vfs: allow copy_file_range to copy across devices

We want to enable cross-filesystem copy_file_range functionality where possible, so push the "same superblock only" checks down to the individual filesystem callouts so they can make their own decisions about cross-superblock copy offload and fallack to generic_copy_file_range() for cross-superblock copy.

[Amir] We do not call ->remap_file_range() in case the files are not on the same sb and do not call ->copy_file_range() in case the files do not belong to the same filesystem driver.

This changes behavior of the copy_file_range(2) syscall, which will now allow cross filesystem in-kernel copy. CIFS already supports cross-superblock copy, between two shares to the same server. This functionality will now be available via the copy_file_range(2) syscall.

Cc Steve French <[email protected]>

Signed-off-by Dave Chinner <[email protected]>

Signed-off-by Amir Goldstein <[email protected]>

---

Darrick,

Per feedback from Olga, I am sending a modified version of this patch to address cross file_system_type copy issue in nfs.

For the sake of global warming I am not re-posting the entire patch set. I removed your RVB because of the change.

Thanks,
Amir.

Changes since v4:
- Check "same filesystem driver" by comapring ->copy_file_range() function pointer

 fs/ceph/file.c    |  4 +++-
 fs/cifs/cifsfs.c  |  2 +-
 fs/fuse/file.c    |  5 ++++-
 fs/nfs/nfs4file.c |  5 ++++-
 fs/read_write.c   | 18 ++++++++++++------
 5 files changed, 24 insertions(+), 10 deletions(-)

fs/ceph/file.c +3 -1

index e87f7b2023af..4cd41ed5cc53 100644
@@ -1909,6 +1909,8 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
 
        if (src_inode == dst_inode)
                return -EINVAL;
+       if (src_inode->i_sb != dst_inode->i_sb)
+               return -EXDEV;
        if (ceph_snap(dst_inode) != CEPH_NOSNAP)
                return -EROFS;
 

@@ -2109,7 +2111,7 @@ static ssize_t ceph_copy_file_range(struct file *src_file, loff_t src_off,
        ret = __ceph_copy_file_range(src_file, src_off, dst_file, dst_off,
                                     len, flags);
 
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(src_file, src_off, dst_file,
                                              dst_off, len, flags);
        return ret;

fs/cifs/cifsfs.c +1 -1

index c65823270313..f11eea6125c1 100644
@@ -1149,7 +1149,7 @@ static ssize_t cifs_copy_file_range(struct file *src_file, loff_t off,
                                        len, flags);
        free_xid(xid);
 
-       if (rc == -EOPNOTSUPP)
+       if (rc == -EOPNOTSUPP || rc == -EXDEV)
                rc = generic_copy_file_range(src_file, off, dst_file,
                                             destoff, len, flags);
        return rc;

fs/fuse/file.c +4 -1

index e03901ae729b..569baf286835 100644
@@ -3126,6 +3126,9 @@ static ssize_t __fuse_copy_file_range(struct file *file_in, loff_t pos_in,
        if (fc->no_copy_file_range)
                return -EOPNOTSUPP;
 
+       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
+               return -EXDEV;
+
        inode_lock(inode_out);
 
        if (fc->writeback_cache) {

@@ -3182,7 +3185,7 @@ static ssize_t fuse_copy_file_range(struct file *src_file, loff_t src_off,
        ret = __fuse_copy_file_range(src_file, src_off, dst_file, dst_off,
                                     len, flags);
 
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(src_file, src_off, dst_file,
                                              dst_off, len, flags);
        return ret;

fs/nfs/nfs4file.c +4 -1

index 4842f3ab3161..f4157eb1f69d 100644
@@ -133,6 +133,9 @@ static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
                                      struct file *file_out, loff_t pos_out,
                                      size_t count, unsigned int flags)
 {
+       /* Only offload copy if superblock is the same */
+       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
+               return -EXDEV;
        if (!nfs_server_capable(file_inode(file_out), NFS_CAP_COPY))
                return -EOPNOTSUPP;
        if (file_inode(file_in) == file_inode(file_out))

@@ -148,7 +151,7 @@ static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
 
        ret = __nfs4_copy_file_range(file_in, pos_in, file_out, pos_out, count,
                                     flags);
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(file_in, pos_in, file_out,
                                              pos_out, count, flags);
        return ret;

fs/read_write.c +12 -7

index cec7e7b1f693..bb594c8f4404 100644
@@ -1599,7 +1599,16 @@ static ssize_t do_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)
 {
-       if (file_out->f_op->copy_file_range)
+       /*
+        * Although we now allow filesystems to handle cross sb copy, 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.
+        * NFS has several different file_system_type's, but they all end up
+        * using the same ->copy_file_range() function pointer.
+        */
+       if (file_out->f_op->copy_file_range &&
+           file_out->f_op->copy_file_range == file_in->f_op->copy_file_range)
                return file_out->f_op->copy_file_range(file_in, pos_in,
                                                       file_out, pos_out,
                                                       len, flags);

