[1/4] libext2fs: Support for orphan file feature

Add support for creating and deleting orphan file and a couple of
utility functions that will be used in other tools.

Signed-off-by: Jan Kara <jack@suse.cz>
---
 lib/e2p/feature.c      |   4 +
 lib/ext2fs/Makefile.in |   2 +
 lib/ext2fs/ext2_fs.h   |  11 +++
 lib/ext2fs/ext2fs.h    |  35 +++++++-
 lib/ext2fs/orphan.c    | 217 +++++++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 267 insertions(+), 2 deletions(-)
 create mode 100644 lib/ext2fs/orphan.c

On Fri, May 22, 2015 at 01:28:54PM +0200, Jan Kara wrote:
> Add support for creating and deleting orphan file and a couple of
> utility functions that will be used in other tools.
> 
> Signed-off-by: Jan Kara <jack@suse.cz>
> ---
>  lib/e2p/feature.c      |   4 +
>  lib/ext2fs/Makefile.in |   2 +
>  lib/ext2fs/ext2_fs.h   |  11 +++
>  lib/ext2fs/ext2fs.h    |  35 +++++++-
>  lib/ext2fs/orphan.c    | 217 +++++++++++++++++++++++++++++++++++++++++++++++++
>  5 files changed, 267 insertions(+), 2 deletions(-)
>  create mode 100644 lib/ext2fs/orphan.c
> 
> diff --git a/lib/e2p/feature.c b/lib/e2p/feature.c
> index 73884f2cf5bf..a8e0d4a4644a 100644
> --- a/lib/e2p/feature.c
> +++ b/lib/e2p/feature.c
> @@ -45,6 +45,8 @@ static struct feature feature_list[] = {
>  			"snapshot_bitmap" },
>  	{	E2P_FEATURE_COMPAT, EXT4_FEATURE_COMPAT_SPARSE_SUPER2,
>  			"sparse_super2" },
> +	{	E2P_FEATURE_COMPAT, EXT4_FEATURE_COMPAT_ORPHAN_FILE,
> +			"orphan_file" },
>  
>  	{	E2P_FEATURE_RO_INCOMPAT, EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER,
>  			"sparse_super" },
> @@ -70,6 +72,8 @@ static struct feature feature_list[] = {
>  			"replica" },
>  	{	E2P_FEATURE_RO_INCOMPAT, EXT4_FEATURE_RO_COMPAT_READONLY,
>  			"read-only" },
> +	{	E2P_FEATURE_RO_INCOMPAT, EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT,
> +			"orphan_file_used" },
>  
>  	{	E2P_FEATURE_INCOMPAT, EXT2_FEATURE_INCOMPAT_COMPRESSION,
>  			"compression" },
> diff --git a/lib/ext2fs/Makefile.in b/lib/ext2fs/Makefile.in
> index 8a7f8ca52902..67120b10438c 100644
> --- a/lib/ext2fs/Makefile.in
> +++ b/lib/ext2fs/Makefile.in
> @@ -109,6 +109,7 @@ OBJS= $(DEBUGFS_LIB_OBJS) $(RESIZE_LIB_OBJS) $(E2IMAGE_LIB_OBJS) \
>  	native.o \
>  	newdir.o \
>  	openfs.o \
> +	orphan.o \
>  	progress.o \
>  	punch.o \
>  	qcow2.o \
> @@ -189,6 +190,7 @@ SRCS= ext2_err.c \
>  	$(srcdir)/native.c \
>  	$(srcdir)/newdir.c \
>  	$(srcdir)/openfs.c \
> +	$(srcdir)/orphan.c \
>  	$(srcdir)/progress.c \
>  	$(srcdir)/punch.c \
>  	$(srcdir)/qcow2.c \
> diff --git a/lib/ext2fs/ext2_fs.h b/lib/ext2fs/ext2_fs.h
> index a755cfac8eae..a77c8fa09938 100644
> --- a/lib/ext2fs/ext2_fs.h
> +++ b/lib/ext2fs/ext2_fs.h
> @@ -52,6 +52,7 @@
>  #define EXT2_JOURNAL_INO	 8	/* Journal inode */
>  #define EXT2_EXCLUDE_INO	 9	/* The "exclude" inode, for snapshots */
>  #define EXT4_REPLICA_INO	10	/* Used by non-upstream feature */
> +#define EXT4_ORPHAN_INO		 9	/* Inode with orphan entries */
>  
>  /* First non-reserved inode for old ext2 filesystems */
>  #define EXT2_GOOD_OLD_FIRST_INO	11
> @@ -769,6 +770,7 @@ struct ext2_super_block {
>  /* #define EXT2_FEATURE_COMPAT_EXCLUDE_INODE	0x0080 not used, legacy */
>  #define EXT2_FEATURE_COMPAT_EXCLUDE_BITMAP	0x0100
>  #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
> +#define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x0400
>  
>  
>  #define EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
> @@ -789,6 +791,7 @@ struct ext2_super_block {
>  #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
>  #define EXT4_FEATURE_RO_COMPAT_REPLICA		0x0800
>  #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
> +#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x2000
>  
>  
>  #define EXT2_FEATURE_INCOMPAT_COMPRESSION	0x0001
> @@ -838,6 +841,14 @@ struct ext2_super_block {
>  #define EXT4_DEFM_DISCARD	0x0400
>  #define EXT4_DEFM_NODELALLOC	0x0800
>  
> +#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
> +
> +/* Structure at the tail of orphan block */
> +struct ext4_orphan_block_tail {
> +	__u32 ob_magic;
> +	__u32 ob_checksum;
> +};
> +
>  /*
>   * Structure of a directory entry
>   */
> diff --git a/lib/ext2fs/ext2fs.h b/lib/ext2fs/ext2fs.h
> index 28c46701da29..1e303d5d59ca 100644
> --- a/lib/ext2fs/ext2fs.h
> +++ b/lib/ext2fs/ext2fs.h
> @@ -555,7 +555,8 @@ typedef struct ext2_icount *ext2_icount_t;
>  					 EXT2_FEATURE_COMPAT_RESIZE_INODE|\
>  					 EXT2_FEATURE_COMPAT_DIR_INDEX|\
>  					 EXT2_FEATURE_COMPAT_EXT_ATTR|\
> -					 EXT4_FEATURE_COMPAT_SPARSE_SUPER2)
> +					 EXT4_FEATURE_COMPAT_SPARSE_SUPER2|\
> +					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
>  
>  #ifdef CONFIG_MMP
>  #define EXT4_LIB_INCOMPAT_MMP		EXT4_FEATURE_INCOMPAT_MMP
> @@ -589,7 +590,8 @@ typedef struct ext2_icount *ext2_icount_t;
>  					 EXT4_FEATURE_RO_COMPAT_BIGALLOC|\
>  					 EXT4_LIB_RO_COMPAT_QUOTA|\
>  					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
> -					 EXT4_FEATURE_RO_COMPAT_READONLY)
> +					 EXT4_FEATURE_RO_COMPAT_READONLY|\
> +					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
>  
>  /*
>   * These features are only allowed if EXT2_FLAG_SOFTSUPP_FEATURES is passed
> @@ -1512,6 +1514,19 @@ errcode_t ext2fs_get_data_io(ext2_filsys fs, io_channel *old_io);
>  errcode_t ext2fs_set_data_io(ext2_filsys fs, io_channel new_io);
>  errcode_t ext2fs_rewrite_to_io(ext2_filsys fs, io_channel new_io);
>  
> +/* orphan.c */
> +/*
> + * Minimum orphan file size (it must be at least 1 block and smaller one isn't
> + * very useful).
> + */
> +#define EXT4_MIN_ORPHAN_FILE_SIZE 16384

