl_iov(struct fuse_conn *fc, struct iovec *iov,
+				 size_t count)
+{
+	size_t n;
+	u32 max = fc->max_pages << PAGE_SHIFT;
+
+	for (n = 0; n < count; n++, iov++) {
+		if (iov->iov_len > (size_t) max)
+			return -ENOMEM;
+		max -= iov->iov_len;
+	}
+	return 0;
+}
+
+static int fuse_copy_ioctl_iovec(struct fuse_conn *fc, struct iovec *dst,
+				 void *src, size_t transferred, unsigned count,
+				 bool is_compat)
+{
+	unsigned i;
+	struct fuse_ioctl_iovec *fiov = src;
+
+	if (fc->minor < 16) {
+		return fuse_copy_ioctl_iovec_old(dst, src, transferred,
+						 count, is_compat);
+	}
+
+	if (count * sizeof(struct fuse_ioctl_iovec) != transferred)
+		return -EIO;
+
+	for (i = 0; i < count; i++) {
+		/* Did the server supply an inappropriate value? */
+		if (fiov[i].base != (unsigned long) fiov[i].base ||
+		    fiov[i].len != (unsigned long) fiov[i].len)
+			return -EIO;
+
+		dst[i].iov_base = (void __user *) (unsigned long) fiov[i].base;
+		dst[i].iov_len = (size_t) fiov[i].len;
+
+#ifdef CONFIG_COMPAT
+		if (is_compat &&
+		    (ptr_to_compat(dst[i].iov_base) != fiov[i].base ||
+		     (compat_size_t) dst[i].iov_len != fiov[i].len))
+			return -EIO;
+#endif
+	}
+
+	return 0;
+}
+
+
+/*
+ * For ioctls, there is no generic way to determine how much memory
+ * needs to be read and/or written.  Furthermore, ioctls are allowed
+ * to dereference the passed pointer, so the parameter requires deep
+ * copying but FUSE has no idea whatsoever about what to copy in or
+ * out.
+ *
+ * This is solved by allowing FUSE server to retry ioctl with
+ * necessary in/out iovecs.  Let's assume the ioctl implementation
+ * needs to read in the following structure.
+ *
+ * struct a {
+ *	char	*buf;
+ *	size_t	buflen;
+ * }
+ *
+ * On the first callout to FUSE server, inarg->in_size and
+ * inarg->out_size will be NULL; then, the server completes the ioctl
+ * with FUSE_IOCTL_RETRY set in out->flags, out->in_iovs set to 1 and
+ * the actual iov array to
+ *
+ * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a) } }
+ *
+ * which tells FUSE to copy in the requested area and retry the ioctl.
+ * On the second round, the server has access to the structure and
+ * from that it can tell what to look for next, so on the invocation,
+ * it sets FUSE_IOCTL_RETRY, out->in_iovs to 2 and iov array to
+ *
+ * { { .iov_base = inarg.arg,	.iov_len = sizeof(struct a)	},
+ *   { .iov_base = a.buf,	.iov_len = a.buflen		} }
+ *
+ * FUSE will copy both struct a and the pointed buffer from the
+ * process doing the ioctl and retry ioctl with both struct a and the
+ * buffer.
+ *
+ * This time, FUSE server has everything it needs and completes ioctl
+ * without FUSE_IOCTL_RETRY which finishes the ioctl call.
+ *
+ * Copying data out works the same way.
+ *
+ * Note that if FUSE_IOCTL_UNRESTRICTED is clear, the kernel
+ * automatically initializes in and out iovs by decoding @cmd with
+ * _IOC_* macros and the server is not allowed to request RETRY.  This
+ * limits ioctl data transfers to well-formed ioctls and is the forced
+ * behavior for all FUSE servers.
