[PATCH 17/18] ext2 xip: replace sendfile with splice

[Jens Axboe <jens.axboe@oracle.com>]

From: Carsten Otte <cotte@de.ibm.com>

This patch removes xip_file_sendfile, and replaces it with
xip_file_splice_read. It passes the ltp testcase on a ext2 filesystem
that is mounted -o xip,rw.
I am not exactly happy with this patch, because it introduces waaaay too
much code duplication from fs/splice.c. I want to dig into making
generic_file_splice_read able to deal with xip files aswell. That would
allow to do the same for splice write.
If I cannot find a nicer solution, I'd vote for removing sendfile from
filemap_xip without replacement: a xip filesystem is intended to be used
for executables and libraries. This content is very rarely subject to
splice.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 fs/ext2/file.c     |    2 +-
 include/linux/fs.h |    4 +-
 mm/filemap_xip.c   |  206 ++++++++++++++++++++++++++++++++++++++++++++++++----
 3 files changed, 192 insertions(+), 20 deletions(-)

diff --git a/fs/ext2/file.c b/fs/ext2/file.c
index 072a190..5f4a17c 100644
--- a/fs/ext2/file.c
+++ b/fs/ext2/file.c
@@ -70,7 +70,7 @@ const struct file_operations ext2_xip_file_operations = {
 	.open		= generic_file_open,
 	.release	= ext2_release_file,
 	.fsync		= ext2_sync_file,
-	.sendfile	= xip_file_sendfile,
+	.splice_read	= xip_file_splice_read,
 };
 #endif
 
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 61ebf32..a5bf63e 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1764,13 +1764,11 @@ extern int nonseekable_open(struct inode * inode, struct file * filp);
 #ifdef CONFIG_FS_XIP
 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
 			     loff_t *ppos);
-extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
-				 size_t count, read_actor_t actor,
-				 void *target);
 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
 			      size_t len, loff_t *ppos);
 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
+extern ssize_t xip_file_splice_read(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
 #else
 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
 {
diff --git a/mm/filemap_xip.c b/mm/filemap_xip.c
index fa360e5..f1f3134 100644
--- a/mm/filemap_xip.c
+++ b/mm/filemap_xip.c
@@ -14,6 +14,7 @@
 #include <linux/uio.h>
 #include <linux/rmap.h>
 #include <linux/sched.h>
+#include <linux/splice.h>
 #include <asm/tlbflush.h>
 #include "filemap.h"
 
@@ -92,7 +93,11 @@ do_xip_mapping_read(struct address_space *mapping,
 		if (unlikely(IS_ERR(page))) {
 			if (PTR_ERR(page) == -ENODATA) {
 				/* sparse */
-				page = ZERO_PAGE(0);
+				page = xip_sparse_page();
+				if (!page) {
+					desc->error = -ENOMEM;
+					goto out;
+				}
 			} else {
 				desc->error = PTR_ERR(page);
 				goto out;
@@ -159,27 +164,196 @@ xip_file_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos)
 }
 EXPORT_SYMBOL_GPL(xip_file_read);
 
-ssize_t
-xip_file_sendfile(struct file *in_file, loff_t *ppos,
-	     size_t count, read_actor_t actor, void *target)
+static int xip_pipe_buf_steal(struct pipe_inode_info *pipe,
+				   struct pipe_buffer *buf)
 {
-	read_descriptor_t desc;
+	return 1;
+}
 