What about 64k block size?  I guess it's fine not to use the whole block in
this (non-default) configuration.

> +
> +extern errcode_t ext2fs_create_orphan_file(ext2_filsys fs, blk_t num_blocks);
> +extern errcode_t ext2fs_truncate_orphan_file(ext2_filsys fs);
> +extern e2_blkcnt_t ext2fs_default_orphan_file_blocks(__u64 num_blocks);
> +extern errcode_t ext2fs_orphan_file_block_csum_set(ext2_filsys fs, char *buf);
> +extern int ext2fs_orphan_file_block_csum_verify(ext2_filsys fs, char *buf);
> +
>  /* get_pathname.c */
>  extern errcode_t ext2fs_get_pathname(ext2_filsys fs, ext2_ino_t dir, ext2_ino_t ino,
>  			       char **name);
> @@ -1645,6 +1660,9 @@ extern int ext2fs_dirent_name_len(const struct ext2_dir_entry *entry);
>  extern void ext2fs_dirent_set_name_len(struct ext2_dir_entry *entry, int len);
>  extern int ext2fs_dirent_file_type(const struct ext2_dir_entry *entry);
>  extern void ext2fs_dirent_set_file_type(struct ext2_dir_entry *entry, int type);
> +extern int ext2fs_inodes_per_orphan_block(ext2_filsys fs);
> +extern struct ext4_orphan_block_tail *ext2fs_orphan_block_tail(ext2_filsys fs,
> +							       char *buf)
>  
>  #endif
>  
> @@ -1915,6 +1933,19 @@ _INLINE_ void ext2fs_dirent_set_file_type(struct ext2_dir_entry *entry, int type
>  	entry->name_len = (entry->name_len & 0xff) | (type << 8);
>  }
>  
> +_INLINE_ int ext2fs_inodes_per_orphan_block(ext2_filsys fs)
> +{
> +	return (fs->blocksize - sizeof(struct ext4_orphan_block_tail)) /
> +		sizeof(__u32);
> +}
> +
> +_INLINE_ struct ext4_orphan_block_tail *
> +ext2fs_orphan_block_tail(ext2_filsys fs, char *buf)
> +{
> +	return (struct ext4_orphan_block_tail *)(buf + fs->blocksize -
> +		sizeof(struct ext4_orphan_block_tail));
> +}
> +
>  #undef _INLINE_
>  #endif
>  
> diff --git a/lib/ext2fs/orphan.c b/lib/ext2fs/orphan.c
> new file mode 100644
> index 000000000000..1fd5c0688218
> --- /dev/null
> +++ b/lib/ext2fs/orphan.c
> @@ -0,0 +1,217 @@
> +/*
> + * orphan.c --- utility function to handle orphan file
> + *
> + * Copyright (C) 2015 Jan Kara.
> + *
> + * %Begin-Header%
> + * This file may be redistributed under the terms of the GNU Library
> + * General Public License, version 2.
> + * %End-Header%
> + */
> +
> +#include "config.h"
> +#include <string.h>
> +
> +#include "ext2_fs.h"
> +#include "ext2fsP.h"
> +
> +errcode_t ext2fs_truncate_orphan_file(ext2_filsys fs)
> +{
> +	struct ext2_inode inode;
> +	errcode_t err;
> +
> +	err = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		return err;
> +
> +	err = ext2fs_punch(fs, EXT4_ORPHAN_INO, &inode, NULL, 0, ~0ULL);
> +	if (err)
> +		return err;
> +
> +	fs->flags &= ~EXT2_FLAG_SUPER_ONLY;
> +	memset(&inode, 0, sizeof(struct ext2_inode));
> +	err = ext2fs_write_inode(fs, EXT4_ORPHAN_INO, &inode);
> +
> +	fs->super->s_feature_compat &= ~EXT4_FEATURE_COMPAT_ORPHAN_FILE;
> +	ext2fs_mark_super_dirty(fs);
> +
> +	return err;
> +}
> +
> +struct mkorphan_info {
> +	char *buf;
> +	char *zerobuf;
> +	blk_t num_blocks;
> +	blk_t alloc_blocks;
> +	errcode_t err;
> +};
> +
> +static int mkorphan_proc(ext2_filsys	fs,
> +			 blk64_t	*blocknr,
> +			 e2_blkcnt_t	blockcnt,
> +			 blk64_t	ref_block EXT2FS_ATTR((unused)),
> +			 int		ref_offset EXT2FS_ATTR((unused)),
> +			 void		*priv_data)
> +{
> +	struct mkorphan_info *oi = (struct mkorphan_info *)priv_data;
> +	blk64_t new_blk;
> +	errcode_t err;
> +
> +	err = ext2fs_new_block2(fs, 0, 0, &new_blk);

