| Message ID | 1440577010-122867-2-git-send-email-tao.peng@primarydata.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Aug 26, 2015 at 04:16:42PM +0800, Peng Tao wrote: > +struct file_clone_range { > + __s64 src_fd; > + __u64 src_offset; > + __u64 src_length; > + __u64 dest_offset; > +}; Might be a good idea to add some spare bytes to the structure. > struct fstrim_range { > __u64 start; > __u64 len; > @@ -159,6 +166,8 @@ struct inodes_stat_t { > #define FIFREEZE _IOWR('X', 119, int) /* Freeze */ > #define FITHAW _IOWR('X', 120, int) /* Thaw */ > #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */ > +#define FICLONE _IOW(0x94, 9, int) /* Clone */ > +#define FICLONERANGE _IOW(0x94, 13, struct file_clone_range) /* Clone range */ FICLONE is a special case of FICLONERANGE. The whole file clone had come historically first and then was refined, I don't think this needs to be copied to the generic API. A zeroed file_clone_range is simple to use for that purpose. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Aug 26, 2015 at 03:37:23PM +0200, David Sterba wrote: > On Wed, Aug 26, 2015 at 04:16:42PM +0800, Peng Tao wrote: > > +struct file_clone_range { > > + __s64 src_fd; > > + __u64 src_offset; > > + __u64 src_length; > > + __u64 dest_offset; > > +}; > > Might be a good idea to add some spare bytes to the structure. But... structure size is encoded in the ioctl definition, so adding bytes to struct file_clone_range now will change the ioctl number and break userland. --D > > > struct fstrim_range { > > __u64 start; > > __u64 len; > > @@ -159,6 +166,8 @@ struct inodes_stat_t { > > #define FIFREEZE _IOWR('X', 119, int) /* Freeze */ > > #define FITHAW _IOWR('X', 120, int) /* Thaw */ > > #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */ > > +#define FICLONE _IOW(0x94, 9, int) /* Clone */ > > +#define FICLONERANGE _IOW(0x94, 13, struct file_clone_range) /* Clone range */ > > FICLONE is a special case of FICLONERANGE. The whole file clone had come > historically first and then was refined, I don't think this needs to be > copied to the generic API. A zeroed file_clone_range is simple to use > for that purpose. > -- > To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Aug 26, 2015 at 09:21:54AM -0700, Darrick J. Wong wrote: > On Wed, Aug 26, 2015 at 03:37:23PM +0200, David Sterba wrote: > > On Wed, Aug 26, 2015 at 04:16:42PM +0800, Peng Tao wrote: > > > +struct file_clone_range { > > > + __s64 src_fd; > > > + __u64 src_offset; > > > + __u64 src_length; > > > + __u64 dest_offset; > > > +}; > > > > Might be a good idea to add some spare bytes to the structure. > > But... structure size is encoded in the ioctl definition, so adding bytes > to struct file_clone_range now will change the ioctl number and break > userland. Oh right, I somehow did not left idea of a new ioctl definition while writing it. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Aug 26, 2015 at 04:16:42PM +0800, Peng Tao wrote: > Now that a few file systems are adding clone functionality, namingly > btrfs, NFS (later in the series) and XFS > (ttp://oss.sgi.com/archives/xfs/2015-06/msg00407.html), it makes sense > to pull the ioctl to common code. > > Add vfs_file_clone_range() helper and .clone_range file operation interface > to allow underlying filesystems to clone between regular files. > > The change in do_vfs_ioctl() is defered to next patch where btrfs > .clone_range is added, just so that we don't break btrfs CLONE ioctl > with this patch. > > Signed-off-by: Peng Tao <tao.peng@primarydata.com> > --- > fs/ioctl.c | 24 ++++++++++++++++++++++++ > fs/read_write.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ > include/linux/fs.h | 4 ++++ > include/uapi/linux/fs.h | 9 +++++++++ > 4 files changed, 82 insertions(+) ..... > +int vfs_file_clone_range(struct file *src_file, struct file *dst_file, > + loff_t off, size_t len, loff_t dstoff) > +{ > + struct inode *src_ino; > + struct inode *dst_ino; > + ssize_t ret; > + > + if (!(src_file->f_mode & FMODE_READ) || > + !(dst_file->f_mode & FMODE_WRITE) || > + (dst_file->f_flags & O_APPEND) || > + !src_file->f_op || !src_file->f_op->clone_range) > + return -EINVAL; > + > + src_ino = file_inode(src_file); > + dst_ino = file_inode(dst_file); > + > + if (S_ISDIR(src_ino->i_mode) || S_ISDIR(dst_ino->i_mode)) > + return -EISDIR; Whacky whitespace. Also, shouldn't this call be restricted to S_ISREG() inodes? This only checks for directories... Cheers, Dave.