@@ -1622,10 +1631,6 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
        if (flags != 0)
                return -EINVAL;
 
-       /* this could be relaxed once a method supports cross-fs copies */
-       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
-               return -EXDEV;
-
        ret = generic_copy_file_checks(file_in, pos_in, file_out, pos_out, &len,
                                       flags);
        if (unlikely(ret))

@@ -1648,7 +1653,8 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
         * Try cloning first, this is supported by more file systems, and
         * more efficient if both clone and copy are supported (e.g. NFS).
         */
-       if (file_in->f_op->remap_file_range) {
+       if (file_in->f_op->remap_file_range &&
+           file_inode(file_in)->i_sb == file_inode(file_out)->i_sb) {
                loff_t cloned;
 
                cloned = file_in->f_op->remap_file_range(file_in, pos_in,
--

2.17.1

Quoted 19 lines · Amir Goldstein

We want to enable cross-filesystem copy_file_range functionality where possible, so push the "same superblock only" checks down to the individual filesystem callouts so they can make their own decisions about cross-superblock copy offload and fallack to generic_copy_file_range() for cross-superblock copy.

[Amir] We do not call ->remap_file_range() in case the files are not on the same sb and do not call ->copy_file_range() in case the files do not belong to the same filesystem driver.

This changes behavior of the copy_file_range(2) syscall, which will now allow cross filesystem in-kernel copy. CIFS already supports cross-superblock copy, between two shares to the same server. This functionality will now be available via the copy_file_range(2) syscall.

Cc Steve French <[email protected]>

Signed-off-by Dave Chinner <[email protected]>

Signed-off-by Amir Goldstein <[email protected]>

Quoted 7 lines

---

Darrick,

Per feedback from Olga, I am sending a modified version of this patch to address cross file_system_type copy issue in nfs.

Thanks Amir, this works for NFS with referrals.

Quoted 144 lines

For the sake of global warming I am not re-posting the entire patch set. I removed your RVB because of the change.

Thanks,
Amir.

Changes since v4:
- Check "same filesystem driver" by comapring ->copy_file_range() function pointer

 fs/ceph/file.c    |  4 +++-
 fs/cifs/cifsfs.c  |  2 +-
 fs/fuse/file.c    |  5 ++++-
 fs/nfs/nfs4file.c |  5 ++++-
 fs/read_write.c   | 18 ++++++++++++------
 5 files changed, 24 insertions(+), 10 deletions(-)

fs/ceph/file.c +3 -1

index e87f7b2023af..4cd41ed5cc53 100644
@@ -1909,6 +1909,8 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
 
        if (src_inode == dst_inode)
                return -EINVAL;
+       if (src_inode->i_sb != dst_inode->i_sb)
+               return -EXDEV;
        if (ceph_snap(dst_inode) != CEPH_NOSNAP)
                return -EROFS;
 

@@ -2109,7 +2111,7 @@ static ssize_t ceph_copy_file_range(struct file *src_file, loff_t src_off,
        ret = __ceph_copy_file_range(src_file, src_off, dst_file, dst_off,
                                     len, flags);
 
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(src_file, src_off, dst_file,
                                              dst_off, len, flags);
        return ret;

fs/cifs/cifsfs.c +1 -1

index c65823270313..f11eea6125c1 100644
@@ -1149,7 +1149,7 @@ static ssize_t cifs_copy_file_range(struct file *src_file, loff_t off,
                                        len, flags);
        free_xid(xid);
 
-       if (rc == -EOPNOTSUPP)
+       if (rc == -EOPNOTSUPP || rc == -EXDEV)
                rc = generic_copy_file_range(src_file, off, dst_file,
                                             destoff, len, flags);
        return rc;

fs/fuse/file.c +4 -1

index e03901ae729b..569baf286835 100644
@@ -3126,6 +3126,9 @@ static ssize_t __fuse_copy_file_range(struct file *file_in, loff_t pos_in,
        if (fc->no_copy_file_range)
                return -EOPNOTSUPP;
 
+       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
+               return -EXDEV;
+
        inode_lock(inode_out);
 
        if (fc->writeback_cache) {

@@ -3182,7 +3185,7 @@ static ssize_t fuse_copy_file_range(struct file *src_file, loff_t src_off,
        ret = __fuse_copy_file_range(src_file, src_off, dst_file, dst_off,
                                     len, flags);
 