Hm.  I think this breaks the cluster allocation rules, since this allocates
a new block for every logical block inside a cluster.

Hopefully ext2fs_fallocate will land soon, then we can get rid of open-coding
file block allocation like this.  You'll still have to have the iterate3 loop
to write out the appropriate block footer, but iirc the library call can be
told to allocate written extents without zeroing the blocks, precisely for
cases like these.

> +	if (err) {
> +		oi->err = err;
> +		return BLOCK_ABORT;
> +	}
> +	ext2fs_block_alloc_stats2(fs, new_blk, +1);
> +	if (blockcnt >= 0)
> +		err = io_channel_write_blk64(fs->io, new_blk, 1, oi->buf);
> +	else
> +		err = io_channel_write_blk64(fs->io, new_blk, 1, oi->zerobuf);
> +	if (err) {
> +		oi->err = err;
> +		return BLOCK_ABORT;
> +	}
> +	oi->alloc_blocks++;
> +	*blocknr = new_blk;
> +	if (blockcnt >= 0 && --oi->num_blocks == 0)
> +		return BLOCK_CHANGED | BLOCK_ABORT;
> +	return BLOCK_CHANGED;
> +}
> +
> +errcode_t ext2fs_create_orphan_file(ext2_filsys fs, blk_t num_blocks)
> +{
> +	struct ext2_inode inode;
> +	errcode_t err;
> +	char *buf = NULL, *zerobuf = NULL;
> +	struct mkorphan_info oi;
> +	struct ext4_orphan_block_tail *ob_tail;
> +
> +	err = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		return err;
> +	if (EXT2_I_SIZE(&inode)) {
> +		err = ext2fs_truncate_orphan_file(fs);
> +		if (err)
> +			return err;
> +	}
> +
> +	memset(&inode, 0, sizeof(struct ext2_inode));
> +	if (fs->super->s_feature_incompat & EXT3_FEATURE_INCOMPAT_EXTENTS) {
> +		inode.i_flags |= EXT4_EXTENTS_FL;
> +		err = ext2fs_write_inode(fs, EXT4_ORPHAN_INO, &inode);
> +		if (err)
> +			return err;
> +	}
> +
> +	err = ext2fs_get_mem(fs->blocksize, &buf);
> +	if (err)
> +		return err;
> +	err = ext2fs_get_mem(fs->blocksize, &zerobuf);
> +	if (err)
> +		goto out;
> +	memset(buf, 0, fs->blocksize);
> +	memset(zerobuf, 0, fs->blocksize);
> +	ob_tail = ext2fs_orphan_block_tail(fs, buf);
> +	ob_tail->ob_magic = ext2fs_cpu_to_le32(EXT4_ORPHAN_BLOCK_MAGIC);
> +	ext2fs_orphan_file_block_csum_set(fs, buf);
> +	oi.num_blocks = num_blocks;
> +	oi.alloc_blocks = 0;
> +	oi.buf = buf;
> +	oi.zerobuf = zerobuf;
> +	oi.err = 0;
> +	err = ext2fs_block_iterate3(fs, EXT4_ORPHAN_INO, BLOCK_FLAG_APPEND,
> +				    0, mkorphan_proc, &oi);
> +	if (err)
> +		goto out;
> +
> +	/* Reread inode after blocks were allocated */
> +	err = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		goto out;
> +	ext2fs_iblk_set(fs, &inode, 0);
> +	inode.i_atime = inode.i_mtime =
> +		inode.i_ctime = fs->now ? fs->now : time(0);
> +	inode.i_links_count = 1;
> +	inode.i_mode = LINUX_S_IFREG | 0600;
> +	ext2fs_iblk_add_blocks(fs, &inode, oi.alloc_blocks);
> +	err = ext2fs_inode_size_set(fs, &inode,
> +			(unsigned long long)fs->blocksize * num_blocks);
> +	if (err)
> +		goto out;
> +	err = ext2fs_write_new_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		goto out;
> +
> +	fs->super->s_feature_compat |= EXT4_FEATURE_COMPAT_ORPHAN_FILE;
> +	ext2fs_mark_super_dirty(fs);
> +out:
> +	if (buf)
> +		ext2fs_free_mem(&buf);
> +	if (zerobuf)
> +		ext2fs_free_mem(&zerobuf);
> +	return err;
> +}
> +
> +/*
> + * Find reasonable size for orphan file. We choose orphan file size to be
> + * between 32 and 512 filesystem blocks and not more than 1/4096 of the
> + * filesystem unless it is really small.
> + */
> +e2_blkcnt_t ext2fs_default_orphan_file_blocks(__u64 num_blocks)
> +{
> +	if (num_blocks < 128 * 1024)
> +		return 32;
> +	if (num_blocks < 2 * 1024 * 1024)
> +		return num_blocks / 4096;
> +	return 512;
> +}
> +
> +static errcode_t ext2fs_orphan_file_block_csum(ext2_filsys fs, char *buf,
> +					       __u32 *crc)
> +{
> +	int inodes_per_ob = ext2fs_inodes_per_orphan_block(fs);
> +	__u32 gen;
> +	ext2_ino_t inum;
> +	struct ext2_inode inode;
> +	errcode_t retval;
> +
> +	retval = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (retval)
