new methods: ->read_iter() and ->write_iter()

Beginning to introduce those.  Just the callers for now, and it's
clumsier than it'll eventually become; once we finish converting
aio_read and aio_write instances, the things will get nicer.

For now, these guys are in parallel to ->aio_read() and ->aio_write();
they take iocb and iov_iter, with everything in iov_iter already
validated.  File offset is passed in iocb->ki_pos, iov/nr_segs -
in iov_iter.

Main concerns in that series are stack footprint and ability to
split the damn thing cleanly.

[fix from Peter Ujfalusi <peter.ujfalusi@ti.com> folded]

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Documentation/filesystems/Locking

@@ -430,6 +430,8 @@ prototypes:
 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
+	ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
+	ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
 	int (*iterate) (struct file *, struct dir_context *);
 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);

Documentation/filesystems/vfs.txt

@@ -806,6 +806,8 @@ struct file_operations {
 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
+	ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
+	ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
 	int (*iterate) (struct file *, struct dir_context *);
 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
@@ -836,11 +838,15 @@ otherwise noted.
 
   read: called by read(2) and related system calls
 
-  aio_read: called by io_submit(2) and other asynchronous I/O operations
+  aio_read: vectored, possibly asynchronous read
+
+  read_iter: possibly asynchronous read with iov_iter as destination
 
   write: called by write(2) and related system calls
 
-  aio_write: called by io_submit(2) and other asynchronous I/O operations
+  aio_write: vectored, possibly asynchronous write
+
+  write_iter: possibly asynchronous write with iov_iter as source
 
   iterate: called when the VFS needs to read the directory contents
 

fs/aio.c

@@ -1241,6 +1241,7 @@ SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx)
 
 typedef ssize_t (aio_rw_op)(struct kiocb *, const struct iovec *,
 			    unsigned long, loff_t);
+typedef ssize_t (rw_iter_op)(struct kiocb *, struct iov_iter *);
 
 static ssize_t aio_setup_vectored_rw(struct kiocb *kiocb,
 				     int rw, char __user *buf,
@@ -1298,7 +1299,9 @@ static ssize_t aio_run_iocb(struct kiocb *req, unsigned opcode,
 	int rw;
 	fmode_t mode;
 	aio_rw_op *rw_op;
+	rw_iter_op *iter_op;
 	struct iovec inline_vec, *iovec = &inline_vec;
+	struct iov_iter iter;
 
 	switch (opcode) {
 	case IOCB_CMD_PREAD:
@@ -1306,6 +1309,7 @@ static ssize_t aio_run_iocb(struct kiocb *req, unsigned opcode,
 		mode	= FMODE_READ;
 		rw	= READ;
 		rw_op	= file->f_op->aio_read;
+		iter_op	= file->f_op->read_iter;
 		goto rw_common;
 
 	case IOCB_CMD_PWRITE:
@@ -1313,12 +1317,13 @@ static ssize_t aio_run_iocb(struct kiocb *req, unsigned opcode,
 		mode	= FMODE_WRITE;
 		rw	= WRITE;
 		rw_op	= file->f_op->aio_write;
+		iter_op	= file->f_op->write_iter;
 		goto rw_common;
 rw_common:
 		if (unlikely(!(file->f_mode & mode)))
 			return -EBADF;
 
-		if (!rw_op)
+		if (!rw_op && !iter_op)
 			return -EINVAL;
 
 		ret = (opcode == IOCB_CMD_PREADV ||
@@ -1347,7 +1352,12 @@ rw_common:
 		if (rw == WRITE)
 			file_start_write(file);
 
-		ret = rw_op(req, iovec, nr_segs, req->ki_pos);
+		if (iter_op) {
+			iov_iter_init(&iter, rw, iovec, nr_segs, req->ki_nbytes);
+			ret = iter_op(req, &iter);
+		} else {
+			ret = rw_op(req, iovec, nr_segs, req->ki_pos);
+		}
 
 		if (rw == WRITE)
 			file_end_write(file);

fs/file_table.c

@@ -175,9 +175,11 @@ struct file *alloc_file(struct path *path, fmode_t mode,
 	file->f_path = *path;
 	file->f_inode = path->dentry->d_inode;
 	file->f_mapping = path->dentry->d_inode->i_mapping;
-	if ((mode & FMODE_READ) && likely(fop->read || fop->aio_read))
+	if ((mode & FMODE_READ) &&
+	     likely(fop->read || fop->aio_read || fop->read_iter))
 		mode |= FMODE_CAN_READ;
-	if ((mode & FMODE_WRITE) && likely(fop->write || fop->aio_write))
+	if ((mode & FMODE_WRITE) &&
+	     likely(fop->write || fop->aio_write || fop->write_iter))
 		mode |= FMODE_CAN_WRITE;
 	file->f_mode = mode;
 	file->f_op = fop;