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(src_file, src_off, dst_file,
                                              dst_off, len, flags);
        return ret;

fs/nfs/nfs4file.c +4 -1

index 4842f3ab3161..f4157eb1f69d 100644
@@ -133,6 +133,9 @@ static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
                                      struct file *file_out, loff_t pos_out,
                                      size_t count, unsigned int flags)
 {
+       /* Only offload copy if superblock is the same */
+       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
+               return -EXDEV;
        if (!nfs_server_capable(file_inode(file_out), NFS_CAP_COPY))
                return -EOPNOTSUPP;
        if (file_inode(file_in) == file_inode(file_out))

@@ -148,7 +151,7 @@ static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
 
        ret = __nfs4_copy_file_range(file_in, pos_in, file_out, pos_out, count,
                                     flags);
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(file_in, pos_in, file_out,
                                              pos_out, count, flags);
        return ret;

fs/read_write.c +12 -7

index cec7e7b1f693..bb594c8f4404 100644
@@ -1599,7 +1599,16 @@ static ssize_t do_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)
 {
-       if (file_out->f_op->copy_file_range)
+       /*
+        * Although we now allow filesystems to handle cross sb copy, 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.
+        * NFS has several different file_system_type's, but they all end up
+        * using the same ->copy_file_range() function pointer.
+        */
+       if (file_out->f_op->copy_file_range &&
+           file_out->f_op->copy_file_range == file_in->f_op->copy_file_range)
                return file_out->f_op->copy_file_range(file_in, pos_in,
                                                       file_out, pos_out,
                                                       len, flags);

@@ -1622,10 +1631,6 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
        if (flags != 0)
                return -EINVAL;
 
-       /* this could be relaxed once a method supports cross-fs copies */
-       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
-               return -EXDEV;
-
        ret = generic_copy_file_checks(file_in, pos_in, file_out, pos_out, &len,
                                       flags);
        if (unlikely(ret))

@@ -1648,7 +1653,8 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
         * Try cloning first, this is supported by more file systems, and
         * more efficient if both clone and copy are supported (e.g. NFS).
         */
-       if (file_in->f_op->remap_file_range) {
+       if (file_in->f_op->remap_file_range &&
+           file_inode(file_in)->i_sb == file_inode(file_out)->i_sb) {
                loff_t cloned;
 
                cloned = file_in->f_op->remap_file_range(file_in, pos_in,
--

2.17.1

I tried the patchset and new xfstests with cifs.ko and looks like the copy_file_range patches are fine

1) xfs_io copy_range worked fine (once I fixed a bug I found in cifs.ko in the "match_server" function)
2) the new xfstest 989 requires other features (swapfiles) which I can add and 990 requires new xfsprogs presumably (xfs_io in current Ubuntu doesn't seem to include xfs_io chmod)

generic/989 [not run] swapfiles are not supported

generic/990 [not run] xfs_io chmod support is missing

Are there other xfstests that you would like me to run (other than the 3 in the 900 series)?

Quoted 171 lines · Olga Kornievskaia

Quoted 19 lines · Amir Goldstein

We want to enable cross-filesystem copy_file_range functionality where possible, so push the "same superblock only" checks down to the individual filesystem callouts so they can make their own decisions about cross-superblock copy offload and fallack to generic_copy_file_range() for cross-superblock copy.

[Amir] We do not call ->remap_file_range() in case the files are not on the same sb and do not call ->copy_file_range() in case the files do not belong to the same filesystem driver.

This changes behavior of the copy_file_range(2) syscall, which will now allow cross filesystem in-kernel copy. CIFS already supports cross-superblock copy, between two shares to the same server. This functionality will now be available via the copy_file_range(2) syscall.

Cc Steve French <[email protected]>

Signed-off-by Dave Chinner <[email protected]>

Signed-off-by Amir Goldstein <[email protected]>

Quoted 7 lines

---

Darrick,

Per feedback from Olga, I am sending a modified version of this patch to address cross file_system_type copy issue in nfs.

Thanks Amir, this works for NFS with referrals.

Quoted 140 lines

For the sake of global warming I am not re-posting the entire patch set. I removed your RVB because of the change.

Thanks,
Amir.