> +		return retval;
> +	inum = ext2fs_cpu_to_le32(EXT4_ORPHAN_INO);
> +	gen = ext2fs_cpu_to_le32(inode.i_generation);
> +	*crc = ext2fs_crc32c_le(fs->csum_seed, (unsigned char *)&inum,
> +			       sizeof(inum));
> +	*crc = ext2fs_crc32c_le(*crc, (unsigned char *)&gen, sizeof(gen));
> +	*crc = ext2fs_crc32c_le(*crc, buf, inodes_per_ob * sizeof(__u32));
> +
> +	return 0;
> +}
> +
> +errcode_t ext2fs_orphan_file_block_csum_set(ext2_filsys fs, char *buf)
> +{
> +	struct ext4_orphan_block_tail *tail;
> +
> +	if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
> +					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
> +		return 0;
> +
> +	tail = ext2fs_orphan_block_tail(fs, buf);
> +	return ext2fs_orphan_file_block_csum(fs, buf, &tail->ob_checksum);
> +}
> +
> +int ext2fs_orphan_file_block_csum_verify(ext2_filsys fs, char *buf)
> +{
> +	struct ext4_orphan_block_tail *tail;
> +	__u32 crc;
> +	errcode_t retval;
> +
> +	if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
> +					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
> +		return 1;
> +	retval = ext2fs_orphan_file_block_csum(fs, buf, &crc);
> +	if (retval)
> +		return 0;
> +	tail = ext2fs_orphan_block_tail(fs, buf);
> +	return ext2fs_le32_to_cpu(tail->ob_checksum) == crc;
> +}
> -- 
> 2.1.4
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" 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-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On May 22, 2015, at 5:28 AM, Jan Kara <jack@suse.cz> wrote:
> 
> Add support for creating and deleting orphan file and a couple of
> utility functions that will be used in other tools.
> 
> Signed-off-by: Jan Kara <jack@suse.cz>
> ---
> lib/e2p/feature.c      |   4 +
> lib/ext2fs/Makefile.in |   2 +
> lib/ext2fs/ext2_fs.h   |  11 +++
> lib/ext2fs/ext2fs.h    |  35 +++++++-
> lib/ext2fs/orphan.c    | 217 +++++++++++++++++++++++++++++++++++++++++++++++++
> 5 files changed, 267 insertions(+), 2 deletions(-)
> create mode 100644 lib/ext2fs/orphan.c
> 
> diff --git a/lib/e2p/feature.c b/lib/e2p/feature.c
> index 73884f2cf5bf..a8e0d4a4644a 100644
> --- a/lib/e2p/feature.c
> +++ b/lib/e2p/feature.c
> @@ -45,6 +45,8 @@ static struct feature feature_list[] = {
> 			"snapshot_bitmap" },
> 	{	E2P_FEATURE_COMPAT, EXT4_FEATURE_COMPAT_SPARSE_SUPER2,
> 			"sparse_super2" },
> +	{	E2P_FEATURE_COMPAT, EXT4_FEATURE_COMPAT_ORPHAN_FILE,
> +			"orphan_file" },
> 
> 	{	E2P_FEATURE_RO_INCOMPAT, EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER,
> 			"sparse_super" },
> @@ -70,6 +72,8 @@ static struct feature feature_list[] = {
> 			"replica" },
> 	{	E2P_FEATURE_RO_INCOMPAT, EXT4_FEATURE_RO_COMPAT_READONLY,
> 			"read-only" },
> +	{	E2P_FEATURE_RO_INCOMPAT, EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT,
> +			"orphan_file_used" },
> 
> 	{	E2P_FEATURE_INCOMPAT, EXT2_FEATURE_INCOMPAT_COMPRESSION,
> 			"compression" },
> diff --git a/lib/ext2fs/Makefile.in b/lib/ext2fs/Makefile.in
> index 8a7f8ca52902..67120b10438c 100644
> --- a/lib/ext2fs/Makefile.in
> +++ b/lib/ext2fs/Makefile.in
> @@ -109,6 +109,7 @@ OBJS= $(DEBUGFS_LIB_OBJS) $(RESIZE_LIB_OBJS) $(E2IMAGE_LIB_OBJS) \
> 	native.o \
> 	newdir.o \
> 	openfs.o \
> +	orphan.o \
> 	progress.o \
> 	punch.o \
> 	qcow2.o \
> @@ -189,6 +190,7 @@ SRCS= ext2_err.c \
> 	$(srcdir)/native.c \
> 	$(srcdir)/newdir.c \
> 	$(srcdir)/openfs.c \
> +	$(srcdir)/orphan.c \
> 	$(srcdir)/progress.c \
> 	$(srcdir)/punch.c \
> 	$(srcdir)/qcow2.c \
> diff --git a/lib/ext2fs/ext2_fs.h b/lib/ext2fs/ext2_fs.h
> index a755cfac8eae..a77c8fa09938 100644
> --- a/lib/ext2fs/ext2_fs.h
> +++ b/lib/ext2fs/ext2_fs.h
> @@ -52,6 +52,7 @@
> #define EXT2_JOURNAL_INO	 8	/* Journal inode */
> #define EXT2_EXCLUDE_INO	 9	/* The "exclude" inode, for snapshots */
> #define EXT4_REPLICA_INO	10	/* Used by non-upstream feature */
> +#define EXT4_ORPHAN_INO		 9	/* Inode with orphan entries */