-	if (!count)
+static void xip_pipe_buf_release(struct pipe_inode_info *pipe,
+					struct pipe_buffer *buf)
+{
+	page_cache_release(buf->page);
+}
+
+static const struct pipe_buf_operations xip_pipe_buf_ops = {
+	.can_merge = 0,
+	.pin = generic_pipe_buf_pin,
+	.get = generic_pipe_buf_get,
+	.map = generic_pipe_buf_map,
+	.unmap = generic_pipe_buf_unmap,
+	.release = xip_pipe_buf_release,
+	.steal = xip_pipe_buf_steal,
+};
+
+static int
+__xip_file_splice_read(struct file *in, loff_t *ppos,
+			   struct pipe_inode_info *pipe, size_t len,
+			   unsigned int flags)
+{
+	struct address_space *mapping = in->f_mapping;
+	unsigned int loff, nr_pages;
+	struct page *pages[PIPE_BUFFERS];
+	struct partial_page partial[PIPE_BUFFERS];
+	struct page *page;
+	pgoff_t index, end_index;
+	loff_t isize;
+	size_t total_len;
+	int error, page_nr;
+	struct splice_pipe_desc spd = {
+		.pages = pages,
+		.partial = partial,
+		.flags = flags,
+		.ops = &xip_pipe_buf_ops,
+	};
+
+	index = *ppos >> PAGE_CACHE_SHIFT;
+	loff = *ppos & ~PAGE_CACHE_MASK;
+	nr_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+
+	if (nr_pages > PIPE_BUFFERS)
+		nr_pages = PIPE_BUFFERS;
+
+	error = 0;
+	total_len = 0;
+
+	spd.nr_pages = 0;
+	while (spd.nr_pages < nr_pages) {
+		page = mapping->a_ops->get_xip_page(mapping,
+			index*(PAGE_SIZE/512), 0);
+		if (!page) {
+			error = -EIO;
+			break;
+		}
+		if (unlikely(IS_ERR(page))) {
+			if (PTR_ERR(page) == -ENODATA) {
+				/* sparse */
+				page = xip_sparse_page();
+				if (!page) {
+					error = -ENOMEM;
+					break;
+				}
+			} else {
+				error = PTR_ERR(page);
+				break;
+			}
+		}
+		page_cache_get(page);
+		pages[spd.nr_pages++] = page;
+		index++;
+	}
+
+	/*
+	 * Now loop over the map and fill in the partial map
+	 */
+	index = *ppos >> PAGE_CACHE_SHIFT;
+	nr_pages = spd.nr_pages;
+	spd.nr_pages = 0;
+	for (page_nr = 0; page_nr < nr_pages; page_nr++) {
+		unsigned int this_len;
+
+		if (!len)
+			break;
+
+		/*
+		 * this_len is the max we'll use from this page
+		 */
+		this_len = min_t(unsigned long, len, PAGE_CACHE_SIZE - loff);
+		page = pages[page_nr];
+		isize = i_size_read(mapping->host);
+		end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
+		if (unlikely(!isize || index > end_index))
+			break;
+
+		/*
+		 * if this is the last page, see if we need to shrink
+		 * the length and stop
+		 */
+		if (end_index == index) {
+			unsigned int plen;
+
+			/*
+			 * max good bytes in this page
+			 */
+			plen = ((isize - 1) & ~PAGE_CACHE_MASK) + 1;
+			if (plen <= loff)
+				break;
+
+			/*
+			 * force quit after adding this page
+			 */
+			this_len = min(this_len, plen - loff);
+			len = this_len;
+		}
+
+		partial[page_nr].offset = loff;
+		partial[page_nr].len = this_len;
+		len -= this_len;
+		loff = 0;
+		spd.nr_pages++;
+		index++;
+	}
+
+	if (spd.nr_pages)
+		return splice_to_pipe(pipe, &spd);
+
+	return error;
+}
+
+/**
+ * xip_file_splice_read - splice data from file to a pipe
+ * @in:		file to splice from
+ * @pipe:	pipe to splice to
+ * @len:	number of bytes to splice
+ * @flags:	splice modifier flags
+ *
+ * derived from generic_file_splice_read in fs/splice_read.c
+ */
+ssize_t xip_file_splice_read(struct file *in, loff_t *ppos,
+				 struct pipe_inode_info *pipe, size_t len,
+				 unsigned int flags)
+{
+	ssize_t spliced;
+	int ret;
+	loff_t isize, left;
+
+	BUG_ON(!in->f_mapping->a_ops->get_xip_page);
+
+	isize = i_size_read(in->f_mapping->host);
+	if (unlikely(*ppos >= isize))
 		return 0;
 
-	desc.written = 0;
-	desc.count = count;
-	desc.arg.data = target;
-	desc.error = 0;
+	left = isize - *ppos;
+	if (unlikely(left < len))
+		len = left;
 
-	do_xip_mapping_read(in_file->f_mapping, &in_file->f_ra, in_file,
-			    ppos, &desc, actor);
-	if (desc.written)
-		return desc.written;
-	return desc.error;
+	ret = 0;
+	spliced = 0;
+	while (len) {
+		ret = __xip_file_splice_read(in, ppos, pipe, len, flags);
+
+		if (ret < 0)
+			break;
+		else if (!ret) {
+			if (spliced)
+				break;
+			if (flags & SPLICE_F_NONBLOCK) {
+				ret = -EAGAIN;
+				break;
+			}
+		}
+
+		*ppos += ret;
+		len -= ret;
+		spliced += ret;
+	}
+
+	if (spliced)
+		return spliced;
+
+	return ret;
 }
-EXPORT_SYMBOL_GPL(xip_file_sendfile);
+EXPORT_SYMBOL(xip_file_splice_read);
 
 /*
  * __xip_unmap is invoked from xip_unmap and
-- 
1.5.2.1.174.gcd03