Hello, developers.
sock_sendfile() and generic_file_sendpage() were implemented
and presented in the attached patch.
Such methods allows to use sendfile() for any file descriptor <-> file
descriptor usage, especially usefull it is in the case socket -> file,
when there are no copy_from_user() cases when writing the data.
I do not have nice hairy beard which all VFS developers must have,
so probably something is completely incorrect, but it passed my tests
with ext3 filesystem.
Signed-off-by: Evgeniy Polyakov <[EMAIL PROTECTED]>
diff --git a/fs/ext3/file.c b/fs/ext3/file.c
--- a/fs/ext3/file.c
+++ b/fs/ext3/file.c
@@ -119,6 +119,7 @@ struct file_operations ext3_file_operati
.release = ext3_release_file,
.fsync = ext3_sync_file,
.sendfile = generic_file_sendfile,
+ .sendpage = generic_file_sendpage,
};
struct inode_operations ext3_file_inode_operations = {
diff --git a/fs/read_write.c b/fs/read_write.c
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -14,6 +14,13 @@
#include <linux/security.h>
#include <linux/module.h>
#include <linux/syscalls.h>
+#include <linux/mm.h>
+#include <linux/aio.h>
+#include <linux/swap.h>
+#include <linux/mman.h>
+#include <linux/pagemap.h>
+#include <linux/writeback.h>
+#include <linux/pagevec.h>
#include <asm/uaccess.h>
#include <asm/unistd.h>
@@ -325,6 +332,81 @@ static inline void file_pos_write(struct
file->f_pos = pos;
}
+extern struct page * __grab_cache_page(struct address_space *, unsigned long,
struct page **, struct pagevec *);
+
+ssize_t generic_file_sendpage(struct file *file, struct page *in_page, int
offset, size_t size, loff_t *ppos, int more)
+{
+ ssize_t err;
+ struct address_space * mapping = file->f_mapping;
+ struct address_space_operations *a_ops = mapping->a_ops;
+ struct inode *inode = mapping->host;
+ struct page *out_page;
+ struct page *cached_page = NULL;
+ struct pagevec lru_pvec;
+ unsigned long index;
+ unsigned long page_offset;
+ unsigned long bytes;
+ size_t written;
+ loff_t pos = *ppos;
+
+ err = generic_write_checks(file, &pos, &size, S_ISBLK(inode->i_mode));
+ if (err)
+ goto err_out_exit;
+
+ pagevec_init(&lru_pvec, 0);
+
+ written = 0;
+
+ while (size) {
+ page_offset = (pos & (PAGE_CACHE_SIZE -1));
+ index = pos >> PAGE_CACHE_SHIFT;
+ bytes = PAGE_CACHE_SIZE - page_offset;
+ if (bytes > size)
+ bytes = size;
+
+ out_page = __grab_cache_page(mapping, index, &cached_page,
&lru_pvec);
+ if (!out_page) {
+ err = -ENOMEM;
+ goto err_out_exit;
+ }
+
+ err = a_ops->prepare_write(file, out_page, page_offset,
page_offset+bytes);
+ if (unlikely(err))
+ goto err_out_unlock;
+
+ memcpy(page_address(out_page)+page_offset,
page_address(in_page)+offset, bytes);
+
+ flush_dcache_page(out_page);
+ err = a_ops->commit_write(file, out_page, page_offset,
page_offset+bytes);
+ unlock_page(out_page);
+ mark_page_accessed(out_page);
+ page_cache_release(out_page);
+
+ if (err < 0)
+ goto err_out_exit;
+
+ balance_dirty_pages_ratelimited(mapping);
+
+ size -= bytes;
+ written += bytes;
+ }
+
+ if (cached_page)
+ page_cache_release(cached_page);
+
+ pagevec_lru_add(&lru_pvec);
+ file_pos_write(file, pos + written);
+
+ return written;
+
+err_out_unlock:
+ unlock_page(out_page);
+ page_cache_release(out_page);
+err_out_exit:
+
+ return err;
+}
+
asmlinkage ssize_t sys_read(unsigned int fd, char __user * buf, size_t count)
{
struct file *file;
@@ -667,8 +749,10 @@ static ssize_t do_sendfile(int out_fd, i
if (!(out_file->f_mode & FMODE_WRITE))
goto fput_out;
retval = -EINVAL;
- if (!out_file->f_op || !out_file->f_op->sendpage)
+ if (!out_file->f_op || !out_file->f_op->sendpage) {
+ printk("%s: out_file->f_op->sendpage=%p.\n", __func__,
out_file->f_op->sendpage);
goto fput_out;
+ }
out_inode = out_file->f_dentry->d_inode;
retval = rw_verify_area(WRITE, out_file, &out_file->f_pos, count);
if (retval)
@@ -685,7 +769,7 @@ static ssize_t do_sendfile(int out_fd, i
retval = -EINVAL;
if (unlikely(pos < 0))
goto fput_out;
- if (unlikely(pos + count > max)) {
+ if (unlikely((unsigned long long)(pos + count) > (unsigned long
long)max)) {
retval = -EOVERFLOW;
if (pos >= max)
goto fput_out;
diff --git a/include/linux/fs.h b/include/linux/fs.h
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1511,6 +1511,7 @@ extern ssize_t do_sync_write(struct file
ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
unsigned long nr_segs, loff_t *ppos);
extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t,
read_actor_t, void *);
+extern ssize_t generic_file_sendpage(struct file *, struct page *, int,
size_t, loff_t *, int);
extern void do_generic_mapping_read(struct address_space *mapping,
struct file_ra_state *, struct file *,
loff_t *, read_descriptor_t *,
read_actor_t);
diff --git a/mm/filemap.c b/mm/filemap.c
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1653,7 +1653,7 @@ EXPORT_SYMBOL(read_cache_page);
* caller's lru-buffering pagevec. This function is specifically for
* generic_file_write().