This still has a problem here, and can't be safely landed until it is resolved.
At a minimum, it shouldn't be possible to create a filesystem with COMPAT_ORPHAN_FILE
at the same time as COMPAT_EXCLUDE_BITMAP.  Since EXCLUDE_BITMAP never made it
upstream, that might be a reasonable compromise for now.

That said, we still need to do something about the lack of reserved inodes.

Cheers, Andreas

> /* First non-reserved inode for old ext2 filesystems */
> #define EXT2_GOOD_OLD_FIRST_INO	11
> @@ -769,6 +770,7 @@ struct ext2_super_block {
> /* #define EXT2_FEATURE_COMPAT_EXCLUDE_INODE	0x0080 not used, legacy */
> #define EXT2_FEATURE_COMPAT_EXCLUDE_BITMAP	0x0100
> #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
> +#define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x0400
> 
> 
> #define EXT2_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
> @@ -789,6 +791,7 @@ struct ext2_super_block {
> #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
> #define EXT4_FEATURE_RO_COMPAT_REPLICA		0x0800
> #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
> +#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x2000
> 
> 
> #define EXT2_FEATURE_INCOMPAT_COMPRESSION	0x0001
> @@ -838,6 +841,14 @@ struct ext2_super_block {
> #define EXT4_DEFM_DISCARD	0x0400
> #define EXT4_DEFM_NODELALLOC	0x0800
> 
> +#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
> +
> +/* Structure at the tail of orphan block */
> +struct ext4_orphan_block_tail {
> +	__u32 ob_magic;
> +	__u32 ob_checksum;
> +};
> +
> /*
>  * Structure of a directory entry
>  */
> diff --git a/lib/ext2fs/ext2fs.h b/lib/ext2fs/ext2fs.h
> index 28c46701da29..1e303d5d59ca 100644
> --- a/lib/ext2fs/ext2fs.h
> +++ b/lib/ext2fs/ext2fs.h
> @@ -555,7 +555,8 @@ typedef struct ext2_icount *ext2_icount_t;
> 					 EXT2_FEATURE_COMPAT_RESIZE_INODE|\
> 					 EXT2_FEATURE_COMPAT_DIR_INDEX|\
> 					 EXT2_FEATURE_COMPAT_EXT_ATTR|\
> -					 EXT4_FEATURE_COMPAT_SPARSE_SUPER2)
> +					 EXT4_FEATURE_COMPAT_SPARSE_SUPER2|\
> +					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
> 
> #ifdef CONFIG_MMP
> #define EXT4_LIB_INCOMPAT_MMP		EXT4_FEATURE_INCOMPAT_MMP
> @@ -589,7 +590,8 @@ typedef struct ext2_icount *ext2_icount_t;
> 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC|\
> 					 EXT4_LIB_RO_COMPAT_QUOTA|\
> 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
> -					 EXT4_FEATURE_RO_COMPAT_READONLY)
> +					 EXT4_FEATURE_RO_COMPAT_READONLY|\
> +					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
> 
> /*
>  * These features are only allowed if EXT2_FLAG_SOFTSUPP_FEATURES is passed
> @@ -1512,6 +1514,19 @@ errcode_t ext2fs_get_data_io(ext2_filsys fs, io_channel *old_io);
> errcode_t ext2fs_set_data_io(ext2_filsys fs, io_channel new_io);
> errcode_t ext2fs_rewrite_to_io(ext2_filsys fs, io_channel new_io);
> 
> +/* orphan.c */
> +/*
> + * Minimum orphan file size (it must be at least 1 block and smaller one isn't
> + * very useful).
> + */
> +#define EXT4_MIN_ORPHAN_FILE_SIZE 16384
> +
> +extern errcode_t ext2fs_create_orphan_file(ext2_filsys fs, blk_t num_blocks);
> +extern errcode_t ext2fs_truncate_orphan_file(ext2_filsys fs);
> +extern e2_blkcnt_t ext2fs_default_orphan_file_blocks(__u64 num_blocks);
> +extern errcode_t ext2fs_orphan_file_block_csum_set(ext2_filsys fs, char *buf);
> +extern int ext2fs_orphan_file_block_csum_verify(ext2_filsys fs, char *buf);
> +
> /* get_pathname.c */
> extern errcode_t ext2fs_get_pathname(ext2_filsys fs, ext2_ino_t dir, ext2_ino_t ino,
> 			       char **name);
> @@ -1645,6 +1660,9 @@ extern int ext2fs_dirent_name_len(const struct ext2_dir_entry *entry);
> extern void ext2fs_dirent_set_name_len(struct ext2_dir_entry *entry, int len);