Changes since v4:
- Check "same filesystem driver" by comapring ->copy_file_range() function pointer

 fs/ceph/file.c    |  4 +++-
 fs/cifs/cifsfs.c  |  2 +-
 fs/fuse/file.c    |  5 ++++-
 fs/nfs/nfs4file.c |  5 ++++-
 fs/read_write.c   | 18 ++++++++++++------
 5 files changed, 24 insertions(+), 10 deletions(-)

fs/ceph/file.c +3 -1

index e87f7b2023af..4cd41ed5cc53 100644
@@ -1909,6 +1909,8 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
 
        if (src_inode == dst_inode)
                return -EINVAL;
+       if (src_inode->i_sb != dst_inode->i_sb)
+               return -EXDEV;
        if (ceph_snap(dst_inode) != CEPH_NOSNAP)
                return -EROFS;
 

@@ -2109,7 +2111,7 @@ static ssize_t ceph_copy_file_range(struct file *src_file, loff_t src_off,
        ret = __ceph_copy_file_range(src_file, src_off, dst_file, dst_off,
                                     len, flags);
 
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(src_file, src_off, dst_file,
                                              dst_off, len, flags);
        return ret;

fs/cifs/cifsfs.c +1 -1

index c65823270313..f11eea6125c1 100644
@@ -1149,7 +1149,7 @@ static ssize_t cifs_copy_file_range(struct file *src_file, loff_t off,
                                        len, flags);
        free_xid(xid);
 
-       if (rc == -EOPNOTSUPP)
+       if (rc == -EOPNOTSUPP || rc == -EXDEV)
                rc = generic_copy_file_range(src_file, off, dst_file,
                                             destoff, len, flags);
        return rc;

fs/fuse/file.c +4 -1

index e03901ae729b..569baf286835 100644
@@ -3126,6 +3126,9 @@ static ssize_t __fuse_copy_file_range(struct file *file_in, loff_t pos_in,
        if (fc->no_copy_file_range)
                return -EOPNOTSUPP;
 
+       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
+               return -EXDEV;
+
        inode_lock(inode_out);
 
        if (fc->writeback_cache) {

@@ -3182,7 +3185,7 @@ static ssize_t fuse_copy_file_range(struct file *src_file, loff_t src_off,
        ret = __fuse_copy_file_range(src_file, src_off, dst_file, dst_off,
                                     len, flags);
 
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(src_file, src_off, dst_file,
                                              dst_off, len, flags);
        return ret;

fs/nfs/nfs4file.c +4 -1

index 4842f3ab3161..f4157eb1f69d 100644
@@ -133,6 +133,9 @@ static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
                                      struct file *file_out, loff_t pos_out,
                                      size_t count, unsigned int flags)
 {
+       /* Only offload copy if superblock is the same */
+       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
+               return -EXDEV;
        if (!nfs_server_capable(file_inode(file_out), NFS_CAP_COPY))
                return -EOPNOTSUPP;
        if (file_inode(file_in) == file_inode(file_out))

@@ -148,7 +151,7 @@ static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
 
        ret = __nfs4_copy_file_range(file_in, pos_in, file_out, pos_out, count,
                                     flags);
-       if (ret == -EOPNOTSUPP)
+       if (ret == -EOPNOTSUPP || ret == -EXDEV)
                ret = generic_copy_file_range(file_in, pos_in, file_out,
                                              pos_out, count, flags);
        return ret;

fs/read_write.c +12 -6

index cec7e7b1f693..bb594c8f4404 100644
@@ -1599,7 +1599,16 @@ static ssize_t do_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)
 {
-       if (file_out->f_op->copy_file_range)
+       /*
+        * Although we now allow filesystems to handle cross sb copy, 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.
+        * NFS has several different file_system_type's, but they all end up
+        * using the same ->copy_file_range() function pointer.
+        */
+       if (file_out->f_op->copy_file_range &&
+           file_out->f_op->copy_file_range == file_in->f_op->copy_file_range)
                return file_out->f_op->copy_file_range(file_in, pos_in,
                                                       file_out, pos_out,
                                                       len, flags);

@@ -1622,10 +1631,6 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
        if (flags != 0)
                return -EINVAL;
 
-       /* this could be relaxed once a method supports cross-fs copies */
-       if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
-               return -EXDEV;
-
        ret = generic_copy_file_checks(file_in, pos_in, file_out, pos_out, &len,
                                       flags);
        if (unlikely(ret))

@@ -1648,7 +1653,8 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
         * Try cloning first, this is supported by more file systems, and
         * more efficient if both clone and copy are supported (e.g. NFS).
         */
-       if (file_in->f_op->remap_file_range) {
+       if (file_in->f_op->remap_file_range &&
+           file_inode(file_in)->i_sb == file_inode(file_out)->i_sb) {
                loff_t cloned;
 

Quoted 4 lines

Quoted 4 lines

fs/read_write.c +0 -1

                cloned = file_in->f_op->remap_file_range(file_in, pos_in,
--

2.17.1

--
Thanks,

Steve