On Thu, Aug 27, 2015 at 6:52 AM, Dave Chinner <david@fromorbit.com> wrote: > On Wed, Aug 26, 2015 at 04:16:42PM +0800, Peng Tao wrote: >> Now that a few file systems are adding clone functionality, namingly >> btrfs, NFS (later in the series) and XFS >> (ttp://oss.sgi.com/archives/xfs/2015-06/msg00407.html), it makes sense >> to pull the ioctl to common code. >> >> Add vfs_file_clone_range() helper and .clone_range file operation interface >> to allow underlying filesystems to clone between regular files. >> >> The change in do_vfs_ioctl() is defered to next patch where btrfs >> .clone_range is added, just so that we don't break btrfs CLONE ioctl >> with this patch. >> >> Signed-off-by: Peng Tao <tao.peng@primarydata.com> >> --- >> fs/ioctl.c | 24 ++++++++++++++++++++++++ >> fs/read_write.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ >> include/linux/fs.h | 4 ++++ >> include/uapi/linux/fs.h | 9 +++++++++ >> 4 files changed, 82 insertions(+) > ..... >> +int vfs_file_clone_range(struct file *src_file, struct file *dst_file, >> + loff_t off, size_t len, loff_t dstoff) >> +{ >> + struct inode *src_ino; >> + struct inode *dst_ino; >> + ssize_t ret; >> + >> + if (!(src_file->f_mode & FMODE_READ) || >> + !(dst_file->f_mode & FMODE_WRITE) || >> + (dst_file->f_flags & O_APPEND) || >> + !src_file->f_op || !src_file->f_op->clone_range) >> + return -EINVAL; >> + >> + src_ino = file_inode(src_file); >> + dst_ino = file_inode(dst_file); >> + >> + if (S_ISDIR(src_ino->i_mode) || S_ISDIR(dst_ino->i_mode)) >> + return -EISDIR; > > Whacky whitespace. > > Also, shouldn't this call be restricted to S_ISREG() inodes? This > only checks for directories... Good point. I'll change it. Thanks! Cheers, Tao -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Aug 26, 2015 at 9:37 PM, David Sterba <dsterba@suse.cz> wrote: > On Wed, Aug 26, 2015 at 04:16:42PM +0800, Peng Tao wrote: >> +struct file_clone_range { >> + __s64 src_fd; >> + __u64 src_offset; >> + __u64 src_length; >> + __u64 dest_offset; >> +}; > > Might be a good idea to add some spare bytes to the structure. > >> struct fstrim_range { >> __u64 start; >> __u64 len; >> @@ -159,6 +166,8 @@ struct inodes_stat_t { >> #define FIFREEZE _IOWR('X', 119, int) /* Freeze */ >> #define FITHAW _IOWR('X', 120, int) /* Thaw */ >> #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */ >> +#define FICLONE _IOW(0x94, 9, int) /* Clone */ >> +#define FICLONERANGE _IOW(0x94, 13, struct file_clone_range) /* Clone range */ > > FICLONE is a special case of FICLONERANGE. The whole file clone had come > historically first and then was refined, I don't think this needs to be > copied to the generic API. A zeroed file_clone_range is simple to use > for that purpose. oh, sorry I missed this one... BTRFS_IOC_CLONE is being used by cp(1) and widely shipped with distros. So we cannot just abandon the API. And my intention is to keep the clone ioctl "JUST WORK" (TM) with cp(1). Cheers, Tao -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/ioctl.c b/fs/ioctl.c index 5d01d26..726c5d7 100644 --- a/fs/ioctl.c +++ b/fs/ioctl.c @@ -215,6 +215,30 @@ static int ioctl_fiemap(struct file *filp, unsigned long arg) return error; } +static long ioctl_file_clone(struct file *dst_file, unsigned long srcfd, + u64 off, u64 olen, u64 destoff) +{ + struct fd src_file = fdget(srcfd); + int ret; + + if (!src_file.file) + return -EBADF; + ret = vfs_file_clone_range(src_file.file, dst_file, off, olen, destoff); + + fdput(src_file); + return ret; +} + +static long ioctl_file_clone_range(struct file *file, void __user *argp) +{ + struct file_clone_range args; + + if (copy_from_user(&args, argp, sizeof(args))) + return -EFAULT; + return ioctl_file_clone(file, args.src_fd, args.src_offset, + args.src_length, args.dest_offset); +} + #ifdef CONFIG_BLOCK static inline sector_t logical_to_blk(struct inode *inode, loff_t offset) diff --git a/fs/read_write.c b/fs/read_write.c index 819ef3f..beaad2c 100644 --- a/fs/read_write.c +++ b/fs/read_write.c @@ -16,6 +16,7 @@ #include <linux/pagemap.h> #include <linux/splice.h> #include <linux/compat.h> +#include <linux/mount.h> #include "internal.h" #include <asm/uaccess.h> @@ -1327,3 +1328,47 @@ COMPAT_SYSCALL_DEFINE4(sendfile64, int, out_fd, int, in_fd, return do_sendfile(out_fd, in_fd, NULL, count, 0); } #endif + +int vfs_file_clone_range(struct file *src_file, struct file *dst_file, + loff_t off, size_t len, loff_t dstoff) +{ + struct inode *src_ino; + struct inode *dst_ino; + ssize_t ret; + + if (!