> extern int ext2fs_dirent_file_type(const struct ext2_dir_entry *entry);
> extern void ext2fs_dirent_set_file_type(struct ext2_dir_entry *entry, int type);
> +extern int ext2fs_inodes_per_orphan_block(ext2_filsys fs);
> +extern struct ext4_orphan_block_tail *ext2fs_orphan_block_tail(ext2_filsys fs,
> +							       char *buf)
> 
> #endif
> 
> @@ -1915,6 +1933,19 @@ _INLINE_ void ext2fs_dirent_set_file_type(struct ext2_dir_entry *entry, int type
> 	entry->name_len = (entry->name_len & 0xff) | (type << 8);
> }
> 
> +_INLINE_ int ext2fs_inodes_per_orphan_block(ext2_filsys fs)
> +{
> +	return (fs->blocksize - sizeof(struct ext4_orphan_block_tail)) /
> +		sizeof(__u32);
> +}
> +
> +_INLINE_ struct ext4_orphan_block_tail *
> +ext2fs_orphan_block_tail(ext2_filsys fs, char *buf)
> +{
> +	return (struct ext4_orphan_block_tail *)(buf + fs->blocksize -
> +		sizeof(struct ext4_orphan_block_tail));
> +}
> +
> #undef _INLINE_
> #endif
> 
> diff --git a/lib/ext2fs/orphan.c b/lib/ext2fs/orphan.c
> new file mode 100644
> index 000000000000..1fd5c0688218
> --- /dev/null
> +++ b/lib/ext2fs/orphan.c
> @@ -0,0 +1,217 @@
> +/*
> + * orphan.c --- utility function to handle orphan file
> + *
> + * Copyright (C) 2015 Jan Kara.
> + *
> + * %Begin-Header%
> + * This file may be redistributed under the terms of the GNU Library
> + * General Public License, version 2.
> + * %End-Header%
> + */
> +
> +#include "config.h"
> +#include <string.h>
> +
> +#include "ext2_fs.h"
> +#include "ext2fsP.h"
> +
> +errcode_t ext2fs_truncate_orphan_file(ext2_filsys fs)
> +{
> +	struct ext2_inode inode;
> +	errcode_t err;
> +
> +	err = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		return err;
> +
> +	err = ext2fs_punch(fs, EXT4_ORPHAN_INO, &inode, NULL, 0, ~0ULL);
> +	if (err)
> +		return err;
> +
> +	fs->flags &= ~EXT2_FLAG_SUPER_ONLY;
> +	memset(&inode, 0, sizeof(struct ext2_inode));
> +	err = ext2fs_write_inode(fs, EXT4_ORPHAN_INO, &inode);
> +
> +	fs->super->s_feature_compat &= ~EXT4_FEATURE_COMPAT_ORPHAN_FILE;
> +	ext2fs_mark_super_dirty(fs);
> +
> +	return err;
> +}
> +
> +struct mkorphan_info {
> +	char *buf;
> +	char *zerobuf;
> +	blk_t num_blocks;
> +	blk_t alloc_blocks;
> +	errcode_t err;
> +};
> +
> +static int mkorphan_proc(ext2_filsys	fs,
> +			 blk64_t	*blocknr,
> +			 e2_blkcnt_t	blockcnt,
> +			 blk64_t	ref_block EXT2FS_ATTR((unused)),
> +			 int		ref_offset EXT2FS_ATTR((unused)),
> +			 void		*priv_data)
> +{
> +	struct mkorphan_info *oi = (struct mkorphan_info *)priv_data;
> +	blk64_t new_blk;
> +	errcode_t err;
> +
> +	err = ext2fs_new_block2(fs, 0, 0, &new_blk);
> +	if (err) {
> +		oi->err = err;
> +		return BLOCK_ABORT;
> +	}
> +	ext2fs_block_alloc_stats2(fs, new_blk, +1);
> +	if (blockcnt >= 0)
> +		err = io_channel_write_blk64(fs->io, new_blk, 1, oi->buf);
> +	else
> +		err = io_channel_write_blk64(fs->io, new_blk, 1, oi->zerobuf);
> +	if (err) {
> +		oi->err = err;
> +		return BLOCK_ABORT;
> +	}
> +	oi->alloc_blocks++;
> +	*blocknr = new_blk;
> +	if (blockcnt >= 0 && --oi->num_blocks == 0)
> +		return BLOCK_CHANGED | BLOCK_ABORT;
> +	return BLOCK_CHANGED;
> +}
> +
> +errcode_t ext2fs_create_orphan_file(ext2_filsys fs, blk_t num_blocks)
> +{
> +	struct ext2_inode inode;
> +	errcode_t err;
> +	char *buf = NULL, *zerobuf = NULL;
> +	struct mkorphan_info oi;
> +	struct ext4_orphan_block_tail *ob_tail;
> +
> +	err = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		return err;
> +	if (EXT2_I_SIZE(&inode)) {
> +		err = ext2fs_truncate_orphan_file(fs);
> +		if (err)
> +			return err;
> +	}
> +
> +	memset(&inode, 0, sizeof(struct ext2_inode));
> +	if (fs->super->s_feature_incompat & EXT3_FEATURE_INCOMPAT_EXTENTS) {
> +		inode.i_flags |= EXT4_EXTENTS_FL;