fs/open.c

@@ -725,9 +725,11 @@ static int do_dentry_open(struct file *f,
 	}
 	if ((f->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
 		i_readcount_inc(inode);
-	if ((f->f_mode & FMODE_READ) && likely(f->f_op->read || f->f_op->aio_read))
+	if ((f->f_mode & FMODE_READ) &&
+	     likely(f->f_op->read || f->f_op->aio_read || f->f_op->read_iter))
 		f->f_mode |= FMODE_CAN_READ;
-	if ((f->f_mode & FMODE_WRITE) && likely(f->f_op->write || f->f_op->aio_write))
+	if ((f->f_mode & FMODE_WRITE) &&
+	     likely(f->f_op->write || f->f_op->aio_write || f->f_op->write_iter))
 		f->f_mode |= FMODE_CAN_WRITE;
 
 	f->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

fs/read_write.c

@@ -25,6 +25,7 @@
 typedef ssize_t (*io_fn_t)(struct file *, char __user *, size_t, loff_t *);
 typedef ssize_t (*iov_fn_t)(struct kiocb *, const struct iovec *,
 		unsigned long, loff_t);
+typedef ssize_t (*iter_fn_t)(struct kiocb *, struct iov_iter *);
 
 const struct file_operations generic_ro_fops = {
 	.llseek		= generic_file_llseek,
@@ -390,6 +391,27 @@ ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *pp
 
 EXPORT_SYMBOL(do_sync_read);
 
+ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos)
+{
+	struct iovec iov = { .iov_base = buf, .iov_len = len };
+	struct kiocb kiocb;
+	struct iov_iter iter;
+	ssize_t ret;
+
+	init_sync_kiocb(&kiocb, filp);
+	kiocb.ki_pos = *ppos;
+	kiocb.ki_nbytes = len;
+	iov_iter_init(&iter, READ, &iov, 1, len);
+
+	ret = filp->f_op->read_iter(&kiocb, &iter);
+	if (-EIOCBQUEUED == ret)
+		ret = wait_on_sync_kiocb(&kiocb);
+	*ppos = kiocb.ki_pos;
+	return ret;
+}
+
+EXPORT_SYMBOL(new_sync_read);
+
 ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
 {
 	ssize_t ret;
@@ -406,8 +428,10 @@ ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
 		count = ret;
 		if (file->f_op->read)
 			ret = file->f_op->read(file, buf, count, pos);
-		else
+		else if (file->f_op->aio_read)
 			ret = do_sync_read(file, buf, count, pos);
+		else
+			ret = new_sync_read(file, buf, count, pos);
 		if (ret > 0) {
 			fsnotify_access(file);
 			add_rchar(current, ret);
@@ -439,6 +463,27 @@ ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, lof
 
 EXPORT_SYMBOL(do_sync_write);
 
+ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos)
+{
+	struct iovec iov = { .iov_base = (void __user *)buf, .iov_len = len };
+	struct kiocb kiocb;
+	struct iov_iter iter;
+	ssize_t ret;
+
+	init_sync_kiocb(&kiocb, filp);
+	kiocb.ki_pos = *ppos;
+	kiocb.ki_nbytes = len;
+	iov_iter_init(&iter, WRITE, &iov, 1, len);
+
+	ret = filp->f_op->write_iter(&kiocb, &iter);
+	if (-EIOCBQUEUED == ret)
+		ret = wait_on_sync_kiocb(&kiocb);
+	*ppos = kiocb.ki_pos;
+	return ret;
+}
+
+EXPORT_SYMBOL(new_sync_write);
+
 ssize_t __kernel_write(struct file *file, const char *buf, size_t count, loff_t *pos)
 {
 	mm_segment_t old_fs;
@@ -455,8 +500,10 @@ ssize_t __kernel_write(struct file *file, const char *buf, size_t count, loff_t
 		count =  MAX_RW_COUNT;
 	if (file->f_op->write)
 		ret = file->f_op->write(file, p, count, pos);
-	else
+	else if (file->f_op->aio_write)
 		ret = do_sync_write(file, p, count, pos);
+	else
+		ret = new_sync_write(file, p, count, pos);
 	set_fs(old_fs);
 	if (ret > 0) {
 		fsnotify_modify(file);
@@ -483,8 +530,10 @@ ssize_t vfs_write(struct file *file, const char __user *buf, size_t count, loff_
 		file_start_write(file);
 		if (file->f_op->write)
 			ret = file->f_op->write(file, buf, count, pos);
-		else
+		else if (file->f_op->aio_write)
 			ret = do_sync_write(file, buf, count, pos);
+		else
+			ret = new_sync_write(file, buf, count, pos);
 		if (ret > 0) {
 			fsnotify_modify(file);
 			add_wchar(current, ret);
@@ -601,6 +650,25 @@ unsigned long iov_shorten(struct iovec *iov, unsigned long nr_segs, size_t to)
 }
 EXPORT_SYMBOL(iov_shorten);
 
