#include <linux/syscalls.h>
#include <linux/uio.h>
-/*
- * Passed to the actors
- */
-struct splice_desc {
- unsigned int len, total_len; /* current and remaining length */
- unsigned int flags; /* splice flags */
- struct file *file; /* file to read/write */
- loff_t pos; /* file position */
-};
-
struct partial_page {
unsigned int offset;
unsigned int len;
};
/*
- * Passed to move_to_pipe
+ * Passed to splice_to_pipe
*/
struct splice_pipe_desc {
struct page **pages; /* page map */
return 1;
}
- buf->flags |= PIPE_BUF_FLAG_STOLEN | PIPE_BUF_FLAG_LRU;
+ buf->flags |= PIPE_BUF_FLAG_LRU;
return 0;
}
{
page_cache_release(buf->page);
buf->page = NULL;
- buf->flags &= ~(PIPE_BUF_FLAG_STOLEN | PIPE_BUF_FLAG_LRU);
+ buf->flags &= ~PIPE_BUF_FLAG_LRU;
}
-static void *page_cache_pipe_buf_map(struct file *file,
- struct pipe_inode_info *info,
- struct pipe_buffer *buf)
+static int page_cache_pipe_buf_pin(struct pipe_inode_info *info,
+ struct pipe_buffer *buf)
{
struct page *page = buf->page;
int err;
}
/*
- * Page is ok afterall, fall through to mapping.
+ * Page is ok afterall, we are done.
*/
unlock_page(page);
}
- return kmap(page);
+ return 0;
error:
unlock_page(page);
- return ERR_PTR(err);
-}
-
-static void page_cache_pipe_buf_unmap(struct pipe_inode_info *info,
- struct pipe_buffer *buf)
-{
- kunmap(buf->page);
-}
-
-static void *user_page_pipe_buf_map(struct file *file,
- struct pipe_inode_info *pipe,
- struct pipe_buffer *buf)
-{
- return kmap(buf->page);
-}
-
-static void user_page_pipe_buf_unmap(struct pipe_inode_info *pipe,
- struct pipe_buffer *buf)
-{
- kunmap(buf->page);
-}
-
-static void page_cache_pipe_buf_get(struct pipe_inode_info *info,
- struct pipe_buffer *buf)
-{
- page_cache_get(buf->page);
+ return err;
}
static struct pipe_buf_operations page_cache_pipe_buf_ops = {
.can_merge = 0,
- .map = page_cache_pipe_buf_map,
- .unmap = page_cache_pipe_buf_unmap,
+ .map = generic_pipe_buf_map,
+ .unmap = generic_pipe_buf_unmap,
+ .pin = page_cache_pipe_buf_pin,
.release = page_cache_pipe_buf_release,
.steal = page_cache_pipe_buf_steal,
- .get = page_cache_pipe_buf_get,
+ .get = generic_pipe_buf_get,
};
static int user_page_pipe_buf_steal(struct pipe_inode_info *pipe,
static struct pipe_buf_operations user_page_pipe_buf_ops = {
.can_merge = 0,
- .map = user_page_pipe_buf_map,
- .unmap = user_page_pipe_buf_unmap,
+ .map = generic_pipe_buf_map,
+ .unmap = generic_pipe_buf_unmap,
+ .pin = generic_pipe_buf_pin,
.release = page_cache_pipe_buf_release,
.steal = user_page_pipe_buf_steal,
- .get = page_cache_pipe_buf_get,
+ .get = generic_pipe_buf_get,
};
/*
* Pipe output worker. This sets up our pipe format with the page cache
* pipe buffer operations. Otherwise very similar to the regular pipe_writev().
*/
-static ssize_t move_to_pipe(struct pipe_inode_info *pipe,
- struct splice_pipe_desc *spd)
+static ssize_t splice_to_pipe(struct pipe_inode_info *pipe,
+ struct splice_pipe_desc *spd)
{
int ret, do_wakeup, page_nr;
pgoff_t index, end_index;
loff_t isize;
size_t total_len;
- int error;
+ int error, page_nr;
struct splice_pipe_desc spd = {
.pages = pages,
.partial = partial,
* read-ahead if this is a non-zero offset (we are likely doing small
* chunk splice and the page is already there) for a single page.
*/
- if (!loff || spd.nr_pages > 1)
- do_page_cache_readahead(mapping, in, index, spd.nr_pages);
+ if (!loff || nr_pages > 1)
+ page_cache_readahead(mapping, &in->f_ra, in, index, nr_pages);
/*
* Now fill in the holes:
*/
error = 0;
total_len = 0;
- for (spd.nr_pages = 0; spd.nr_pages < nr_pages; spd.nr_pages++, index++) {
- unsigned int this_len;
- if (!len)
- break;
+ /*
+ * Lookup the (hopefully) full range of pages we need.