+ */
+long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
+		   unsigned int flags)
+{
+	struct fuse_file *ff = file->private_data;
+	struct fuse_mount *fm = ff->fm;
+	struct fuse_ioctl_in inarg = {
+		.fh = ff->fh,
+		.cmd = cmd,
+		.arg = arg,
+		.flags = flags
+	};
+	struct fuse_ioctl_out outarg;
+	struct iovec *iov_page = NULL;
+	struct iovec *in_iov = NULL, *out_iov = NULL;
+	unsigned int in_iovs = 0, out_iovs = 0, max_pages;
+	size_t in_size, out_size, c;
+	ssize_t transferred;
+	int err, i;
+	struct iov_iter ii;
+	struct fuse_args_pages ap = {};
+
+#if BITS_PER_LONG == 32
+	inarg.flags |= FUSE_IOCTL_32BIT;
+#else
+	if (flags & FUSE_IOCTL_COMPAT) {
+		inarg.flags |= FUSE_IOCTL_32BIT;
+#ifdef CONFIG_X86_X32
+		if (in_x32_syscall())
+			inarg.flags |= FUSE_IOCTL_COMPAT_X32;
+#endif
+	}
+#endif
+
+	/* assume all the iovs returned by client always fits in a page */
+	BUILD_BUG_ON(sizeof(struct fuse_ioctl_iovec) * FUSE_IOCTL_MAX_IOV > PAGE_SIZE);
+
+	err = -ENOMEM;
+	ap.pages = fuse_pages_alloc(fm->fc->max_pages, GFP_KERNEL, &ap.descs);
+	iov_page = (struct iovec *) __get_free_page(GFP_KERNEL);
+	if (!ap.pages || !iov_page)
+		goto out;
+
+	fuse_page_descs_length_init(ap.descs, 0, fm->fc->max_pages);
+
+	/*
+	 * If restricted, initialize IO parameters as encoded in @cmd.
+	 * RETRY from server is not allowed.
+	 */
+	if (!(flags & FUSE_IOCTL_UNRESTRICTED)) {
+		struct iovec *iov = iov_page;
+
+		iov->iov_base = (void __user *)arg;
+
+		switch (cmd) {
+		case FS_IOC_GETFLAGS:
+		case FS_IOC_SETFLAGS:
+			iov->iov_len = sizeof(int);
+			break;
+		default:
+			iov->iov_len = _IOC_SIZE(cmd);
+			break;
+		}
+
+		if (_IOC_DIR(cmd) & _IOC_WRITE) {
+			in_iov = iov;
+			in_iovs = 1;
+		}
+
+		if (_IOC_DIR(cmd) & _IOC_READ) {
+			out_iov = iov;
+			out_iovs = 1;
+		}
+	}
+
+ retry:
+	inarg.in_size = in_size = iov_length(in_iov, in_iovs);
+	inarg.out_size = out_size = iov_length(out_iov, out_iovs);
+
+	/*
+	 * Out data can be used either for actual out data or iovs,
+	 * make sure there always is at least one page.