*/
-static inline struct page *
+struct page *
__grab_cache_page(struct address_space *mapping, unsigned long index,
struct page **cached_page, struct pagevec *lru_pvec)
{
@@ -1682,6 +1682,8 @@ repeat:
return page;
}
+EXPORT_SYMBOL(__grab_cache_page);
+
/*
* The logic we want is
*
diff --git a/net/socket.c b/net/socket.c
--- a/net/socket.c
+++ b/net/socket.c
@@ -44,6 +44,7 @@
* Tigran Aivazian : sys_send(args) calls sys_sendto(args,
NULL, 0)
* Tigran Aivazian : Made listen(2) backlog sanity checks
* protocol-independent
+ * Evgeniy Polyakov: Added
sock_sendfile()/generic_file_sendpage().
*
*
* This program is free software; you can redistribute it and/or
@@ -114,6 +115,7 @@ static ssize_t sock_writev(struct file *
unsigned long count, loff_t *ppos);
static ssize_t sock_sendpage(struct file *file, struct page *page,
int offset, size_t size, loff_t *ppos, int more);
+ssize_t sock_sendfile(struct file *file, loff_t *ppos, size_t count,
read_actor_t actor, void *target);
/*
@@ -134,7 +136,8 @@ static struct file_operations socket_fil
.fasync = sock_fasync,
.readv = sock_readv,
.writev = sock_writev,
- .sendpage = sock_sendpage
+ .sendpage = sock_sendpage,
+ .sendfile = sock_sendfile,
};
/*
@@ -720,6 +723,82 @@ static ssize_t sock_aio_write(struct kio
return __sock_sendmsg(iocb, sock, &x->async_msg, size);
}
+ssize_t sock_sendfile(struct file *file, loff_t *ppos, size_t count,
read_actor_t actor, void *target)
+{
+ struct socket *sock;
+ struct page *page;
+ int err = 0;
+ struct msghdr msg;
+ struct kvec iov;
+ size_t size;
+ read_descriptor_t desc;
+ unsigned long offset = 0;
+ size_t written = 0;
+
+ if (!count)
+ return 0;
+
+ if (!ppos != 0)
+ return -ERANGE;
+
+ desc.written = 0;
+ desc.count = count;
+ desc.arg.data = target;
+ desc.error = 0;
+
+ page = alloc_page(GFP_KERNEL);
+ if (!page)
+ return -ENOMEM;
+
+ sock = SOCKET_I(file->f_dentry->d_inode);
+
+ while (count) {
+
+ size = min(count, PAGE_SIZE);
+
+ sock->sk->sk_allocation |= GFP_NOIO;
+ iov.iov_base = page_address(page);
+ iov.iov_len = size;
+ msg.msg_name = NULL;
+ msg.msg_namelen = 0;
+ msg.msg_control = NULL;
+ msg.msg_controllen = 0;
+ msg.msg_namelen = 0;
+ msg.msg_flags = MSG_NOSIGNAL;
+
+ memset(iov.iov_base, 0, PAGE_SIZE);
+
+ err = kernel_recvmsg(sock, &msg, &iov, 1, size, 0);
+
+ if (signal_pending(current)) {
+ flush_signals(current);
+ printk("Interrupted by signal\n");
+ return -ERESTARTSYS;
+ }
+
+ if (err < 0) {
+ printk("Failed to receive message: size=%zx,
err=%d.\n", iov.iov_len, err);
+ break;
+ }
+
+ if (err == 0)
+ break;
+
+ count -= err;
+
+ err = actor(&desc, page, offset, err);
+ *ppos += err;
+ written += err;
+ }
+
+ __free_pages(page, 0);
+
+ if (err)
+ return err;
+
+ return written;
+}
+
ssize_t sock_sendpage(struct file *file, struct page *page,
int offset, size_t size, loff_t *ppos, int more)
{
--
Evgeniy Polyakov
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