> +		err = ext2fs_write_inode(fs, EXT4_ORPHAN_INO, &inode);
> +		if (err)
> +			return err;
> +	}
> +
> +	err = ext2fs_get_mem(fs->blocksize, &buf);
> +	if (err)
> +		return err;
> +	err = ext2fs_get_mem(fs->blocksize, &zerobuf);
> +	if (err)
> +		goto out;
> +	memset(buf, 0, fs->blocksize);
> +	memset(zerobuf, 0, fs->blocksize);
> +	ob_tail = ext2fs_orphan_block_tail(fs, buf);
> +	ob_tail->ob_magic = ext2fs_cpu_to_le32(EXT4_ORPHAN_BLOCK_MAGIC);
> +	ext2fs_orphan_file_block_csum_set(fs, buf);
> +	oi.num_blocks = num_blocks;
> +	oi.alloc_blocks = 0;
> +	oi.buf = buf;
> +	oi.zerobuf = zerobuf;
> +	oi.err = 0;
> +	err = ext2fs_block_iterate3(fs, EXT4_ORPHAN_INO, BLOCK_FLAG_APPEND,
> +				    0, mkorphan_proc, &oi);
> +	if (err)
> +		goto out;
> +
> +	/* Reread inode after blocks were allocated */
> +	err = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		goto out;
> +	ext2fs_iblk_set(fs, &inode, 0);
> +	inode.i_atime = inode.i_mtime =
> +		inode.i_ctime = fs->now ? fs->now : time(0);
> +	inode.i_links_count = 1;
> +	inode.i_mode = LINUX_S_IFREG | 0600;
> +	ext2fs_iblk_add_blocks(fs, &inode, oi.alloc_blocks);
> +	err = ext2fs_inode_size_set(fs, &inode,
> +			(unsigned long long)fs->blocksize * num_blocks);
> +	if (err)
> +		goto out;
> +	err = ext2fs_write_new_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (err)
> +		goto out;
> +
> +	fs->super->s_feature_compat |= EXT4_FEATURE_COMPAT_ORPHAN_FILE;
> +	ext2fs_mark_super_dirty(fs);
> +out:
> +	if (buf)
> +		ext2fs_free_mem(&buf);
> +	if (zerobuf)
> +		ext2fs_free_mem(&zerobuf);
> +	return err;
> +}
> +
> +/*
> + * Find reasonable size for orphan file. We choose orphan file size to be
> + * between 32 and 512 filesystem blocks and not more than 1/4096 of the
> + * filesystem unless it is really small.
> + */
> +e2_blkcnt_t ext2fs_default_orphan_file_blocks(__u64 num_blocks)
> +{
> +	if (num_blocks < 128 * 1024)
> +		return 32;
> +	if (num_blocks < 2 * 1024 * 1024)
> +		return num_blocks / 4096;
> +	return 512;
> +}
> +
> +static errcode_t ext2fs_orphan_file_block_csum(ext2_filsys fs, char *buf,
> +					       __u32 *crc)
> +{
> +	int inodes_per_ob = ext2fs_inodes_per_orphan_block(fs);
> +	__u32 gen;
> +	ext2_ino_t inum;
> +	struct ext2_inode inode;
> +	errcode_t retval;
> +
> +	retval = ext2fs_read_inode(fs, EXT4_ORPHAN_INO, &inode);
> +	if (retval)
> +		return retval;
> +	inum = ext2fs_cpu_to_le32(EXT4_ORPHAN_INO);
> +	gen = ext2fs_cpu_to_le32(inode.i_generation);
> +	*crc = ext2fs_crc32c_le(fs->csum_seed, (unsigned char *)&inum,
> +			       sizeof(inum));
> +	*crc = ext2fs_crc32c_le(*crc, (unsigned char *)&gen, sizeof(gen));
> +	*crc = ext2fs_crc32c_le(*crc, buf, inodes_per_ob * sizeof(__u32));
> +
> +	return 0;
> +}
> +
> +errcode_t ext2fs_orphan_file_block_csum_set(ext2_filsys fs, char *buf)
> +{
> +	struct ext4_orphan_block_tail *tail;
> +
> +	if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
> +					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
> +		return 0;
> +
> +	tail = ext2fs_orphan_block_tail(fs, buf);
> +	return ext2fs_orphan_file_block_csum(fs, buf, &tail->ob_checksum);
> +}
> +
> +int ext2fs_orphan_file_block_csum_verify(ext2_filsys fs, char *buf)
> +{
> +	struct ext4_orphan_block_tail *tail;
> +	__u32 crc;
> +	errcode_t retval;
> +
> +	if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
> +					EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
> +		return 1;
> +	retval = ext2fs_orphan_file_block_csum(fs, buf, &crc);
> +	if (retval)
> +		return 0;
> +	tail = ext2fs_orphan_block_tail(fs, buf);
> +	return ext2fs_le32_to_cpu(tail->ob_checksum) == crc;
> +}
> -- 
> 2.1.4
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html


