Introduces the FOP_CROSS_TYPE_COPYFILE flag, and implements it for
overlayfs. This enables copy_file_range between an overlayfs mount and
its constituent layers.
Signed-off-by: Han-Wen Nienhuys <hanwen@xxxxxxxxxxx>
---
fs/overlayfs/file.c | 60 ++++++++++++++++++++++++++++++---------------
fs/read_write.c | 15 ++++++++----
include/linux/fs.h | 4 +++
3 files changed, 54 insertions(+), 25 deletions(-)
diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c
index 969b458100fe..97b394737251 100644
--- a/fs/overlayfs/file.c
+++ b/fs/overlayfs/file.c
@@ -536,6 +536,9 @@ static int ovl_fadvise(struct file *file, loff_t offset, loff_t len, int advice)
return ret;
}
+static ssize_t ovl_copy_file_range(struct file *file_in, loff_t pos_in,
+ struct file *file_out, loff_t pos_out,
+ size_t len, unsigned int flags);
enum ovl_copyop {
OVL_COPY,
OVL_CLONE,
@@ -547,30 +550,42 @@ static loff_t ovl_copyfile(struct file *file_in, loff_t pos_in,
loff_t len, unsigned int flags, enum ovl_copyop op)
{
struct inode *inode_out = file_inode(file_out);
- struct file *realfile_in, *realfile_out;
+ struct file *realfile_in = file_in;
+ struct file *realfile_out = file_out;
const struct cred *old_cred;
loff_t ret;
+ bool in_overlay = file_in->f_op->copy_file_range == &ovl_copy_file_range;
+ bool out_overlay = file_out->f_op->copy_file_range == &ovl_copy_file_range;
- inode_lock(inode_out);
- if (op != OVL_DEDUPE) {
- /* Update mode */
- ovl_copyattr(inode_out);
- ret = file_remove_privs(file_out);
- if (ret)
- goto out_unlock;
+ if (WARN_ON_ONCE(!in_overlay && !out_overlay))
+ return -EXDEV;
+
+ if (in_overlay) {
+ realfile_in = ovl_real_file(file_in);
+ ret = PTR_ERR(realfile_in);
+ if (IS_ERR(realfile_in))
+ return ret;
}
- realfile_out = ovl_real_file(file_out);
- ret = PTR_ERR(realfile_out);
- if (IS_ERR(realfile_out))
- goto out_unlock;
+ if (out_overlay) {
+ inode_lock(inode_out);
- realfile_in = ovl_real_file(file_in);
- ret = PTR_ERR(realfile_in);
- if (IS_ERR(realfile_in))
- goto out_unlock;
+ if (op != OVL_DEDUPE) {
+ /* Update mode */
+ ovl_copyattr(inode_out);
+ ret = file_remove_privs(file_out);
+ if (ret)
+ goto out_unlock;
+ }
+
+ realfile_out = ovl_real_file(file_out);
+ ret = PTR_ERR(realfile_out);
+ if (IS_ERR(realfile_out))
+ goto out_unlock;
+
+ old_cred = ovl_override_creds(file_inode(file_out)->i_sb);
+ }
- old_cred = ovl_override_creds(file_inode(file_out)->i_sb);
switch (op) {
case OVL_COPY:
ret = vfs_copy_file_range(realfile_in, pos_in,
@@ -588,13 +603,16 @@ static loff_t ovl_copyfile(struct file *file_in, loff_t pos_in,
flags);
break;
}
- ovl_revert_creds(old_cred);
/* Update size */
- ovl_file_modified(file_out);
+ if (out_overlay) {
+ ovl_file_modified(file_out);
+ ovl_revert_creds(old_cred);
+ }
out_unlock:
- inode_unlock(inode_out);
+ if (out_overlay)
+ inode_unlock(inode_out);
return ret;
}
@@ -654,6 +672,8 @@ static int ovl_flush(struct file *file, fl_owner_t id)
}
const struct file_operations ovl_file_operations = {
+ .fop_flags = FOP_CROSS_TYPE_COPYFILE,
+
.open = ovl_open,
.release = ovl_release,
.llseek = ovl_llseek,
diff --git a/fs/read_write.c b/fs/read_write.c
index a6133241dfb8..93618441a02d 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -1489,7 +1489,7 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
* We allow some filesystems to handle cross sb copy, but passing
* a file of the wrong filesystem type to filesystem driver can result
* in an attempt to dereference the wrong type of ->private_data, so
- * avoid doing that until we really have a good reason.
+ * avoid doing unless FOP_CROSS_TYPE_COPYFILE is set.
*
* nfs and cifs define several different file_system_type structures
* and several different sets of file_operations, but they all end up
@@ -1497,6 +1497,9 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
*/
if (flags & COPY_FILE_SPLICE) {
/* cross sb splice is allowed */
+ } else if (file_in->f_op->fop_flags & FOP_CROSS_TYPE_COPYFILE ||
+ file_out->f_op->fop_flags & FOP_CROSS_TYPE_COPYFILE) {
+ /* file system understands how to cross FS types */
} else if (file_out->f_op->copy_file_range) {
if (file_in->f_op->copy_file_range !=
file_out->f_op->copy_file_range)
@@ -1576,10 +1579,12 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
* same sb using clone, but for filesystems where both clone and copy
* are supported (e.g. nfs,cifs), we only call the copy method.
*/
- if (!splice && file_out->f_op->copy_file_range) {
- ret = file_out->f_op->copy_file_range(file_in, pos_in,
- file_out, pos_out,
- len, flags);
+ if (!splice && (file_in->f_op->copy_file_range || file_out->f_op->copy_file_range)) {
+ ret = (file_in->f_op->copy_file_range ?
+ file_in->f_op->copy_file_range :
+ file_out->f_op->copy_file_range)(file_in, pos_in,
+ file_out, pos_out,
+ len, flags);
} else if (!splice && file_in->f_op->remap_file_range && samesb) {
ret = file_in->f_op->remap_file_range(file_in, pos_in,
file_out, pos_out,
diff --git a/include/linux/fs.h b/include/linux/fs.h
index a4af70367f8a..1248a2542758 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2143,6 +2143,10 @@ struct file_operations {
#define FOP_ASYNC_LOCK ((__force fop_flags_t)(1 << 6))
/* File system supports uncached read/write buffered IO */
#define FOP_DONTCACHE ((__force fop_flags_t)(1 << 7))
+/* copy_file_range accepts source and destination on different types of file
+ * system. If set, file_operations.copy_file_range must also be set.
+ */
+#define FOP_CROSS_TYPE_COPYFILE ((__force fop_flags_t)(1 << 8))
/* Wrap a directory iterator that needs exclusive inode access */
int wrap_directory_iterator(struct file *, struct dir_context *,
--
2.43.0