+static ssize_t do_iter_readv_writev(struct file *filp, int rw, const struct iovec *iov,
+		unsigned long nr_segs, size_t len, loff_t *ppos, iter_fn_t fn)
+{
+	struct kiocb kiocb;
+	struct iov_iter iter;
+	ssize_t ret;
+
+	init_sync_kiocb(&kiocb, filp);
+	kiocb.ki_pos = *ppos;
+	kiocb.ki_nbytes = len;
+
+	iov_iter_init(&iter, rw, iov, nr_segs, len);
+	ret = fn(&kiocb, &iter);
+	if (ret == -EIOCBQUEUED)
+		ret = wait_on_sync_kiocb(&kiocb);
+	*ppos = kiocb.ki_pos;
+	return ret;
+}
+
 static ssize_t do_sync_readv_writev(struct file *filp, const struct iovec *iov,
 		unsigned long nr_segs, size_t len, loff_t *ppos, iov_fn_t fn)
 {
@@ -738,6 +806,7 @@ static ssize_t do_readv_writev(int type, struct file *file,
 	ssize_t ret;
 	io_fn_t fn;
 	iov_fn_t fnv;
+	iter_fn_t iter_fn;
 
 	ret = rw_copy_check_uvector(type, uvector, nr_segs,
 				    ARRAY_SIZE(iovstack), iovstack, &iov);
@@ -753,13 +822,18 @@ static ssize_t do_readv_writev(int type, struct file *file,
 	if (type == READ) {
 		fn = file->f_op->read;
 		fnv = file->f_op->aio_read;
+		iter_fn = file->f_op->read_iter;
 	} else {
 		fn = (io_fn_t)file->f_op->write;
 		fnv = file->f_op->aio_write;
+		iter_fn = file->f_op->write_iter;
 		file_start_write(file);
 	}
 
-	if (fnv)
+	if (iter_fn)
+		ret = do_iter_readv_writev(file, type, iov, nr_segs, tot_len,
+						pos, iter_fn);
+	else if (fnv)
 		ret = do_sync_readv_writev(file, iov, nr_segs, tot_len,
 						pos, fnv);
 	else
@@ -912,6 +986,7 @@ static ssize_t compat_do_readv_writev(int type, struct file *file,
 	ssize_t ret;
 	io_fn_t fn;
 	iov_fn_t fnv;
+	iter_fn_t iter_fn;
 
 	ret = compat_rw_copy_check_uvector(type, uvector, nr_segs,
 					       UIO_FASTIOV, iovstack, &iov);
@@ -927,13 +1002,18 @@ static ssize_t compat_do_readv_writev(int type, struct file *file,
 	if (type == READ) {
 		fn = file->f_op->read;
 		fnv = file->f_op->aio_read;
+		iter_fn = file->f_op->read_iter;
 	} else {
 		fn = (io_fn_t)file->f_op->write;
 		fnv = file->f_op->aio_write;
+		iter_fn = file->f_op->write_iter;
 		file_start_write(file);
 	}
 
-	if (fnv)
+	if (iter_fn)
+		ret = do_iter_readv_writev(file, type, iov, nr_segs, tot_len,
+						pos, iter_fn);
+	else if (fnv)
 		ret = do_sync_readv_writev(file, iov, nr_segs, tot_len,
 						pos, fnv);
 	else

include/linux/fs.h

@@ -1451,6 +1451,8 @@ struct block_device_operations;
 #define HAVE_COMPAT_IOCTL 1
 #define HAVE_UNLOCKED_IOCTL 1
 
+struct iov_iter;
+
 struct file_operations {
 	struct module *owner;
 	loff_t (*llseek) (struct file *, loff_t, int);
@@ -1458,6 +1460,8 @@ struct file_operations {
 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
+	ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
+	ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
 	int (*iterate) (struct file *, struct dir_context *);
 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
@@ -2415,6 +2419,8 @@ extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff
 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
+extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
+extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
 
 /* fs/block_dev.c */
 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,