Cheers, Andreas





--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Fri 22-05-15 10:35:21, Darrick J. Wong wrote:
> On Fri, May 22, 2015 at 01:28:54PM +0200, Jan Kara wrote:
...
> > diff --git a/lib/ext2fs/ext2fs.h b/lib/ext2fs/ext2fs.h
> > index 28c46701da29..1e303d5d59ca 100644
> > --- a/lib/ext2fs/ext2fs.h
> > +++ b/lib/ext2fs/ext2fs.h
> > @@ -555,7 +555,8 @@ typedef struct ext2_icount *ext2_icount_t;
> >  					 EXT2_FEATURE_COMPAT_RESIZE_INODE|\
> >  					 EXT2_FEATURE_COMPAT_DIR_INDEX|\
> >  					 EXT2_FEATURE_COMPAT_EXT_ATTR|\
> > -					 EXT4_FEATURE_COMPAT_SPARSE_SUPER2)
> > +					 EXT4_FEATURE_COMPAT_SPARSE_SUPER2|\
> > +					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
> >  
> >  #ifdef CONFIG_MMP
> >  #define EXT4_LIB_INCOMPAT_MMP		EXT4_FEATURE_INCOMPAT_MMP
> > @@ -589,7 +590,8 @@ typedef struct ext2_icount *ext2_icount_t;
> >  					 EXT4_FEATURE_RO_COMPAT_BIGALLOC|\
> >  					 EXT4_LIB_RO_COMPAT_QUOTA|\
> >  					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
> > -					 EXT4_FEATURE_RO_COMPAT_READONLY)
> > +					 EXT4_FEATURE_RO_COMPAT_READONLY|\
> > +					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
> >  
> >  /*
> >   * These features are only allowed if EXT2_FLAG_SOFTSUPP_FEATURES is passed
> > @@ -1512,6 +1514,19 @@ errcode_t ext2fs_get_data_io(ext2_filsys fs, io_channel *old_io);
> >  errcode_t ext2fs_set_data_io(ext2_filsys fs, io_channel new_io);
> >  errcode_t ext2fs_rewrite_to_io(ext2_filsys fs, io_channel new_io);
> >  
> > +/* orphan.c */
> > +/*
> > + * Minimum orphan file size (it must be at least 1 block and smaller one isn't
> > + * very useful).
> > + */
> > +#define EXT4_MIN_ORPHAN_FILE_SIZE 16384
> 
> What about 64k block size?  I guess it's fine not to use the whole block in
> this (non-default) configuration.

We round up file size to be a multiple of fs block size.

> > +struct mkorphan_info {
> > +	char *buf;
> > +	char *zerobuf;
> > +	blk_t num_blocks;
> > +	blk_t alloc_blocks;
> > +	errcode_t err;
> > +};
> > +
> > +static int mkorphan_proc(ext2_filsys	fs,
> > +			 blk64_t	*blocknr,
> > +			 e2_blkcnt_t	blockcnt,
> > +			 blk64_t	ref_block EXT2FS_ATTR((unused)),
> > +			 int		ref_offset EXT2FS_ATTR((unused)),
> > +			 void		*priv_data)
> > +{
> > +	struct mkorphan_info *oi = (struct mkorphan_info *)priv_data;
> > +	blk64_t new_blk;
> > +	errcode_t err;
> > +
> > +	err = ext2fs_new_block2(fs, 0, 0, &new_blk);
> 
> Hm.  I think this breaks the cluster allocation rules, since this allocates
> a new block for every logical block inside a cluster.

Yes, I didn't think of cluster allocation. Frankly, I have just mirrored
what a code creating journal does. If this is wrong, then that code needs
fixing as well ;)

> Hopefully ext2fs_fallocate will land soon, then we can get rid of open-coding
> file block allocation like this.  You'll still have to have the iterate3 loop
> to write out the appropriate block footer, but iirc the library call can be
> told to allocate written extents without zeroing the blocks, precisely for
> cases like these.

So should I wait for that or do you have any pointer to code which does
it correctly?

								Honza

On Fri 22-05-15 15:59:19, Andreas Dilger wrote:
> On May 22, 2015, at 5:28 AM, Jan Kara <jack@suse.cz> wrote:
> > diff --git a/lib/ext2fs/ext2_fs.h b/lib/ext2fs/ext2_fs.h
> > index a755cfac8eae..a77c8fa09938 100644
> > --- a/lib/ext2fs/ext2_fs.h
> > +++ b/lib/ext2fs/ext2_fs.h
> > @@ -52,6 +52,7 @@
> > #define EXT2_JOURNAL_INO	 8	/* Journal inode */
> > #define EXT2_EXCLUDE_INO	 9	/* The "exclude" inode, for snapshots */
> > #define EXT4_REPLICA_INO	10	/* Used by non-upstream feature */
> > +#define EXT4_ORPHAN_INO		 9	/* Inode with orphan entries */
> 
> This still has a problem here, and can't be safely landed until it is
> resolved.  At a minimum, it shouldn't be possible to create a filesystem
> with COMPAT_ORPHAN_FILE at the same time as COMPAT_EXCLUDE_BITMAP.  Since
> EXCLUDE_BITMAP never made it upstream, that might be a reasonable
> compromise for now.

Yeah, for now I've chosen inode number 9 as for testing it's good enough.
We can make this feature incompatible with COMPAT_EXCLUDE_BITMAP as you
suggest or we can use some higher inode number and require increased number
of reserved inodes. I don't mind either too much.

> That said, we still need to do something about the lack of reserved inodes.

Agreed. I've tried to get some decision from Ted regarding this a few times
but failed.

								Honza