(src_file->f_mode & FMODE_READ) || + !(dst_file->f_mode & FMODE_WRITE) || + (dst_file->f_flags & O_APPEND) || + !src_file->f_op || !src_file->f_op->clone_range) + return -EINVAL; + + src_ino = file_inode(src_file); + dst_ino = file_inode(dst_file); + + if (S_ISDIR(src_ino->i_mode) || S_ISDIR(dst_ino->i_mode)) + return -EISDIR; + + /* sanity check on offsets and length */ + if (off + len < off || dstoff + len < dstoff || + off + len > i_size_read(src_ino)) + return -EINVAL; + + if (src_ino->i_sb != dst_ino->i_sb || + src_file->f_path.mnt != dst_file->f_path.mnt) + return -EXDEV; + + ret = mnt_want_write_file(dst_file); + if (ret) + return ret; + + ret = src_file->f_op->clone_range(src_file, dst_file, off, len, dstoff); + if (!ret) { + fsnotify_access(src_file); + fsnotify_modify(dst_file); + } + + mnt_drop_write_file(dst_file); + + return ret; +} +EXPORT_SYMBOL(vfs_file_clone_range); diff --git a/include/linux/fs.h b/include/linux/fs.h index cc008c3..612d7f4 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1628,6 +1628,8 @@ struct file_operations { long (*fallocate)(struct file *file, int mode, loff_t offset, loff_t len); void (*show_fdinfo)(struct seq_file *m, struct file *f); + int (*clone_range)(struct file *src_file, struct file *dst_file, + loff_t off, size_t len, loff_t dstoff); #ifndef CONFIG_MMU unsigned (*mmap_capabilities)(struct file *); #endif @@ -2678,6 +2680,8 @@ int __dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t, int dax_pfn_mkwrite(struct vm_area_struct *, struct vm_fault *); #define dax_mkwrite(vma, vmf, gb, iod) dax_fault(vma, vmf, gb, iod) #define __dax_mkwrite(vma, vmf, gb, iod) __dax_fault(vma, vmf, gb, iod) +int vfs_file_clone_range(struct file *src_file, struct file *dst_file, + loff_t off, size_t len, loff_t dstoff); #ifdef CONFIG_BLOCK typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode, diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h index 9b964a5..ac7f1c5 100644 --- a/include/uapi/linux/fs.h +++ b/include/uapi/linux/fs.h @@ -39,6 +39,13 @@ #define RENAME_EXCHANGE (1 << 1) /* Exchange source and dest */ #define RENAME_WHITEOUT (1 << 2) /* Whiteout source */ +struct file_clone_range { + __s64 src_fd; + __u64 src_offset; + __u64 src_length; + __u64 dest_offset; +}; + struct fstrim_range { __u64 start; __u64 len; @@ -159,6 +166,8 @@ struct inodes_stat_t { #define FIFREEZE _IOWR('X', 119, int) /* Freeze */ #define FITHAW _IOWR('X', 120, int) /* Thaw */ #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */ +#define FICLONE _IOW(0x94, 9, int) /* Clone */ +#define FICLONERANGE _IOW(0x94, 13, struct file_clone_range) /* Clone range */ #define FS_IOC_GETFLAGS _IOR('f', 1, long) #define FS_IOC_SETFLAGS _IOW('f', 2, long)
Now that a few file systems are adding clone functionality, namingly btrfs, NFS (later in the series) and XFS (ttp://oss.sgi.com/archives/xfs/2015-06/msg00407.html), it makes sense to pull the ioctl to common code. Add vfs_file_clone_range() helper and .clone_range file operation interface to allow underlying filesystems to clone between regular files. The change in do_vfs_ioctl() is defered to next patch where btrfs .clone_range is added, just so that we don't break btrfs CLONE ioctl with this patch. Signed-off-by: Peng Tao <tao.peng@primarydata.com> --- fs/ioctl.c | 24 ++++++++++++++++++++++++ fs/read_write.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ include/linux/fs.h | 4 ++++ include/uapi/linux/fs.h | 9 +++++++++ 4 files changed, 82 insertions(+)