+ */
+ spd.nr_pages = find_get_pages_contig(mapping, index, nr_pages, pages);
+ /*
+ * If find_get_pages_contig() returned fewer pages than we needed,
+ * allocate the rest.
+ */
+ index += spd.nr_pages;
+ while (spd.nr_pages < nr_pages) {
/*
- * this_len is the max we'll use from this page
- */
- this_len = min_t(unsigned long, len, PAGE_CACHE_SIZE - loff);
-find_page:
- /*
- * lookup the page for this index
+ * Page could be there, find_get_pages_contig() breaks on
+ * the first hole.
*/
page = find_get_page(mapping, index);
if (!page) {
/*
- * page didn't exist, allocate one
+ * page didn't exist, allocate one.
*/
page = page_cache_alloc_cold(mapping);
if (!page)
break;
error = add_to_page_cache_lru(page, mapping, index,
- mapping_gfp_mask(mapping));
+ mapping_gfp_mask(mapping));
if (unlikely(error)) {
page_cache_release(page);
break;
}
-
- goto readpage;
+ /*
+ * add_to_page_cache() locks the page, unlock it
+ * to avoid convoluting the logic below even more.
+ */
+ unlock_page(page);
}
+ pages[spd.nr_pages++] = page;
+ index++;
+ }
+
+ /*
+ * Now loop over the map and see if we need to start IO on any
+ * pages, fill in the partial map, etc.
+ */
+ 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];
+
/*
* If the page isn't uptodate, we may need to start io on it
*/
*/
if (!page->mapping) {
unlock_page(page);
- page_cache_release(page);
break;
}
/*
goto fill_it;
}
-readpage:
/*
* need to read in the page
*/
error = mapping->a_ops->readpage(in, page);
-
if (unlikely(error)) {
- page_cache_release(page);
+ /*
+ * We really should re-lookup the page here,
+ * but it complicates things a lot. Instead
+ * lets just do what we already stored, and
+ * we'll get it the next time we are called.
+ */
if (error == AOP_TRUNCATED_PAGE)
- goto find_page;
+ error = 0;
+
break;
}
*/
isize = i_size_read(mapping->host);
end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
- if (unlikely(!isize || index > end_index)) {
- page_cache_release(page);
+ if (unlikely(!isize || index > end_index))
break;
- }
/*
* if this is the last page, see if we need to shrink
*/
if (end_index == index) {
loff = PAGE_CACHE_SIZE - (isize & ~PAGE_CACHE_MASK);
- if (total_len + loff > isize) {
- page_cache_release(page);
+ if (total_len + loff > isize)
break;
- }
/*
* force quit after adding this page
*/
- nr_pages = spd.nr_pages;
+ len = this_len;
this_len = min(this_len, loff);
loff = 0;
}
}
fill_it:
- pages[spd.nr_pages] = page;
- partial[spd.nr_pages].offset = loff;
- partial[spd.nr_pages].len = this_len;
+ partial[page_nr].offset = loff;
+ partial[page_nr].len = this_len;
len -= this_len;
total_len += this_len;
loff = 0;
+ spd.nr_pages++;
+ index++;
}
+ /*
+ * Release any pages at the end, if we quit early. 'i' is how far
+ * we got, 'nr_pages' is how many pages are in the map.
+ */
+ while (page_nr < nr_pages)
+ page_cache_release(pages[page_nr++]);
+
if (spd.nr_pages)
- return move_to_pipe(pipe, &spd);
+ return splice_to_pipe(pipe, &spd);
return error;
}
{
struct file *file = sd->file;
loff_t pos = sd->pos;
- ssize_t ret;
- void *ptr;
- int more;
+ int ret, more;
- /*
- * Sub-optimal, but we are limited by the pipe ->map. We don't
- * need a kmap'ed buffer here, we just want to make sure we
- * have the page pinned if the pipe page originates from the
- * page cache.