+	 */
+	out_size = max_t(size_t, out_size, PAGE_SIZE);
+	max_pages = DIV_ROUND_UP(max(in_size, out_size), PAGE_SIZE);
+
+	/* make sure there are enough buffer pages and init request with them */
+	err = -ENOMEM;
+	if (max_pages > fm->fc->max_pages)
+		goto out;
+	while (ap.num_pages < max_pages) {
+		ap.pages[ap.num_pages] = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
+		if (!ap.pages[ap.num_pages])
+			goto out;
+		ap.num_pages++;
+	}
+
+
+	/* okay, let's send it to the client */
+	ap.args.opcode = FUSE_IOCTL;
+	ap.args.nodeid = ff->nodeid;
+	ap.args.in_numargs = 1;
+	ap.args.in_args[0].size = sizeof(inarg);
+	ap.args.in_args[0].value = &inarg;
+	if (in_size) {
+		ap.args.in_numargs++;
+		ap.args.in_args[1].size = in_size;
+		ap.args.in_pages = true;
+
+		err = -EFAULT;
+		iov_iter_init(&ii, WRITE, in_iov, in_iovs, in_size);
+		for (i = 0; iov_iter_count(&ii) && !WARN_ON(i >= ap.num_pages); i++) {
+			c = copy_page_from_iter(ap.pages[i], 0, PAGE_SIZE, &ii);
+			if (c != PAGE_SIZE && iov_iter_count(&ii))
+				goto out;
+		}
+	}
+
+	ap.args.out_numargs = 2;
+	ap.args.out_args[0].size = sizeof(outarg);
+	ap.args.out_args[0].value = &outarg;
+	ap.args.out_args[1].size = out_size;
+	ap.args.out_pages = true;
+	ap.args.out_argvar = true;
+
+	transferred = fuse_simple_request(fm, &ap.args);
+	err = transferred;
+	if (transferred < 0)
+		goto out;
+
+	/* did it ask for retry? */
+	if (outarg.flags & FUSE_IOCTL_RETRY) {
+		void *vaddr;
+
+		/* no retry if in restricted mode */
+		err = -EIO;
+		if (!(flags & FUSE_IOCTL_UNRESTRICTED))
+			goto out;
+
+		in_iovs = outarg.in_iovs;
+		out_iovs = outarg.out_iovs;
+
+		/*
+		 * Make sure things are in boundary, separate checks
+		 * are to protect against overflow.
+		 */
+		err = -ENOMEM;
+		if (in_iovs > FUSE_IOCTL_MAX_IOV ||
+		    out_iovs > FUSE_IOCTL_MAX_IOV ||
+		    in_iovs + out_iovs > FUSE_IOCTL_MAX_IOV)
+			goto out;
+
+		vaddr = kmap_atomic(ap.pages[0]);
+		err = fuse_copy_ioctl_iovec(fm->fc, iov_page, vaddr,
+					    transferred, in_iovs + out_iovs,
+					    (flags & FUSE_IOCTL_COMPAT) != 0);
+		kunmap_atomic(vaddr);
+		if (err)
+			goto out;
+
+		in_iov = iov_page;
+		out_iov = in_iov + in_iovs;
+
+		err = fuse_verify_ioctl_iov(fm->fc, in_iov, in_iovs);
+		if (err)
+			goto out;
+
+		err = fuse_verify_ioctl_iov(fm->fc, out_iov, out_iovs);
+		if (err)
+			goto out;
+
+		goto retry;
+	}
+
+	err = -EIO;
+	if (transferred > inarg.out_size)
+		goto out;
+
+	err = -EFAULT;
+	iov_iter_init(&ii, READ, out_iov, out_iovs, transferred);
+	for (i = 0; iov_iter_count(&ii) && !WARN_ON(i >= ap.num_pages); i++) {
+		c = copy_page_to_iter(ap.pages[i], 0, PAGE_SIZE, &ii);
+		if (c != PAGE_SIZE && iov_iter_count(&ii))
+			goto out;
+	}
+	err = 0;
+ out:
+	free_page((unsigned long) iov_page);
+	while (ap.num_pages)
+		__free_page(ap.pages[--ap.num_pages]);
+	kfree(ap.pages);
+
+	return err ? err : outarg.result;
+}
+EXPORT_SYMBOL_GPL(fuse_do_ioctl);
+
+long fuse_ioctl_common(struct file *file, unsigned int cmd,
+		       unsigned long arg, unsigned int flags)
+{
+	struct inode *inode = file_inode(file);
+	struct fuse_conn *fc = get_fuse_conn(inode);
+
+	if (!fuse_allow_current_process(fc))
+		return -EACCES;
+
+	if (fuse_is_bad(inode))
+		return -EIO;
+
+	return fuse_do_ioctl(file, cmd, arg, flags);
+}
+
+long fuse_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
+{
+	return fuse_ioctl_common(file, cmd, arg, 0);
+}
+
+long fuse_file_compat_ioctl(struct file *file, unsigned int cmd,
+			    unsigned long arg)
+{
+	return fuse_ioctl_common(file, cmd, arg, FUSE_IOCTL_COMPAT);
+}