- */
- ptr = buf->ops->map(file, info, buf);
- if (IS_ERR(ptr))
- return PTR_ERR(ptr);
-
- more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
+ ret = buf->ops->pin(info, buf);
+ if (!ret) {
+ more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
- ret = file->f_op->sendpage(file, buf->page, buf->offset, sd->len,
- &pos, more);
+ ret = file->f_op->sendpage(file, buf->page, buf->offset,
+ sd->len, &pos, more);
+ }
- buf->ops->unmap(info, buf);
return ret;
}
unsigned int offset, this_len;
struct page *page;
pgoff_t index;
- char *src;
int ret;
/*
* make sure the data in this buffer is uptodate
*/
- src = buf->ops->map(file, info, buf);
- if (IS_ERR(src))
- return PTR_ERR(src);
+ ret = buf->ops->pin(info, buf);
+ if (unlikely(ret))
+ return ret;
index = sd->pos >> PAGE_CACHE_SHIFT;
offset = sd->pos & ~PAGE_CACHE_MASK;
this_len = PAGE_CACHE_SIZE - offset;
/*
- * Reuse buf page, if SPLICE_F_MOVE is set.
+ * Reuse buf page, if SPLICE_F_MOVE is set and we are doing a full
+ * page.
*/
- if (sd->flags & SPLICE_F_MOVE) {
+ if ((sd->flags & SPLICE_F_MOVE) && this_len == PAGE_CACHE_SIZE) {
/*
* If steal succeeds, buf->page is now pruned from the vm
* side (LRU and page cache) and we can reuse it. The page
goto find_page;
page = buf->page;
- if (add_to_page_cache(page, mapping, index, gfp_mask))
+ if (add_to_page_cache(page, mapping, index, gfp_mask)) {
+ unlock_page(page);
goto find_page;
+ }
+
+ page_cache_get(page);
if (!(buf->flags & PIPE_BUF_FLAG_LRU))
lru_cache_add(page);
} else if (ret)
goto out;
- if (!(buf->flags & PIPE_BUF_FLAG_STOLEN)) {
- char *dst = kmap_atomic(page, KM_USER0);
+ if (buf->page != page) {
+ /*
+ * Careful, ->map() uses KM_USER0!
+ */
+ char *src = buf->ops->map(info, buf);
+ char *dst = kmap_atomic(page, KM_USER1);
memcpy(dst + offset, src + buf->offset, this_len);
flush_dcache_page(page);
- kunmap_atomic(dst, KM_USER0);
+ kunmap_atomic(dst, KM_USER1);
+ buf->ops->unmap(info, buf);
}
ret = mapping->a_ops->commit_write(file, page, offset, offset+this_len);
- if (ret == AOP_TRUNCATED_PAGE) {
+ if (!ret) {
+ /*
+ * Return the number of bytes written and mark page as
+ * accessed, we are now done!
+ */
+ ret = this_len;
+ mark_page_accessed(page);
+ balance_dirty_pages_ratelimited(mapping);
+ } else if (ret == AOP_TRUNCATED_PAGE) {
page_cache_release(page);
goto find_page;
- } else if (ret)
- goto out;
-
- /*
- * Return the number of bytes written.
- */
- ret = this_len;
- mark_page_accessed(page);
- balance_dirty_pages_ratelimited(mapping);
+ }
out:
- if (!(buf->flags & PIPE_BUF_FLAG_STOLEN))
- page_cache_release(page);
-
+ page_cache_release(page);
unlock_page(page);
out_nomem:
- buf->ops->unmap(info, buf);
return ret;
}
-typedef int (splice_actor)(struct pipe_inode_info *, struct pipe_buffer *,
- struct splice_desc *);
-
/*
* Pipe input worker. Most of this logic works like a regular pipe, the
* key here is the 'actor' worker passed in that actually moves the data
* to the wanted destination. See pipe_to_file/pipe_to_sendpage above.
*/
-static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out,
- loff_t *ppos, size_t len, unsigned int flags,
- splice_actor *actor)
+ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
+ loff_t *ppos, size_t len, unsigned int flags,
+ splice_actor *actor)
{
int ret, do_wakeup, err;
struct splice_desc sd;
struct address_space *mapping = out->f_mapping;
ssize_t ret;
- ret = move_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
+ ret = splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file);
if (ret > 0) {
struct inode *inode = mapping->host;
ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, struct file *out,
loff_t *ppos, size_t len, unsigned int flags)
{
- return move_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage);
+ return splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage);
}
EXPORT_SYMBOL(generic_splice_sendpage);
/*
* We don't have an immediate reader, but we'll read the stuff
- * out of the pipe right after the move_to_pipe(). So set
+ * out of the pipe right after the splice_to_pipe(). So set
* PIPE_READERS appropriately.
*/
pipe->readers = 1;
if (spd.nr_pages <= 0)
return spd.nr_pages;
- return move_to_pipe(pipe, &spd);
+ return splice_to_pipe(pipe, &spd);
}
asmlinkage long sys_vmsplice(int fd, const struct iovec __user *iov,